Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Risk factors associated with moderate and serious injuries attributable to the 1994 Northridge earthquake, Los Angeles County
(USC Thesis Other)
Risk factors associated with moderate and serious injuries attributable to the 1994 Northridge earthquake, Los Angeles County
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy subm itted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send U M I a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9’ black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. Bell & Howell Information and Learning 300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA 800-521-0600 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. RISK FACTORS ASSOCIATED WITH MODERATE AND SERIOUS INJURIES ATTRIBUTABLE TO THE 1994 NORTHRIDGE EARTHQUAKE, LOS ANGELES COUNTY Volume I by Maya Louise Mahue-Giangreco A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements of the Degree DOCTOR OF PHILOSOPHY (Epidemiology) December 1999 Copyright 1999 Maya Louise Mahue-Giangreco Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UM I Number 9987628 Copyright 1999 by Mahue-Giangreco. Maya Louise All rights reserved. __ ___ < R > UMI UMI Microform9987628 Copyright 2001 by Bell & Howell Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. Bell & Howell Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY OF SOUTHERN CALIFORNIA THE GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES. CALIFORNIA 90007 This dissertation, written by MAyA . ..M a W Q> . . \ . b. H under the direction of h/LC. Dissertation Committee, and approved by all its members, has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for the degree of DOCTOR OF PHILOSOPHY Dean o f Graduate Studies Novem ber 2 9 , 1999 DISSERTATION COMMITTEE c b P K -g — .................... Chairperson Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This dissertation is dedicated to my mother and father who have always encouraged me to excel, Stanley Azen, who has always advised me wisely, Billie Weiss, who opened the door to injury epidemiology for me, and to my husband who patiently waited four years for me to finish this study, and stood in the office doorway at two o’clock in the morning and said ‘Hi. I’m Chris. I’m your husband’, while I worked on the final draft. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ii A C K N O W L E D G E M E N T S I would like to thank: the United States Geological Survey (Grant USGS-SC- 7107) for financial support; Hope Seligson, Ron Eguchi, and EQE, International for their engineering expertise; the Centers for Disease Control and Prevention for financial support of the pilot study (Award No. 953124); the Los Angeles County Board of Supervisors for approving these studies; Billie Weiss for her unwavering support of injury prevention; Susan Preston-Martin, Wendy Mack, Linda Bourque, Masanobu Shinozuka and Malcolm Pike for their mentoring and careful review of dissertation drafts; Corrine Peek-Asa, Kimberley Shoaf, Robin Wagner, and Pamela Anderson for paving the way before me and for their support, advice, and encouragement; Jess Kraus and the Southern California Injury Prevention Research Center for financial support and loaning of staff; all the student interns, especially Marjorie Estoque and Silvia Dominguez for their reliable and diligent help abstracting data, Marizen Ramirez for her help cleaning and coding the data, Carrie Hartman Casteel for her help designing the data abstraction instrument, documentation, and friendship; Caroline Ervin, DeeAnn Allen, Wendy Cozen, and Tom Mack for their wisdom, encouragement, and friendship; Christian, Mom and Dad, Michelle, Celia, Christina, Lizzie and Dean, Lizzie and Atilio, Emo, Java, and Xerox for their unconditional love. I could not have done this without you. in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS Page Dedication ii Acknowledgements ..............................................................................................iii List of Tables xi List of Figures xvi List of Abbreviations ......................................................................................... xviii Abstract xxii Chapter Section I INTRODUCTION................................................................1 II REVIEW OF THE LITERATURE.................................... 5 1.00 Injury Epidemiology............................................................ 5 1.01 Pioneers in Injury Prevention..............................................6 1.02 Injury Severity Measurements.............................................9 1.03 Assessment of Multiple Injuries...................................... 10 1.04 Coding of Injury Mechanism........................................... 12 1.05 Methodological Challenges of Disaster-Related Injury Epidemiology....................................................................... 13 1.06 Identification of Cases......................................................... 16 1.07 Assessment of the Population at Risk for Earthquake-Related Injuries.................................................................................. 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.08 Calculating Rates of Earthquake-Related Injuries..............26 1.09 Earthquake-Related Injuries: Guatemala, 1976.................29 1.10 Earthquake-Related Injuries: Whittier Narrows, 1987.... 31 1.11 Earthquake-Related Injuries: Northern Armenia, 1988 ... 33 1.12 Earthquake-Related Injuries: Loma Prieta, 1989............. 40 1.13 Earthquake-Related Injuries: Luzon, Phillipines, 1990... 43 1.14 Earthquake-Related Injuries: Northridge, California, 1994 ...................................................................................... 45 1.15 Generalization of Previous Injury Studies to Other Earthquakes............................................................................53 1.16 Sources of Injury Data......................................................... 55 2.00 Earthquake Hazard Assessment and Seismology.............. 60 2.01 Geology Background........................................................... 61 2.02 Fault Terminology................................................................62 2.03 Seismology Background......................................................66 2.04 Comparative Analyses of Seismographs............................ 70 2.05 Liquefaction...........................................................................71 2 06 Earthquake Magnitude Scales.............................................73 2.07 Earthquake Intensity Scales................................................ 74 3 .00 Structural Engineering Background................................... 98 3.01 Earthquake Hazard Estimation Methodology................... 99 3.02 ATC-13 Methodology........................................................ 102 v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 .03 Correlations between Ground Motion and Structural Damage................................................................................. 117 4.00 Recent Advances in Earthquake Hazard Estimation........120 4.01 FEMA-NIBS Hazard Assessment Methodology.............. 120 4.02 EPED AT Methodology....................................................... 127 4.03 Interdisciplinary Advances in Hazard Estimation............ 130 4.04 Summary o f Limitations o f Current Methodologies........132 4.05 Summary o f Recent Advances in Earthquake Hazard Estimation.............................................................................134 III. PILOT STUDY................................................................... 136 1.00 Pilot Study Background........................................................136 2.00 Pilot Study Methodology..................................................... 137 2.01 Case Identification through Emergency Department (ED) Log Data.................................................................................139 2.02 Data Management and Quality Control.............................142 2.03 Statistical Methods............................................................... 144 3.00 Pilot Study Results.................................................................144 3 .01 Proportion o f Emergency Department Visits for Injury Treatment...............................................................................145 3 .02 Validation Study Comparing ED Logs and Medical Records at Hospital D’........................................................................... 151 4.00 Discussion o f Pilot Study......................................................154 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.01 Problems Associated with Calculation of Rates.................155 4.02 Comparison of Study Results Based on Different Methodologies..................................................................... 156 4.03 Discussion of the Validity Study between Emergency Department Logs and Medical Records.............................161 5.00 Limitations to the Pilot Study..............................................162 6.00 Conclusions Based on the Pilot Study................................163 IV STUDY METHODOLOGY............................................... 165 1.00 Intended Study Design......................................................... 165 2.00 Facility Selection..................................................................167 2.01 Facilities in Areas o f ‘High Impact’ ...................................171 2.02 Summary of Facility Selection............................................172 3.00 Case Identification................................................................173 3.01 Data Abstraction...................................................................173 3.02 Quality Control.....................................................................175 3 .03 Identifying Cases of Earthquake-Related Injuries............. 180 4.00 Injury Coding and Associated Assumptions......................187 4.01 Injury Scene Categorization................................................187 4.02 Coding of Body Location and External Cause of Injury .188 4.03 Injury Severity Scoring........................................................189 5.00 Linking Injury Data to Engineering Data.......................... 193 5 .01 Linkage of Injury Characteristics to Building Databases 193 vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 .02 Linkage of Injury and Building Characteristics to Geologic Databases............................................................................. 196 6.00 Statistical Methods................................................................ 199 6.01 Grouping of Demographic Variables...................................199 6.02 Establishing Cut-Points for Year of Building Construction....................................................................... 200 6.03 Grouping of Geologic Variables.........................................201 6.04 Statistical Modeling, Analytic Methods and Computer Software..............................................................................201 V. RESULTS.............................................................................206 1.00 Baseline Versus Post-Event Descriptive Results.............. 206 1.01 Facility Activity....................................................................207 1.02 Demographic Make-Up of Patients by Time Period.........214 1.03 Geographic Comparison of Injured Patients by Time Period.................................................................................. 224 2.00 Descriptive Characteristics of Study Sample.....................229 2.01 Demographic Characteristics of Injured Patients..............229 2.02 Injury Characteristics of Study Patients............................. 231 2 .03 Geographic Distribution of Earthquake-Related Injuries by Estimated Ground-Shaking and Intensity.........................236 3 .00 Logistic Regression: Identification of Factors Associated with Moderate and Serious Injury.....................................245 viii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.01 Demographic Characteristics with Respect to Injury Severity Score......................................................................247 3.02 Injury Characteristics with Respect to Injury Severity Score.................................................................................... 255 3.03 Building Structure Characteristics with Respect to Injury Severity Score......................................................................262 3.04 Geologic Characteristics with Respect to Injury Severity Score......................................................................279 3.05 Demographic, Injury, Structural, and Geologic Characteristics with Respect to Injury Severity Score ...285 VI. DISCUSSION.....................................................................302 1.00 Estimates of Relative R isk................................................ 3 02 1.01 Demographic Characteristics of Patients with More Serious Injuries....................................................................303 1.02 Characteristics of More Serious Injuries.......................... 304 1.03 Characteristics of Structures Associated with More Serious Injuries....................................................................306 1.04 Geologic Characteristics of Scenes Associated with More Serious Injuries....................................................................308 2 .00 Comparison o f Observed and Predicted Injuries Using EPEDAT (Early Post Earthquake Damage Assessment Tool) Software.....................................................................310 ix Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.00 Discussion o f the Methodology........................................... 311 3 .01 Sample Selection...................................................................311 3 .02 Specificity o f Case Identification........................................314 4.00 Baseline Earthquake-Related Injuries.................................315 5.00 Limitations.............................................................................316 VII CONCLUSIONS.................................................................. 319 BIBLIOGRAPHY ........................................................................................... 318 APPENDIXES ................................................................................................ 1 . Grant Proposal Submitted to the United States Geological Survey (Department of the Interior) and Associated Award L etter....................................................................................337 2. Instruments Used for Data Collection................................366 3. Statistical Tests for Assuming Earthquake Attributability...................................................................... 375 4. Abbreviated Injury Severity (AIS) and Injury Severity Score (ISS) Guidelines..................................................................388 5. Memo from EQE International............................................419 6. Final Report from EQE International................................ 435 7. Likelihood Ratio Tests: Model Selection Process............463 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLES Table Page 1 . Example of a Haddon Matrix for Automobile Collision....................... 8 2. Modified Mercalli Intensity Scale (MMI), Adapted from Seiberg’s (1923) Mercalli-Cancani Scale, Modified and Condensed, Quoted from Wood and Neumann (1931)....................................................... 75 3. Japan Meteorological Agency (JMA) Scale........................................ 82 4. Medvedev-Sponheuer-Kamik (MSK) Intensity Scale. Quoted from Barosh (1969).............................................................................. 84 5. Rossi-Forel (RF) Intensity Scale. Quoted from Richter (1958)........91 6. Geofian Intensity Scale. Quoted from Barosh (1969)........................92 7. ATC-13 Methodology: Earthquake Engineering Facility Classification....................................................................................... 104 8. ATC-13 Methodology: Economic Social Function Classification...................................................................................... 107 9. ATC-13 Methodology: Example of a Damage Probability Matrix Based on Expert Opinion....................................................................109 10. ATC-13 Methodology: Estimates of Weighted Loss of Function Restoration Time (in Days) by Social Function Classification I l l 11. ATC-13 Methodology: National Oceanic and Atmospheric Association Review of Historical Earthquakes.............................. 113 12. ATC-13 Methodology: Whitman, Cornell, et al. (1975) Death and Injury Proportions................................................................................115 13. ATC-13 Methodology: Injury and Death Rates per Damage State...................................................................................................... 116 14. Comparison between Levels of Trauma Care Centers..................... 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15. Size of Study Facilities and Licensed Level of Emergency Medical C are.................................................................................................... 145 16. Northridge Earthquake Study: Proportion of Injuries from Nine Emergency Departments, January 1994.......................................... 146 17. Comparison of Injury and Demographic Characteristics of Patients Presenting to Nine Emergency Departments after the 1994 Northridge Earthquake and at Baseline...........................................150 18. Validity Study for Hospital‘D ’: Agreement between Medical Records and Emergency Department Logs.................................... 153 19. Example of Discrepancies in Reported Injuries due to Different Study Methods...................................................................................157 20. Example of Discrepancies in Reported Injuries due to Date of Data Review: Reported Injuries at Hospital #6 on January 17, 1994.................................................................................................... 159 21. Hospitals in Los Angeles County Included in Pilot Study of Earthquake-Related Injuries and Hospitals in Average MMI Areas of VIII Due to the 1994 Northridge Earthquake.............................169 22. Comparison of Numbers of Medical Records Requested and Retrieved at 3 Facilities in Los Angeles County, January 1994.... 174 23. Eligibility and Exclusion Criteria for Injury Incidents Treated at 4 Emergency Departments, Los Angeles County, January 17-31, 1994.................................................................................................... 177 24. Frequencies of Valid Injuries Treated at 4 Facilities in Los Angeles County, January 1994................................................. 179 25. Proportions of Injuries Attributed to the 1994 Northridge Earthquake at 4 Facilities in Los Angeles County, January 1994...................................................................................... 185 26. Injury Severity Scores for Earthquake-Related Injuries from 4 Emergency Departments in Los Angeles County, January 17-31, 1994.......................................................................... 191 xii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27. Injury Severity Scores by Facility for Earthquake-Related Injuries Treated at 4 Emergency Departments in Los Angeles County, January 17-31, 1994...................................... 192 28. Characteristics of Structures Linked to Injury Scenes for Patients Injured from the Northridge Earthquake and Treated at 4 Emergency Departments in Los Angeles County, January 17-31, 1994........................................................................... 193 29. Characteristics of Estimates of Geologic Conditions at Sites Linked to Injury Scenes for Patients Injured from the Northridge Earthquake and Treated at 4 Emergency Departments in Los Angeles County, January 17-31, 1994...................................... 198 30. Variable Groups and Corresponding Variables.................................204 31. Gender (Male: Female) Ratios of Injured Patients Treated at 4 Emergency Departments in Los Angeles County, January 1994 (Pre and Post Earthquake)...........................................215 32. Age Distribution of Injured Patients Treated at 4 Emergency Departments in Los Angeles County, January 1994, Pre and Post Earthquake............................................................................................216 33. Ethnic Distributions of Injured Patients Treated at 4 Emergency Departments in Los Angeles County, January 1994, Pre and Post Earthquake...........................................................................................220 34. External Mechanisms o f Injury for Patients Treated at 4 Emergency Departments in Los Angeles County, January 1994 (Pre and Post Northridge Earthquake)...............................................225 35. Demographic Characteristics (Including Missing Values) of Patients Injured from the Northridge Earthquake that Presented at 4 Emergency Departments in Los Angeles County, January 17-31, 1994............................................................................ 230 36. Injury Characteristics o f Patients Treated at 4 Emergency Departments in Los Angeles County, January 17-31, 1994, with Earthquake-Related Injuries...................................................... 233 37. Minimum Detectable Odds Ratios for Various Sample Sizes Based on Selected Exposure Prevalence Values.............................. 246 xiii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38. Polytomous Logistic Regression: Associations between Demographic Characteristics of Injured Patients and Severity of Injury from the Northridge Earthquake (n=4I8)........................248 39. Dichotomous Logistic Regression: Associations between Demographic Characteristics of Patients and Severity of Injury from the Northridge Earthquake (n=438).........................................253 40. Polytomous Logistic Regression: Associations between Injury Characteristics of Patients and Severity of Injury from the Northridge Earthquake (n=330)........................................................ 258 41. Dichotomous Logistic Regression: Associations between Injury Characteristics of Patients and Severity o f Injury from the Northridge Earthquake (n=375)........................................................260 42. Polytomous Logistic Regression: Associations between Structural Factors Linked to Scene of Injury and Severity of Injury from the Northridge Earthquake (n=393)........................265 43. Polytomous Logistic Regression: Odds Ratios for Interaction between Structure Use and Year Built with Respect to Severity of Injury (n=393)............................................................................... 268 44. Polytomous Logistic Regression: Associations between Selected Structural Factors Linked to Injury Scenes with Respect to Injury Severity (n=259).................................................................... 270 45. Polytomous Logistic Regression: Odds Ratios for Interaction between Structure Use and Year Built with Respect to Severity of Injury, Adjusting for Patient Demographics (n=259).................271 46. Dichotomous Logistic Regression: Associations between Structural Factors Linked to Scene of Injury and Severity of Injury from the Northridge Earthquake (n=393)........................ 273 47. Dichotomous Logistic Regression. Odds Ratios for Interaction between Structure Use and Year Built with Respect to Severity of Injury from the Northridge Earthquake (n=393).........................275 48. Dichotomous Logistic Regression: Associations between Structural Factors Linked to Scene of Injury with Respect to Severity of Injury from the Northridge Earthquake, Adjusting for Patient Demographics (n=275)................................. 277 xiv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49. Dichotomous Logistic Regression: Odds Ratios for Interaction between Structure Use and Year Built with Respect to Severity of Injury from the Northridge Earthquake, Adjusting for Patient Demographics (n=275)......................................................................278 50. Polytomous Logistic Regression: Associations between Geology of the Injury Scene and Severity of Injury from the Northridge Earthquake (n=279)........................................................................... 281 51. Dichotomous Logistic Regression: Associations between Geology of the Injury Scene and Severity of Injury from the Northridge Earthquake (n=295)........................................................................... 283 52. Polytomous Logistic Regression: Associations between Demographic, Injury, Structural, and Geologic Characteristics of the Injury Scene with Respect to Severity of Injury from the Northridge Earthquake (n=196)........................................................287 53. Dichotomous Logistic Regression: Associations between Demographic, Injury, Structural, and Geologic Characteristics of the Injury Scene and Severity of Injury from the Northridge Earthquake (n=230)........................................................................... 294 xv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. F I G U R E S Figure Page 1 . The Injury Pyramid from the Burden of Injury: United States, 1995.............................................................................. 19 2. Graphical Presentation of Angular Elements of a Fault..................... 63 3. Graphical Presentation of Structural Elements of a Fault..................64 4. Graphical Presentation of Components of Displacement in an Oblique Fault..........................................................................................65 5. Graphical Presentation of Particle Motion of Body Waves (P-waves), and Surface Waves (S-waves, Love waves, and Rayleigh waves).................................................................................... 67 6. Essential Components of a Seismograph............................................69 7. Graphical Comparison of Seismology Intensity Scales......................81 8. Isoseismal Map of Modified Mercalli Intensity Estimates (MMI) due to the Northridge Earthquake, January 17, 1994............ 95 9. Isoseismal Map of Log Peak Ground Acceleration Estimates (PGA) due to the Northridge Earthquake, January 17, 1994........... 97 10. Graphical Comparison of Elastic Strength Demand Spectra Recorded during the Northridge Earthquake, January 17, 1994 ... 119 11. HAZUS Methodology Modules...........................................................121 12. Examples of Building Fragility Curves for Slight, Moderate, Extensive, and Complete Damage..................................................... 124 13. Daily Frequencies of Injury Visits to 4 Emergency Departments in Los Angeles County, January 1994.............................................. 208 14. Daily Frequencies of Injury Visits to 4 Emergency Departments in Los Angeles County, 1/94, Deleting 1/17/94...............................209 xvi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15. Frequencies of Injury Visits to Hospital ‘B ’ Emergency Department by Weekday, January 1994.......................................... 211 16. Frequencies of Injury Visits to Hospital ‘C’ Emergency Department by Weekday, January 1994.......................................... 212 17. Frequencies of Injury Visits to Hospital ‘D ’ Emergency Department by Weekday, January 1994............................................213 18. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7), Regional Extent................................................................................... 227 19. Injuries Recorded at Emergency Departments between 1/1/94 and 1/16/94, Regional Extent............................................................. 228 20. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Modified Mercalli Intensity, Regional Extent..........................237 21. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Modified Mercalli Intensity, San Fernando Extent..................238 22. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Modified Mercalli Intensity, Southwestern Extent.................239 23. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Peak Ground Acceleration Contours (% g), Regional Extent....................................................................................................241 24. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Peak Ground Acceleration Contours (% g), San Fernando Extent........................................................................... 242 25. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Peak Ground Acceleration Contours (% g), Southwestern Extent........................................................................... 244 xvii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A B B R E V I A T I O N S ACD Acute Communicable Diseases AIS Abbreviated Injury Severity ATC Applied Technology Council CATI Computer-Assisted Telephone Interview CDC Centers for Disease Control and Prevention CDF Central Damage Factor CT Computed Tomography CUBE Caltech-USGS Broadcast-of -Earthquakes System CUREe California Universities for Research in Earthquake Engineering DAC Disaster Application Center DHS Department of Health Services DPM Damage Probability Matrices DS Damage State E-Codes External Cause of Injury Codes ED Emergency Department EEA Engineering-Economics Associates EERI Earthquake Engineering Research Institute EMS Emergency Medical Services EPEDAT Early Post-Earthquake Damage Assessment Tool xviii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. EQ EQE FDHAP FEDLOSS FEIMS FEMA FF GIS HAZUS HMO ICD ICD-9 INCAP ISS IVPP JCAHO J\1A LA LAC LAMB LRT Earthquake EQE, International Center for Advanced Planning and Research Federal Disaster Housing Assistance Program FEMA Earthquake Damage and Loss Estimation System FEMA Earthquake Impacts Modeling System Federal Emergency Management Association Free Field Geographic Information System Hazards in the United States Health Maintenance Organization International Classification of Diseases International Classification o f Diseases, Ninth Revision Institute of Nutrition for Central America and Panama Injury Severity Score Injury and Violence Prevention Program Joint Commission on the Accreditation of Healthcare Organizations Japan Meteorological Agency Scale of Earthquake Intensity Los Angeles Los Angeles County Loss Assessment of Memphis Buildings Likelihood Ratio Test \L \ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. mb Body Wave Magnitude Scale of Earthquake M l Local Magnitude Scale of Earthquake MLE Maximum Likelihood Estimate MLGW Memphis Light, Gas, and Water Division MMI Modified Mercalli Intensity of Earthquake Ms Surface Wave Magnitude Scale of Earthquake MSK Medvedev-Sponuheuer-Kamik Scale of Earthquake-Intensity Mw Moment Magnitude Scale of Earthquake NCEER National Center for Earthquake Engineering Research NIBS National Institute of Building Sciences NOAA National Oceanic and Atmospheric Agency OES California Governor's Office of Emergency Services OR Odds Ratio PESH Potential Earth Science Hazard PGA Peak Ground Acceleration P-Waves Primary Waves RF Rossi-Forel Earthquake Intensity Scale RMS Risk Management Solutions, Inc. SAS Statistical Analysis System SPA Service Planning Area S-Waves Secondary Waves UBC Uniform Build Codes Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UCLA University of California Los Angeles US United States of America USGS United States Geological Survey WHO World Health Organization Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Risk Factors Associated with Moderate and Serious Injuries Attributable to the 1994 Northridge Earthquake, Los Angeles County Patients from four emergency departments in Los Angeles County were included in a study of earthquake-related injuries. O f 4,190 medical records reviewed, 3,596 had valid injury and address information. Baseline data (January 1-16, 1994, n=l,323) were compared to post-event data (January 17-31, 1994, n=2,273) to investigate excess injuries by facility, weekday, and mechanism. Of 684 injuries linked to the Northridge earthquake, 250 were clearly denoted as such in the medical records; 434 were assumed to be earthquake-related. Injury Severity Scores were assigned to all earthquake-related injuries. Injury scene addresses were linked to the County Assessor's Building Inventory. Successfully matched parcels were subsequently linked to geologic characteristics from the January 17 Northridge Earthquake. Logistic regression was used to identify risk factors for more serious injuries. Variables were analyzed by category (demographic, injury, structural, and geologic), then merged. Analysis showed patients over age 59 (versus 30-39) had six times the risk for more serious injury; those sustaining injuries to the upper versus lower extremities had 2.6 times the risk of more serious injury; those injured from falls versus cutting or piercing mechanisms had 5.3 times the risk of more serious injury; although only 1% of the injuries was associated with any structural collapse, patients injured in multi-family structures versus single or duplex housing had 4.8 times the risk of more serious injury, xxii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. those injured in structures built before 1960 versus after 1975 had 4.6 times the risk of more serious injury, and patients injured in areas of peak ground acceleration (PGA) at least 0.68 g versus less than 0.68 g had 13 .5 times the risk of more serious injury (adjusting for age, gender, ethnicity, mechanism and body location of injury, structure use, year of construction, Modified Mercalli Intensity, PGA, and soil conditions). Residents over age 59, those in multi-family and older structures may benefit most from targeted education to prepare (bolting furniture, keeping flashlights and slippers handy) and respond (covering head and arms with pillows, moving slowly with slippers and flashlight) in urban night-time earthquakes. xxiii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C H A P T E R I: I N T R O D U C T I O N The Northridge earthquake occurred on Monday, January 17, 1994, Martin Luther King Day, at 4:31 in the morning. Most of the nine million residents o f Los Angeles County, as well as some residents in adjacent Ventura, Orange, San Bernardino and Riverside Counties were jolted awake by this magnitude 6.7 earthquake. Natural catastrophes such as earthquakes introduce a number of important health care questions regarding injury and disease control, health care delivery, cost of health care services, and recommendations for reducing injuries and disease in future disasters. The discipline o f earthquake hazard mitigation has evolved to measure factors that affect risk and loss due to earthquakes in order to forecast and reduce economic loss and health problems. These forecasts provide policy-makers and health-care workers with information used to focus prevention programs, determine seismic safety codes, guide earthquake preparedness, and design protocols for appropriate responses during and after an earthquake. Earthquake hazard mitigation assesses property and health losses (including mortality) together under the category of casualties. Estimates of casualties associated with earthquakes do not typically employ real data, but depend heavily on simulated data Conclusions based on models from simulated data may be misleading, and may lead to inadequate or inappropriate recommendations and mobilization of resources. The purpose of this dissertation is to help improve the methodology used to estimate injuries and death due to earthquakes by identifying risk (or protective) factors associated with real injuries attributable to the Northridge earthquake. A review of 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. pertinent literature (Chapter If) will provide a background from an epidemiologic point of view. A review of suggested risk factors for injury due to seismic events is presented, as well as a discussion of methodological challenges faced in investigations of earthquake- related injuries. The pilot study is presented in Chapter III that shows an increase in injury visits to a sample of emergency departments in Los Angeles County after the 1994 Northridge Earthquake. This increase helped justify the detailed review of medical records that was required to obtain information on: a) patient demographics; b) characteristics of the injury; and c) the scene and circumstances surrounding earthquake-related injuries. The subsequent study was funded by the United States Geological Survey to obtain these epidemiologic details, and to link the injury database to existing engineering databases detailing building and geologic characteristics of the injury scene. That study is the basis for this dissertation. Documentation associated with the grant proposal is presented in Appendix I. The study methodology for this dissertation is presented in Chapter IV. In brief, actual data for patients treated at emergency departments in the Santa Clarita Valley, West Los Angeles, and South-Central Los Angeles was abstracted for the month of January 1994. The two-week time period before the earthquake (January 1-16, 1994) served as the baseline period or period of non-exposure. The two-week period during and after the earthquake (January 17-31, 1994) served as the post-event or exposure period. Injury- specific information abstracted from medical records was matched by injury scene address to structural and geologic information. Data from the Los Angeles County Assessor’s 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. building inventory, Los Angeles City building inspection data, the United States Geological Survey, the California Governor’s Office of Emergency Services (OES), and medical record data were incorporated in a multivariate analysis. Proportional polytomous logistic regression and dichotomous logistic regression techniques were used to estimate epidemiologic risk of more severe injury associated with specific demographic, mechanical, structural, and geologic elements after the Northridge earthquake. Results are presented in Chapter V Theories regarding the stability o f specific structures and estimates of site-specific geologic activity are investigated in order to help confirm or disprove their contributions in risk estimation. Additionally, injury-specific characteristics are investigated regarding their contributions to risk estimation. A discussion section is presented in Chapter VI. Results of this study are compared with other earthquake-related injury studies. Deficiencies in existing databases are identified, and suggested improvements in these databases are also presented. Additionally, existing study results will be compared with hypothetical results predicted from an engineering model. By comparing parameter estimates obtained using simulated data to parameter estimates obtained using real data, assumptions regarding injury patterns that are used in current models may be validated or calibrated. Finally, limitations of the study will be summarized and discussed. Final conclusions based on the results are presented in Chapter VII. These conclusions will be helpful to policy makers, engineers and health professionals in focusing prevention efforts for reducing earthquake-related injuries in future events. They will improve techniques of hazard assessment and disaster mitigation strategies by providing 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. more precise estimates of the mechanisms and severities of earthquake-related injuries, and by describing how and where people were injured during the earthquake. Theoretically, if injury events can be patterned from past earthquakes and incorporated into a comprehensive model, more realistic outcomes may be predicted for future earthquakes. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C H A P T E R II: R E V I E W OF T H E L I T E R A T U R E This dissertation is based on a multi-disciplinary approach to injury epidemiology. Therefore, a review of relevant literature from injury epidemiology, structural engineering, geology and seismology will be presented. The review of the literature in each field will include an orientation to the terms and measurements that will be used in subsequent analyses of data, as well as a critical review o f pertinent studies, methodologies and standards or established guidelines for measuring various parameters. 1.00. INJURY EPIDEMIOLOGY Over the past twenty years, a sophisticated science has developed in the field of injury prevention. Researchers from diverse fields including physics, engineering, biology, public health, and the military have shared knowledge and contributed independently to this discipline. Collaborative efforts have yielded standard research tools including terms to describe injuries, codes for grouping the mechanisms or causes of injuries, and quantitative methods for measuring injury severity. Through epidemiologic research, measurement of the mechanical or physical displacement of energy that results in injuries has helped identify potential junctures for intervention. Injuries such as those caused by vehicular collisions, drownings, and violence are now viewed as understandable and preventable public health problems rather than isolated 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. random acts of fate. Additionally, surveillance of injuries allows application of focused interventions by identifying characteristics of populations at increased risk for specific injuries. 1.01. Pioneers in Iniurv Prevention. Measures of injury severity for epidemiological research evolved from the study of injuries resulting from airplane and automobile collisions. Hugh DeHaven was one of the first researchers to recognize the importance of the structural environment as an opportunity for intervention. DeHaven studied cases in which individuals plunged up to 150 feet in free fall without sustaining injuries. His findings, published in 1942, demonstrated that structural elements may be positioned in the environment to provide a means to control deceleration and redistribute energy forces across the human body thus enhancing survival (The National Committee for Injury Prevention and Control, 1989; DeHaven, 1942). Years later in 1961, an experimental psychologist, James J. Gibson, developed a method for classifying the exchange of energy that results in injury. He described this physical exchange as either mechanical, thermal, radiant, chemical, or electrical (Gibson, 1961). Dr. William Haddon, Jr., working for the New York State Department of Health, revised Gibson’s energy-transfer analysis to include physical elements whose absence also resulted in injury. For example, frostbite results from insufficient heat (Control, 1989). Haddon contributed one o f the most useful tools in injury epidemiology to date: the phase factor matrix, or Haddon’s Matrix (Haddon, 1980). Each dimension of this matrix decomposes an injury incident to its epidemiologic components. These 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. components include the host, the agent (or pseudo-vector), and the environment, as they interact over time and contribute toward injury. The actual agent o f injury is implicitly identified as energy transfer, but the matrix denotes a pseudo-vector that is the vehicle or mechanism through which energy is displaced. An example of a Haddon matrix for a vehicular injury is shown in Table 1. Specific factors are listed in relation to the element of time (pre-crash, crash, and post crash). By focusing on the cells within this type o f matrix, epidemiologists working with engineers and physicians have been able to reduce the frequency and severity of injuries, and improve recovery from injuries. For example, during the pre-crash phase o f the matrix, if a driver has eye disease, corrective lenses may be prescribed to compensate for vision problems and legislation restricts an individual’s driving privileges to those times when lenses are used. During the crash phase, the vehicle size and velocity will determine the momentum that will be transferred during the crash. Similarly, characteristics of the object with which the vehicle collides (density, height, and weight) will effect the impact of energy transfer, and characteristics of restraints available for occupants of the vehicle (lap belts, shoulder harnesses, and airbags) will effect how this energy transfer impacts the occupants. During the post-crash phase, fuel system integrity may effect whether a fire complicates existing damage and injury. Additionally, the age and health condition of the occupants may be associated with survival from injuries, and the distance and quality of emergency medical care will determine whether severe injuries will be treated within a time-frame that will enhance the likelihood of survival. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1. Example o f Haddon Matrix for Automobile Collision * Phase Host Factors (Human) V ector Factors (Vehicle) Physical Environment Socioeconom ic Environment Pre-Crash Eye disease in driver Inebriation Experience Travel time for this incident Wear on brakes & tires Center of gravity Tendency to jackkmfe Speed capability Characteristics of load Ease of control Visibility of road hazards Road curvature Surface friction Type of road Type of intersection Signage & road signals Attitudes about alcohol Laws related to licensing Speed limits Support for injury prevention efforts Enforcement of laws Number of passengers Crash Use of safety belts Use of air bags Osteoporosis Size of passengers Use of child safety seats Vehicle size & velocity Anti-lock brakes Load containment Placement of contact surfaces Hardness & sharpness of contact surfaces Guard rails Characteristics of fixed objects Median barriers Roadside embankments Recovery areas Attitudes about safety belt use Laws about safety belt use Laws about air bags Motorcycle helmet laws Enforcement of laws Post-Crash Age of injured patient Fuel system integrity Distance from emergency medical care Quality of emergency medical care Availability of communication to request help Rehabilitation programs Support for trauma systems Pre-existing health problems of injured patient Training of EMS personnel Ratio of EMS personnel to injured patients ‘Adapted from Injury Prevention: Meeting the Challenge, American Journal of Preventive Medicine, supplement to Volume 5 :3 ,19B9. o o 1.02. Injury Severity Measurements. Multidisciplinary teams of physicists, engineers, biologists, public health, and military professionals have collaborated to develop scales to standardize injury measurement. In 1971, the first Abbreviated Injury Scale (AIS) was published (Committee on Medical Aspects of Automotive Safety, 1971). The American Medical Association, the Association for the Advancement of Automotive Medicine, and the Society of Automotive Engineers sponsored this project. Representatives from each of these organizations, as well as about 35 other specialists from the previously mentioned disciplines were involved in the process. The AIS grouped patients and assigned injury scores based on the body region of injury (head, face, neck, thorax, abdomen, spine, upper extremity, lower extremity, and unspecified body location), type of anatomic structure affected (whole area (skin), vessels, nerves, organs (including muscles and ligaments), skeletal (including joints), and head), and the injury-specific level of effect (i.e., minor, moderate, serious, or severe). The scale helped standardize the terminology used to describe injuries, and to provide a quantitative method for ranking injuries by severity in order to compare them. The scale was used mainly by researchers investigating the physics of vehicular crashes for presenting summary descriptive statistics on related injuries. The AIS was developed to measure the severity of the injury itself rather than the consequences o f the injury. The scale was modified on a fairly regular annual basis, and by 1976 listed more than 500 injury descriptions (American Association for Automotive Medicine, 1976; American Association for Automotive Medicine, 1980; American Association for 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Automotive Medicine, 1985). However, in the 1980s with the advent of trauma centers, the AIS was expanded to include more detail on penetrating trauma and more clinically useful injury descriptions. Some of the clinical descriptions reflected the immediate physiological consequences of the injury (loss of consciousness, amount of hemorrhage, pneumothorax), but, consistent with the principles upon which the AIS was developed, did not measure impairments or disabilities resulting from the injury. These changes were included in the 1990 revision that was used for this dissertation (Association for the Advancement of Automotive Medicine, 1990). The 1990 revision assigns a 7-digit code based on a dictionary of injuries. The dictionary is organized by body region of injury. The code is determined by body region (same categories as specified earlier), type o f anatomic structure (same categories as specified earlier), specific anatomic structure or nature (whole area amputations, burns, crushes, etc.; loss of consciousness, concussion, level of consciousness; three categories for spine: cervical, thoracic, lumbar), level o f injury within a specific body region and anatomic structure, and a 6-level severity code (l=minor, 2=moderate, 3=serious, 4=severe, 5=critical, 6=maximum). An AIS code of zero denotes a non-injury, and an AIS code of 6 is considered a non-survivable injury. 1.03. Assessment of Multiple Injuries. Although the AIS was shown to be correlated with death from injuries that were life-threatening (AIS > 3), the system does not allow for a combined assessment of the effects of multiple injuries. However, a landmark injury epidemiology study found that the AIS had a non-linear relationship with survival (Baker, O'Neill, Haddon, & Long, 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1974). Based on this study, a method was developed to assess overall severity of injury for patients suffering multiple injuries, and to adjust for this non-linear relationship. This method is called the Injury Severity Score (ISS), and is calculated by summing the squares of the highest AIS scores from up to three different body locations. The ISS ranges from 1-76, with ISS=76 denoting a non-survivable injury. Additionally, an ISS=99 indicates that insufficient information was available to score injury severity. The AIS and ISS serve as general indices of severity, and are also useful as planning aids for medical management of patients. For example, they have been used for the evaluation of the impact of time delays in treating injuries (Noji, Jones, Smith, & Krimgold, 1989). Specifically, complications are associated with delayed treatment of injuries of varying severity. A mild injury that is left untreated may become a severe injury, and a severe injury left untreated may become fatal. All injury severity scales are limited by inadequacies of available information and by the individuals responsible for extracting and interpreting medical information. Some of the limitations of individual coding preferences have been reduced by increasing the dictionary of injuries from 75 descriptions in 1971 to over 2,000 in 1990. Additionally, training is available for those who code injuries. However, little can be done to overcome limitations o f available information. Injuries described as ‘multiple cuts and abrasions’ or ‘severe lacerations to the trunk’ provide insufficient information to code. Other complications to coding involve contradictions from different information sources (i.e., a nurse may record a given injury as a hand fracture and a physician may record the same injury as a foot fracture). Presumably, some diagnostic confirmation such as an 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X-ray is available within the medical record, or a follow-up visit may provide confirmation of previously recorded data. In cases where this is not available and contradictory information is recorded, the observation is essentially useless because it is not definitive. 1.04. Coding of Injury Mechanism. The primary method of coding mechanism (or cause) of injury is application of the International Classification of Diseases, Ninth Revision (ICD-9) External Cause codes (called E-codes) (Karaffa. 1993). Disease and diagnostic coding were initially developed primarily for documenting causes of mortality. The activity dates to the 17th century in the London Bills o f Mortality (Seare, Speirs, & Driscoll, 1993). This method of surveillance was developed and applied to other nationalities over several centuries. The World Health Organization (WHO) published the first version of the International Classification of Diseases (ICD) in 1948. The intended purpose of the ICD was for surveillance o f global causes of morbidity and mortality. This system is an important tool for epidemiologists, but it has also gained notoriety through applications to health care policy. For example, a mandate was passed in 1988 requiring that physicians provide ICD-9 codes for all Medicare claims submitted. The use of E-Codes has currently impacted issues relating to the effectiveness of clinical decision-making, identification of additional sources of reimbursement to reduce institutional financial loss, assessment of effectiveness of interventions, and to focus prevention strategies. (Alberts, Goldie, Edler, & Svanstrom, 1991; Irving, Norton, & Langley, 1994; Langley, 1995; Payne & Waller, 1989; Ribbeck, 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Runge, Thomason, & Baker. 1992; Rivara, Morgan, Bergman, & Maier, 1990; Runyan, Bowling, & Bangdiwala, 1992; Williams, Furbee, Prescott, & Paulson, 1995) In 1989, a conversion between AIS and ICD-9 codes based on hospital discharge data was published by MacKenzie, et al. (MacKenzie, Steinwachs, & Shankar, 1989). If diagnoses are coded with ICD-9 codes, it is possible (by applying a list of assumptions) to translate those diagnoses into body regions and severity codes. However, not all injuries are coded with ICD-9 codes, and there is some concern about the validity and reliability of this coding. In California, the State Department of Health mandates coding of hospital discharge data by ICD-9 categories, but also mandates reporting within 30 days of discharge. How this time constraint effects coding is not known. Additionally, inter-rater and intra-rater reliability of coding is not monitored. Therefore, unless standard software is used uniformly throughout the state, reporting is likely to vary between and within hospitals. Another obstacle to using the conversion table is that currently only hospital discharge data is coded with ICD-9 classifications. Since only a small proportion of injuries actually result in hospitalizations, the majority of injuries are not coded with ICD-9 codes. 1.05. Methodological Challenges of Disaster-Related Iniurv Epidemiology. Lively controversy exists surrounding many aspects of injury epidemiology. These discussions are important, especially when data are used for planning and management. Many issues focus on standard epidemiologic methods. Topics that are debated include how the population at risk for injuries is identified, how injured people are identified, how factors that contribute to risk of injury are identified, measurement of 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. risk factors, quantifying injury and severity, and limitations, strengths, and standardization of various sources of data. As this chapter is developed, selected population sampling methodologies, data collection techniques, sources of data and data collection time-frames will be discussed. It is important to remember while reading these pages that this review is not exhaustive. Examples are presented that apply mainly to existing studies of earthquake-related injuries. Regardless of the sampling methodology that is discussed, (i.e., cluster, population-based, convenience), the way in which the data is collected may vary (i.e., personal interview, medical record review, observation), and the time-frame in which the data is collected may also vary (i.e., 2-6 days after the event, 2 months after the event, 2 years after the event). Therefore caution is issued regarding implications of the examples presented, since they are not intended to cover all possible combinations of sampling methodologies, data collection techniques, sources of data, and data collection time frames. Disaster epidemiology is a subset of injury epidemiology, and the epidemiology of earthquake-related injuries is a special subset of disaster epidemiology. Earthquakes have the potential to be quite powerful with wide geographic areas of impact. However, they are relatively rare events compared to other weather-related disasters such as floods or windstorms. Additionally, it is not possible to predict the likelihood o f an earthquake in the same manner as weather-related disasters. The combined elements of surprise, infrequency, strength, and wide-dissemination o f force provide an environment that is particularly difficult to study scientifically. Practical public health issues such as 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. preparation for earthquakes, recommendations for responses during earthquakes, and mitigation efforts to reduce damage and injury in future earthquakes are especially challenging. Damage subsequent to a disaster often necessitates rapid mobilization of equipment, supplies, and funds. This mobilization should be based not only on what is thought to be necessary, but also on experience. As will be mentioned in Chapter 3 of this dissertation, data varies depending on the manner of information retrieval, the time period after the earthquake, and the criteria for inclusion in studies o f damage and injury caused by the earthquake. If damage can be anticipated or if protective measures are identified prior to a disaster, persons at risk for injury can be better prepared in order to reduce the likelihood o f injury. If disaster planning and management are not based on sound scientific investigations, flawed assessment at different stages of a disaster may result in inappropriate allocation of resources. This, in turn, will limit attempts at mitigation and may result in the neediest populations receiving inappropriate or insufficient aid. An important issue in earthquake-related injuries addresses secular trends. This has been an epidemiological issue in disaster-related injury for more than fifteen years (Logue, Melick, & Hansen, 1981). Reported trends have been difficult to interpret due to variation in methodology. For example, declining death to injury ratios apparent over time could reflect actual decreases in mortality due to improved disaster relief or preparedness resulting in fewer deaths and injuries. Alternatively, the trends could be 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. due to more complete reporting of less severe injuries, resulting in an increase in the denominator (injuries) while the numerator (deaths) stays constant (Logue et al., 1981). Another issue regarding reporting of earthquake-related injuries addresses variation in case definition between investigators or studies. For example, reports of deaths due to the Northridge earthquake have varied depending on the case definition. One investigator with primary training in architecture reported 72 fatalities due to earthquake-related injuries (Durkin, 1995). In contrast, the Director of the United States Geological Survey reported 57 fatalities in a testimony before the Subcommittee on Science, Technology, and Space at the United States Senate (Eaton, 1994). However, Peek-Asa and colleagues, using standard epidemiologic methods (i.e., review of coroner’s reports for earthquake-related deaths), reported 33 fatal injuries resulting from the earthquake (Peek-Asa et al., 1998). Variation in numbers of events results from variation in case definitions, sources of data, time periods of reference, time periods elapsed since the disaster, and other methodological issues. Standardization of these methodological issues has not been well established and in some studies, the methodology is only vaguely described. 1.06. Identification of Cases. Selection of cases for earthquake-related injuries is not as straightforward as might be expected. Although an injury (by definition) must have some physical manifestation and mechanical cause, many earthquake injury studies have included heart attacks and other disease-related conditions (Durkin, 1995; Durkin & Thiel, 1991; Durkin & Thiel, 1992; Holmes & Somers, 1996; Noji, 1997). Certainly an individual who is in 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. poor health may be more likely to experience more complications during or immediately after an earthquake than a healthy individual. However, to assume that a death due to a coronary event (for example) is attributable to the earthquake requires knowing that an individual would have survived beyond that date had there not been an earthquake. This information is not readily available. Researchers investigating cardiac events after the Northridge earthquake have approached the problem of case identification using different sources of data (Los Angeles County Coroner’s Reports and death certificates for Los Angeles County obtained from the State Department of Health Services) (Kloner, Loer, Poole, & Perritt, 1997; Leor & Kloner, 1996; Leor, Poole, & Kloner, 1996). Findings based on one data set are confirmed in another, and this type of methodological approach strengthens the validity of relationships that are described. In these studies, both sources of data showed what is described as a compensatory effect of specific cardiac events that was apparent after the Northridge earthquake (Kloner et al., 1997; Leor & Kloner, 1996; Leor et al., 1996). The meaning of a compensatory effect in the context of this study is that those individuals who died on the day of the earthquake due to cardiac failure probably would have died within the next few days had there not been an earthquake. An important limitation to the study based on Coroner’s reports (Leor et al., 1996) is due to the source of data and selection bias. Under normal conditions, mortality due to atherosclerotic cardiovascular disease does not come to the attention of the coroner. Only if a death occurs in suspicious circumstances (suspected foul play), if the patient has not been under the care of a physician, or if the family requests an autopsy, does the coroner 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. routinely perform an autopsy. Therefore, baseline or expected fatalities due to this cause are probably underestimated by studying trends in the coroner’s data. Similarly, an important limitation to the study based on death certificates (Kloner et al., 1997) is also due to the source of data and selection bias. Death certificates allow for one primary cause of death as well as multiple underlying causes o f death. Although there are standards for coding causes of death, variation exists between nosologists regarding adherence to these rules. Overall, both studies (Kloner et al., 1997; Leor et al., 1996) show some methodological weaknesses independently. However, when the study results are considered simultaneously, the findings are complementary, and methodological strength is gained. Assuming one has a clear definition of mechanically-caused injuries, there are many problems enumerating these, particularly in times of disaster. One method of identifying injury cases is to concentrate on a subset of injuries. Researchers have studied hospitalized injuries and deaths that are traceable through mandatory reports (Peek-Asa et al., 1998). A limitation with this approach is that emergency department visits and injuries treated outside of the hospital are not included. By reviewing Figure 1. The Injury Pyramid (Fingerhut & Warner, 1997), one can see that deaths and hospitalizations accounted for less than 3% of all reported injuries in the nation in 1995. These are usually the most severe injuries, and mechanisms that cause them may be very different than mechanisms that cause less serious injuries. Another limitation to using death and hospitalized injuries is due to the data sources. Medical records, hospital discharge data, and coroner’s reports may be incomplete or erroneous. Specific examples 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of these problems will be expanded upon later in this chapter in Section 1.16, Sources o f Injury Data. Fi&ure I. The Injury Pyramid from the Burden o f Injury: United States. 1995. (Source: Finger hut and Warner. 1997). Deaths 147.891 Hospital Discharges 2 .5 9 1 .0 0 0 Emergency Department Visits 3 6 .9 6 1 .0 0 0 Episodes of Injuries Reported 5 9 .1 2 7 .0 0 0 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Another method used to estimate the number of earthquake-related injuries is random telephonic sampling of the population (Bourque et al., 1997; Shoaf, Sareen, Nguyen, & Bourque, 1999; Shoaf & L.B., 1999). Methodological problems with this approach include the absence of diagnostic validity (since injuries are self-reported), recall bias, and misrepresentation of injury cases (non-sampling errors, over-sampling errors). Invalid reports of injury and recall bias are both types of information bias (Rothman, 1986). Misclassification of injuries due to self-reporting has not been studied well in the published disaster literature, but this is likely to be more problematic for respondents with no medical background or with less education than for others (Armenian, Melkonian, Noji, & Hovanesian, 1997). Recall bias has also not been studied well in disaster situations, but it is probably not a large problem in this context because the impact of the event was so memorable that individuals probably remembered better than in non-disaster situations (Shoaf, 1998; Wagner, 1999). If present, an example of recall bias would be an increase in reports of injuries in those whose housing sustained damage. The perception of damage within the house may influence individual perception of injury in such a way that those respondents whose homes were damaged ascertain injury more frequently than those whose homes were not damaged. Non-sampling and over-sampling errors may be associated with selection bias. It has been suggested that the statistical theory behind sampling for surveys is concerned more with the problem of sampling error than non-sampling error, therefore this emphasis may lead investigators to overlook the importance of non-sampling errors (Armitage & Berry, 1994). If these errors are not systematic, they will contribute to the 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. variability associated with point estimates (i.e., estimates of risk for a specified outcome due to one factor in reference to another), but they should cause no bias. However, if the errors are systematic, corresponding point estimates may be biased either away or towards the true mean value in the population. Armitage and Berry state that ‘One of the most important types of systematic error is that due to inadequate coverage of the sampling frame, either because of non cooperation by the individual or because the investigator finds it difficult to make the correct observations’ (Armitage & Berry, 1994, p. 182). If the relationship between exposure and disease is different for those who participate in the survey compared to those who are eligible but do not participate, selection bias is theoretically possible (Greenland, 1977; Rothman, 1986). Examples of individuals that might be eligible for a telephone survey but are not included are those who refuse to participate, those who have died, and those who cannot be reached for other reasons such as emigration, job-related working patterns, educational attainment, non-subscription to residential telephone services, disease, institutionalization, language barriers, and cultural preferences. In contrast, an example of those who might be over-represented in a telephonic-based sample includes those households with multiple telephone lines. Estimates of the number of people and characteristics of those who emigrated out of Los Angeles due to the earthquake are not available. However, it is likely that they are different in some systematic way than those who remained. Characteristics on which these individuals might show differences include socio-demographic factors (i.e., proximity to blood relatives, ethnicity, age, education, income, employment status and 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. occupation) injury patterns (i.e., more injuries per household, severity of injuries per household), and severity and type of housing damage. If these characteristics are different between emigrants and non-emigrants that were exposed to the earthquake, respondent injury patterns may not be representative o f all injury cases and estimates of relative risk will be biased. Similar logic may be applied to those individuals who have refused to participate in the study, who do not subscribe to residential telephone services, and those who were never at home when contact was attempted. It is possible that those households in these categories were different regarding exposure to the earthquake, demographic characteristics, exposure to potential risk factors, and subsequent injury patterns, than identified cases. Again, this would lead to biased estimates of relative risk based on available data. Several standard epidemiology texts emphasize the importance of attempting to assess the impact of non-sampling errors (Armenian & Shapiro, 1998; Armitage & Berry, 1994; Kelsey, Whittemore, Evans, & Thompson, 1996; Kleinbaum, Kupper, & Morgenstem, 1982; Rothman, 1986; Shlesselman, 1982). This can be accomplished by attempting to characterize non-respondents so that an epidemiologic judgment regarding the direction and magnitude of bias can be made based on knowledge of the subject matter and observed differences between participants and non-participants. Despite the limitations, however, telephone surveys have certain advantages over other methodologies. They are relatively inexpensive to administer, can be administered to a very large segment o f the nation (since 94% of the population subscribes to 22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. residential telephone services) and can give insight regarding how an event such as an earthquake impacts the community in general. One final methodological issue surrounding case definitions relates to direct and indirect attribution of an injury to an earthquake. Earthquake-related injuries may be subdivided into those that can be traced directly to the earthquake and those that are indirectly related to the earthquake. An example of a directly related injury is when an object topples during an earthquake, falls onto a person and injures that person. In comparison, a person that falls from a ladder and is injured while trying to replace ceiling tiles that were dislodged during an earthquake suffers from an indirect earthquake-related injury. Differentiating between the two is important for policy makers and health professionals. Policy makers may find that mobilization of resources is necessary to respond adequately to directly-related injuries, whereas this is not such an important issue in most indirectly-related injuries. Similarly, directly-related injuries may require modification or new recommendations from injury prevention experts, whereas indirectly-related injuries may be reduced using standard injury prevention strategies. 1.07. Assessment of the Population at Risk for Earthquake-Related Injuries. Another important methodological issue in epidemiologic investigations of earthquake-related injuries is assessment of the appropriate denominator for population- based rates. This denominator is defined as the population at risk for injuries due to the earthquake. If the appropriate denominator is not selected, rates are misleading (artificially high or low). If injury rates do not represent the population at risk, it is 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. inappropriate to make generalizations or informed decisions based on these rates. This is a very important public health issue. Traditional sampling techniques such as cluster-sampling have been used successfully in hurricane-related injury studies (Centers for Disease Control and Prevention, 1992; Hlady, Quenemoen. & Armenia-Cope, 1994). Cluster investigations usually collect data via questionnaire, which provides optimal assessment of much disaster-related and injury-specific information (see Section 1.12 for an example of this). Since the damage-area of hurricanes dissipates as the storm moves inland, and the eye of the storm is relatively easy to track over time and geographically, the areas at risk for damage are more easily identified. These areas are usually coastal communities in the path of the storm. Clusters of the population may be selected systematically from the impact area using a grid that is overlaid onto aerial photographs or maps of the area (Noji, 1997). However, epidemiologists who attempted to apply cluster-sampling to identify earthquake-related damage and casualties found the technique inadequate (Noji, 1997). Earthquake-related damage is not necessarily located in contiguous geographic areas. Instead, the ground-shaking produces pockets of damage. Additionally, ground-shaking may be attenuated by soil conditions and structural fragility. A Rapid Health Needs Assessment conducted after the Northridge Earthquake by the Los Angeles County Department of Health Services’ Acute Communicable Disease Unit reported that houses with severe damage were often surrounded by houses with only minor damage (Los Angeles County Department of Health Services Disease Control Program, 1994). The 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. same report also acknowledged that the greatest limitation to the study was the methodology used to select the study area. This methodology was a simplified cluster sampling technique that was modeled after an investigation done by the CDC’s Disaster Assessment and Epidemiology Section Rapid Health Needs Assessment during Hurricane Andrew. The cluster investigation was problematic because it was not possible to identify areas of high impact during the first few days after the earthquake, therefore it was thought that the population most affected was underrepresented resulting in artificially low estimates of earthquake-related injury. It has been noted elsewhere in the literature that the use of a cluster-sampling to assess damage after an earthquake may result in missed assessment of high-impact areas that will subsequently lead to underestimates in the overall earthquake-related damage and injury (Noji, 1994a; Noji, 1997). To complicate matters, moderate and large earthquakes often have a wide geographic area o f impact. Therefore, the population at risk for earthquake-related injuries often crosses political boundaries. For example, the Northridge earthquake affected five counties (Los Angeles, Ventura, Orange, Riverside and San Bernardino). Conducting investigations across county boundaries requires either consensual collaboration or politically superior intervention that may result in alienated participation and biased results. Therefore these types of collaborations are rarely conducted. Even when multi-county collaboration is not required, for a county the size of Los Angeles (approximately 4,000 square miles), the goal of identifying all injuries is unrealistic. It is more reasonable to expect that injuries might be identified by smaller 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. political boundaries (city, zip code, census tract, and health district). However, these boundaries do not coincide with geologic boundaries of the earthquake. This is important because geologic and structural characteristics of the environment are acknowledged in the literature to be the main risk factors for serious earthquake-related injuries and death (Kircher, Nassar, Kustu, & Holmes, 1997b; Noji, 1997; Peek-Asa et al., 1998). Therefore calculating rates for selected geo-political subsets (i.e., census tracts) may overestimate or underestimate the overall impact on the county. 1.08. Calculating Rates of Earthouake-Related Injuries. In order to determine whether an increase in injuries is due to an earthquake (or any other disaster), existing information (baseline data) on the rates o f injuries in the population must be available. Once existing injury rates are available, one can statistically test for increases in specific mechanisms of injury beyond what might be expected under usual conditions. However, baseline injury morbidity data are rarely available in Southern California, even though disasters occur nearly every year. If surveillance is established, however, it is important that the system have high sensitivity (the ability to accurately detect true cases) in order to provide some assurance regarding reports of increase or decrease (Noji, 1997). If sensitivity is low, an apparent increase could be due to better detection o f usual injuries rather than injuries from the event of interest (i.e., the earthquake). Surveillance programs and disease registries also have limitations. For example, in reportable disease surveillance systems, underreporting is well documented; therefore these registries may be viewed as representing a minimum number of outcomes (Kimball, 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thacker, & Levy, 1980; Noji, 1997; Vogt, Clark, & Kappel, 1986). In spite of this limitation, minimal documentation is better than none, and injury surveillance is clearly an area that needs to be augmented if policy makers are serious about reducing losses and providing appropriate care after disasters such as earthquakes. The study time-frame may also influence calculation of rates. For example, a study that investigates injuries during the first 72 hours after an earthquake may capture most of the directly-related injuries. However, a study that investigates injuries during the 30-day period after the event will capture both directly- and indirectly-related injuries. Additionally, persons injured directly from the earthquake that delayed seeking treatment might also be detected if the time-frame is extended. It is therefore important for researchers to clearly delineate the study time-frame along with other methodological processes. Rates of injury may be calculated by hospital. Problems with this approach are based on determining the appropriate catchment area for the hospital or identifying those patients that present during a disaster that would not usually present at a specific facility. Where a patient usually seeks health care is not a bit of information that is recorded in medical records. Even though this is not well-documented, it is logical to assume that healthcare-seeking behavior of injured persons was probably altered after the Northridge earthquake. Some hospitals were closed, sections of freeways collapsed, explosions occurred under city streets, and electrical and communication outages all probably had an adverse effect on access to facilities and traffic patterns in general. This is supported by information gleaned through interviews with employees at hospitals in high-impact areas 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Bourque et al., 1997). Anecdotes have been noted regarding hospital employees that were injured at home, who, under normal conditions, would have sought care, but in the aftermath of the earthquake chose to stay at home because the damage was deemed more serious than the injury. Accounts of people seeking care from unusual sources such as veterinarians and neighbors have also been noted (Shoaf, 1998). Therefore hospital- specific injury rates from times of disaster may not be comparable to baseline rates because they may not represent the hospital’s usual clientele. Additionally, the behavior of health-care providers will impact hospital-specific rates. Again, although these events are not well documented, it is logical to assume that provider care was altered after the Northridge earthquake. Personal accounts from practitioners on duty in emergency departments in high-impact areas noted that patients with rather severe injuries who would have been admitted under normal conditions were treated and released, either because of facility inadequacies or urging from the patient (Mahue & Weiss, 1996b). Peek-Asa confirms this finding and reports that patients admitted to facilities for treatment of earthquake-related injuries were (as a group) less serious than what would have been expected under usual conditions(Peek-Asa et al., 1998). Additionally, the study findings indicate that many of those injuries that resulted in admission were due to delayed care-seeking behavior by the patient. This apparent combination of altered care-giving and care-seeking behavior limits the usefulness of hospital-specific rates. 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.09. Earthquake-Related Injuries; Guatemala. 1976. One epidemiologic study of earthquake-related injuries dates back to the February 4, 1976 earthquake in Guatemala. This devastating earthquake measured 7.5 on the Richter scale, occurred at 3:05 in the morning, and was associated with 22,778 deaths and 76,504 injuries. Investigators from the Centers for Disease Control and Prevention in Atlanta, Georgia (then called the Center for Disease Control) and from the Institute of Nutrition for Central America and Panama (INCAP) conducted a cross-sectional field survey investigating patterns of injury and disease in the village of Santa Maria Cauque (Glass et al.. 1977). The study included surveillance for infectious diseases, descriptions of the demographic characteristics of patients sustaining severe and mortal injuries, and administration of surveys to examine environmental and behavioral factors associated with physical injury. This study identified demographic characteristics (age, gender, birth order of children and size of family), as well as cultural practices (sleeping patterns), and structural characteristics (adobe and unreinforced concrete) that appeared to be associated with more severe injury. Children less than one year of age were at lower risk for severe injury. It was hypothesized that this might be due to the cultural practice of the youngest child sleeping with the mother. Indirect support for this theory was found by investigating the outcome o f mother/child sleeping pairs. In most cases the child sustained a similar if not identical outcome to the mother when they slept together. Age-specific mortality rates were highest for the young and the old, as expected since these extreme age groups are more vulnerable to physical stress. Additionally, risk 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of injury was greater for women than for men in all age groups. The authors point out similarities between this injury pattern and the pattern o f injury for hip fractures in the United States, as well as similarities between this pattern and injury patterns observed in other disasters (de Ville de Goyet, del Cid, Romero, Jeannee, & Lechat, 1976; Glass et al., 1977; Sommer & Mosley, 1970). It is suggested that age-dependent physical stamina is strongly associated with survival in life-threatening disaster situations. The authors also found a two-fold risk o f injury for families with 7 or more members compared to smaller families. This increased risk could be explained in part by an increased pool for potential victims. Additionally, larger families were probably more likely to include individuals from the oldest and younger age categories since generations of families tended to reside in the same abode. Adjusting for age may have controlled this apparent effect, but this was not discussed in the article. Structural characteristics that were associated with increased risk of severe injury included buildings made of adobe and unreinforced concrete. Doorjambs (doorways) were not associated with protection from injury in adobe homes although these areas were thought to be more structurally sound. The size of the structure and the number of rooms were also not associated with increased risk of severe injury, which was contrary to common beliefs. Additionally, no protective value was noted for sleeping in comers or near doors, although those areas were thought to be more structurally sound. Although the study has clear definitions of physical injury, no information is presented regarding how the authors determine that injuries were due to the earthquake. Additionally, mortality was measured globally rather than by cause. Therefore, no cause- 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. specific differentiation is possible. Additionally, post-event mortality was not compared to baseline mortality. Therefore, excess mortality that might be due to the earthquake is not reported. 1.10. Earthauake-Related Injuries: Whittier Narrows. 1987. Shoaf and colleagues report on a population-based telephone survey of residents of Los Angeles County subsequent to the 1987 Whittier Narrows earthquake (Shoaf et al., 1999). The total weighted sample size for this survey was 1,309 respondents. The survey inquired about household damage, injuries to members of the household, utility outages, evacuation and information-seeking behaviors during and after the earthquake, and general socio-demographic characteristics of the household and the respondent. This event was a relatively minor earthquake (5.9 on the Richter scale), with ground-shaking lasting less than ten seconds, and most of the communities were undamaged. Only eleven of the respondents reported injuries (< 1%). O f these 11 injuries, 23% were emotional injuries, 11% were of unknown type, 11% were aggravations of prior conditions or injuries, 11% were cuts, and 44% were minor head injuries. The causes of these 11 injuries included mental (21%), non-structural objects falling (51%), falling during the event (19%), and unknown mechanisms of injury (10%). Although this survey reveals a small number of injuries, the authors note that if the proportion of injuries from this population-based sample is applied to 1990 census data for Los Angeles County, approximately 24,000 households in the Whittier Narrows earthquake would have had at least one injured occupant. An important finding in this survey is that approximately half of the identified injuries were caused by displacement 31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of objects that were not part of the structure itself. These types of injuries are easily prevented by anchoring or bolting objects so that they are less likely to fall, and by installing safety latches on cupboards and cabinets so the contents are contained. Another report of the Whittier Narrows survey reveals initial behavioral responses that may have been effective protective actions for injury reduction (Goltz, Russell, & Bourque, 1992). Since the event occurred at 7:42 in the morning, the morning commuter traffic was in peak flow. The modal response for those people driving was to pull to the side of the road and stop. The modal response for those who were at work or at home was to take cover in a doorway, hall or under furniture. Response actions were associated with fear, the presence and identity of other people, and gender of the respondent for those who were at home when the earthquake struck. Women sought cover at home more than men. Men who reported being very frightened were twice as likely to take cover than men who were less frightened. The relationship between fear and taking cover was strongest when other adults were not present. Additionally, those who were very fearful and reported having children present in the home were twice as likely to take cover than those who were not as fearful and had children in the home. For those who were at work during the main shock, the response behavior was associated with fear, gender, and the presence of others. When others were present in the workplace, those who were more fearful were twice as likely to take cover than those who were less fearful. Women with high levels of fear were more than twice as likely to take cover than women with lower levels o f fear. Seventy percent of those who were at 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. work during the earthquake remained at work rather than leaving to check on family members or property in the two hours following the earthquake. An important finding in this study relates to the role that fear plays in behavioral protective responses during an earthquake. The preconceived notions that fear would induce flight responses leading to injury and that people would take evasive actions regardless of fear were not confirmed. On the contrary, fear was not associated with inappropriate responses. Additionally, it was found that as expressed fear increased, so did self-protective activity. These findings are important for health professionals and community-based information centers to consider when advising individuals on appropriate actions. Those that are less fearful might be targeted for improving their behavioral responses, and those that are more fearful might be encouraged to be prepared to respond appropriately (pull to the side of the road and stop, take cover under heavy furniture, avoid catching or being hit by falling objects). f.ll. Other Studies of Earthquake-Related Injuries: Northern Armenia. 1988. An earthquake that registered 6.9 on the Richter scale devastated Northern Armenia on December 7, 1988 at 11:41 in the morning. This earthquake has been the focus of well-planned ongoing epidemiologic studies conducted by Armenian and colleagues (Armenian, Melkonian, & Hovanesian, 1998; Armenian et al., 1997; Armenian, Noji, & Oganesian, 1992). Researchers from Johns Hopkins University, the CDC, and the Armenia Ministry of Health have collaborated on a case-control study and a cohort study that investigated the short and long-term effects of the earthquake. These 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. three studies represent the most complete epidemiologic analysis of adverse health effects from an earthquake to date. Case-Control Study. The goal of the case-control study was to attempt to identify predictors of injury among hospitalized individuals after the earthquake. Cases and controls were clearly defined and a standardized interview was administered to both. Cases and controls were attempted to be matched by neighborhood, sex and age (± 5 years). However this was not possible and the analysis was unmatched. Multivariate logistic regression was used to estimate risk factors for those who were hospitalized for injuries (n=189) compared with those who were hospitalized without injuries (n-156). Risk factors that were identified included being inside a building at the time of the earthquake compared with being outside a building. Furthermore, being in a building with five or more floors compared to being in a building with fewer floors was identified as a risk factor. Within floors o f buildings, people who were on higher floors o f multi story buildings were at increasing risk for injury relative to those who were located on lower floors. Analysis also included investigation of first reactions of individuals within buildings. Those who moved at all within the building compared with those who ran out of the building were at increased risk for injury. Limitations o f this study include problems using hospitalized cases and controls. These cases represent only the most severe injuries; therefore they are not representative of the general population. In fact, some of the controls were injured but not hospitalized, hence these individuals were analyzed as a separate group of minor injuries. Similarly, hospital controls are not representative of the general population, so estimates o f relative 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. risk may be higher or lower than expected if non-hospitalized controls were used. However, the study does acknowledge that the selection of cases and controls is not ideal. Additionally, the authors point out that it is quite difficult to obtain a random sample from a community that has been destroyed by a disaster. The authors also acknowledge that although the findings in general suggest that running out of a building was protective for injury, the study does not differentiate between those who could run out of the building and those who could not because of their injuries. Specifically, the authors point out that the injury itself could determine an individual’s behavior. The issue of attempting versus succeeding in escaping a structure as a possible risk factor is not addressed directly, but the authors acknowledge that in most of the multi-story structures, the stair-wells were often the least structurally sound part of the building. If attempting to exit via a stairwell resulted in an injury due to structural collapse of the stairwell, this action might be discouraged in future earthquakes. The authors also warn against generalizing these findings to geopolitical areas that have developed and enforced construction codes intended to lower seismic damage. The Armenian government has fewer construction standards and regulations compared to California and Japan. Additionally, crowding of tall buildings that is characteristic of urban centers may create an environment that is more hazardous to those outside the building than inside because of falling external structural debris. Cohort Approach: Short-Term Outcomes. This study was conducted primarily to investigate determinants of death and injury immediately after the earthquake (Armenian et al., 1997). The cohort was defined as employees of the Ministry of Health living in the 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. earthquake region on the day prior to the earthquake as well as their first degree family members. This cohort was identified through listings (payroll and personnel) that were readily available at the Ministry of Health. The cohort was thought to be representative of a broad sector of the population except that they probably had better access to medical care than the general population. Not all individuals were traceable, but participation in the study is impressive and includes 32,743 people from the affected area. Multivariate logistic regression was used to analyze risk factors. Details o f the interview questions were not presented, but analytic results were presented for demographic characteristics (age, gender and geopolitical region), location at the moment of the earthquake, and characteristics of the building (building type, height, and location of participant within the building at the moment of the earthquake). Crude death rates were higher for those aged 61 or older compared to all other ten-year age groups. Crude injury rates were higher in females compared to males, and were lower in those aged ten years or less and older than age 70, compared to all other ten-year age groups. Additionally, those who were inside a building at the moment of impact had 2.3 times the risk for injury relative to those who were outside a building, a confirmation of the finding reported in the preceding case-control study. Those who were in taller buildings were reported to be at increased risk for death relative to non- injury (adjusting for age, geopolitical location, structure type, and floor location). However, this risk was not observed for injuries relative to non-injuries. The authors report that a separate analysis comparing deaths to injuries showed that location within the upper floors of buildings in conjunction with building height were predictors of death. 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. These results would be interesting to see, but they are not presented. Another risk factor for injury and death was panel-type construction compared to other types of construction. This study is a fine contribution to disaster epidemiology. The cohort approach has obvious advantages over case-control and cross-sectional studies (reduced selection bias and better representation of the population at risk for injury). The 95% confidence intervals (95% Cl) around most of the estimates of relative risk are quite narrow, leading one to believe that these estimates may actually be very close to the true value. However some unexplained oddities are noted in the findings. For example, the relative risk for death compared to non-injury for individuals on floors five or more (relative to floor 1) was 1.7 (95% Cl: 1.3-2.3), controlling for age, building height, and construction type. In contrast, the relative risk for death versus non-injury for individuals on floors 2-4 (relative to floor 1) was 1.9 (95% Cl: 1.6-2.4), controlling for age, building height, and construction type. This seems counterintuitive to the authors’ statement of association between increased risk for death compared to injury with building height and floor location within the building. It seems that this same type of relationship should also be apparent when comparing deaths to non-injuries. Additionally, the confidence interval around the relative risk estimate for death in buildings higher than 8 stories is very wide, indicating high variability around this estimate, and potentially wide dispersion from the true value o f the parameter. To complicate this finding, the elevated risk was not confirmed for injured persons relative to non-injured persons. This is also counter intuitive, and attempts to explain this were not presented. 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. High estimates of relative risk and wide variation are also reported by geopolitical region, particularly for deaths relative to non-injuries. This could be due to the use of ‘Other Regions’ as a reference category. This category (‘Other Regions’) contained the largest fraction of the cohort (19,041 of 32,743 individuals in the cohort) compared to the ‘Gumri Region’, an urban center with 11,581 individuals, and the ‘Spitak Region’, located near the epicenter, with 2,121 individuals. However, it contained the smallest fraction of deaths (31 compared to Gumri’s 561 and Spitak’s 239) and injuries (278 compared to Gumri’s 682 and Spitak’s 494). The use of the ‘Other’ category as a reference value for estimating relative risk is probably the best option available, but it is limited due to imprecision in defining this category (‘Other Region’ includes all regions other them Gumri and Spitak) and the relative paucity of observed outcomes as a point of reference. Another aspect of this study that is not clearly explained is the manner that was used to determine earthquake-relatedness, as well as the time-period of interest for making this determination. The absence of a clear definition for earthquake-related deaths and injuries introduces the potential for the previously described problem of differentiating deaths due to natural causes from those that were due to the earthquake. The classic example of this is deaths from cardiac arrhythmia. Those people who were predisposed to death or injury because of preexisting health conditions are not clearly differentiated from those normal, healthy individuals that suffer from mechanical physical injuries that may have led to death. Additionally, the element of time is important when estimating relative risk. If the time period extends too far into the future, 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. it becomes difficult to differentiate directly-related injuries from indirectly-related injuries. These subtle differentiations are not clearly presented, but it is implied that the study outcomes were injuries and death caused by the earthquake itself as opposed to clean-up activities or exacerbation of pre-existing physical or mental conditions that resulted in injury or death. The authors emphasize that buildings in Armenia are structurally fragile, and note that many nine-story residential buildings that were wide-spread around the epicenter collapsed completely and were important causes of death in this earthquake. They also emphasized that determination of appropriate protective (behavioral) responses is probably dependent on specific construction types as well as the geopolitical environment including density of the affected population. The authors noted that available estimates of earthquake-related casualties are based on ‘superficial observations of limited technical and statistical validity. Therefore casualty extrapolations to other earthquakes and other geophysical settings have generally low credibility’ (Armenian et al., p. 811, 1997). Therefore, future study recommendations include obtaining more detail regarding building design, soil conditions, and the population at risk in individual buildings. Cohort Approach: Long-Term Outcomes. Armenian and colleagues also conducted a two-year prospective study to assess the relationship between cause-specific increased mortality and morbidity and personal loss and damage following the 1988 earthquake in Armenia (Armenian et al., 1998). Additionally, the investigators conducted a nested case-control study of heart disease, diabetes, arthritis, stroke, and hypertension. Injuries were intentionally omitted from the analysis, and the primary 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. outcomes of interest were incident cases of chronic disease. Injury could have been included in the study from the perspective of disability, dependence, and loss of productivity. It is possible that the authors have more publications pending that might further detail injury events. 1.12. Earthauake-Related Injuries: Loma Prieta. 1989. The Centers for Disease Control and Prevention published a brief report on earthquake-associated deaths due to the 1989 Loma Prieta earthquake in northern California (Centers for Disease Control and Prevention, 1989). This earthquake registered 7.1 on the Richter scale and occurred on October 17, 1989, at 5:04 p.m. The earthquake impacted seven counties (Alameda, Monterey, San Benito, San Francisco, San Mateo, Santa Clara, and Santa Cruz) and approximately 4.6 million people. No standard definition for determining earthquake-related deaths was used. Instead, county medical examiners and coroners reported the numbers of earthquake-related deaths in their jurisdictions from October 17 through October 31, 1989, and additional information about the victim and circumstances of the injury. This information is useful in describing the relatively rare event of mortality due to that earthquake. However, no information is provided regarding morbidity. Sixty-three deaths were reported. Sixty deaths were classified as directly-related to the earthquake, and three were assumed to be indirectly-related to the earthquake. Directly-related deaths included those due to collapse of elevated sections o f freeways (n=41), brick wall collapse onto automobiles (n=5), brick wall collapse onto individuals (n=4), dwelling collapse (n=3), falls on stairways (n=2), falls from towers (n=2), a 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. landslide on a coastal highway (n=l), and smoke inhalation from a gas fire (n=l). Ninety-five percent of the directly-related injuries were sustained within 2 minutes o f the earthquake. The rest were sustained within 8 hours of the earthquake. Indirectly-related deaths included a mortal wound from a firearm while directing traffic, motor-vehicle collision with a damaged bridge section, motor-vehicle collision with a horse on a highway, and carbon-monoxide poisoning due to exhaust from an emergency generator. Mortality rates are provided by county and range from 0 to 3.4 per 100,000 population (Centers for Disease Control and Prevention, 1989). Shoaf and colleagues conducted a population-based telephone survey o f the five- county area of the San Francisco Bay. The total weighted sample size for this survey was 3,416 respondents. O f these, 23 respondents reported injuries that included cuts and bruises (58%), tom cartilage (24%), and unknown types of injuries. Identified mechanisms of injury included falls (56%), motor-vehicle collisions due to the earthquake (27%), cuts from glass (4%), stepping on objects (4%), and unknown mechanisms (9%). Interestingly, the results to this survey reveal that less than 10% of the respondents reported being injured by non-structural objects falling on them, contrary to the Whittier Narrows survey results of 51% of the respondents being injured by falling non-structural objects (Shoaf et al., 1999). The authors do not address this disparity, but it could be due to differences in the time of day, day of the week, and characteristics of the environments (residential versus commercial) that were most affected. The Loma Prieta Earthquake caused major road damage that resulted in essentially isolating the County of Santa Cruz. A case-control study was conducted by 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Wagner and colleagues to study physical injuries to residents of Santa Cruz County caused by the main shock of the Loma Prieta Earthquake (Wagner, 1996; Wagner, Jones, & Smith, 1994). Cases either died, presented at one of the three hospitals in Santa Cruz county, or were transported by helicopter to a hospital outside Santa Cruz County. A detailed algorithm was developed in order to conduct interviews of injured patients (or proxies for those that had died) and abstract information from medical records. Wagner found that medical record documentation was inconsistent and incomplete, and self-reported assessments were not adequately detailed to allow assessment of injury severity (Wagner, 1999). Therefore, both data sources were used to obtain optimal information. Controls were selected through random digit dialing of residential phone numbers in Santa Cruz County. Non-injured controls were frequency- matched to cases on general residential area. Cases were also compared to injured controls that did not seek treatment in order to examine selection factors for seeking medical care. Unconditional logistic regression was performed using the outcome variable of injured versus non-injured. Important findings from this study include behavioral characteristics as well as physical characteristics of the environment that were associated with increased or decreased risk for injury. Rescuing people or exiting a structure were both associated with elevated risk for injury, while holding onto something was protective. In contrast, the combined act of rescuing and exiting a building was associated with reduced risk for injury. A possible explanation that Wagner provides for this finding suggests that individuals who were able to successfully accomplish both acts 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (rescuing and exiting) were possibly in much better physical condition than those that accomplished only one of the two acts (Wagner, 1999). Collapsing walls, ceilings, and floors were associated with statistically significant increases in risk for injury, as were chemical spills. General recommendations from this study include establishing emergency response plans that are based on individual building characteristics and associated estimates of seismic fragility and injury potential. 1.13. Earthquake-Related Injuries: Luzon. Phillipines. 1990. A case-control study of injuries from the 1990 earthquake in Luzon, Philippines is also documented in the literature (Roces, White, Dayrit, & Durkin, 1992). The Luzon earthquake occurred on July 16, 1990 at 4:30 p.m., and measured 7.7 on the Richter scale. The cases were identified from hospital records or Department of Social Welfare and Development lists. The study showed increased risk for injury in buildings made of reinforced concrete compared to wood-framed structures (Roces et al., 1992). This was an important finding, since previous studies had not identified this construction type as a risk factor, and it was thought that reinforcement would help to stabilize structures during seismic events. Previous studies had identified buildings made of unreinforced concrete and adobe as risk factors compared to wood-framed structures (Armenian et al., 1992; Glass et al., 1977; Wagner, 1996). Additionally, the study found that patients were at an increased risk for injury if they were on top or middle floors o f buildings rather than ground floors (Roces et al., 1992). This may be explained in part by a person’s age and their proximity to building exits. The very young and the very old are generally less 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ambulatory than the rest of the population, therefore they may be less likely to escape. Additionally, those on the ground level would have been able to leave the building more quickly than those on the upper stories. The authors also state that ‘escape’ from the building was protective from injury (Roces et al., 1992). This statement is substantiated by a reported increased estimate of risk for injury for those who stayed inside buildings compared to those who left. However, the questionnaire apparently did not assess whether the person attempted to escape versus succeeded in escaping. It is possible that successful escape from a building may be protective, however attempting to escape may be a risk factor. A mass exodus from a multi-story building may lead to increased risk for injuries due to crowd-related mechanisms (i.e., falls in stairwells, falls from pushing and trampling). In fact, the authors acknowledged that students and teachers in schools panicked, and subsequent stampeding resulted in ‘considerable numbers’ of injuries. Therefore risk may seem to be elevated for those who did not escape, but the risk may be due to the act of trying to leave rather than not leaving the building. Additionally, the injury itself could have occurred prior to the patient’s attempt or successful escape from the building. If the injury prevented the patient from escaping, risk may have seemed higher for those that did not escape. This illustrates why extraneous factors should be considered in the analysis, and how wording of survey questions can impact study findings. Other limitations include methodological problems common to disaster epidemiology. Since the controls were not matched to the cases, and no statistical adjustment was applied to control for this in the analysis stage, possible confounding bias 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. exists due to factors such as age and gender. Additionally, how the researchers determined whether the case was attributable to the earthquake is not clearly stated. Since earthquake-relatedness is rarely available in most documentation, fundamental issues regarding ascertainment and selection of cases may have also biased the study results. 1.14. Earthquake-Related Injuries: Northridge. California. 1994. Several different studies have been conducted on injuries resulting from the 1994 Northridge Earthquake in Los Angeles County. Included in these are a study of fatal and hospitalized injuries (Peek-Asa et al., 1998), a population-based survey (Shoaf et al., 1999), and a report of activity at a Disaster Application Center (DAC) (Teeter, 1996). These will each be discussed in the following paragraphs. Fatal and Hospitalized Injuries. Peek-Asa and colleagues investigated fatal and hospitalized injuries due to the 1994 Northridge earthquake (Peek-Asa et al., 1998). Earthquake-related deaths were identified through the Los Angeles Department of the Coroner. Earthquake-related hospitalized injuries were identified through a screening process of all 78 receiving hospitals (hospitals with emergency departments) in Los Angeles County. Sixteen hospitals subsequently showed at least one admission for an earthquake-related injury during the 2-week period after the earthquake. Individual medical records and coroner’s reports were reviewed in order to obtain information regarding patient demographic characteristics, circumstances o f the injury, and to identify causes and types o f injuries. 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The methods are clearly stated in the manuscript and include a detailed description of how cases were identified. Injuries were coded using the Abbreviated Injury Severity Scale (AIS), which was used to calculate the Injury Severity Score (ISS). Buildings in which injuries occurred were linked by address to Los Angeles City Department of Building and Safety reports that denoted whether a building had been inspected and the degree of damage the building sustained. Injury rates were calculated based on the 1990 Census for Los Angeles County. A total of 171 injuries (33 fatal and 138 hospitalized) were identified in this study (1.93 per 100,000 residents). An important finding confirmed that existing knowledge* regarding ratios of hospitalized to fatal injuries (4:1) was also appropriate for this event. Unadjusted injury rates were highest for those of Caucasian ethnicity. Unadjusted rates of earthquake-related injuries were also higher for females than for males, and increased dramatically with age. The trend was more pronounced for hospitalized than fatally injured patients. This is an important finding since people are living longer, and those aged 65 and older continue to comprise an ever-growing fraction of the national population. Future prevention and response strategies should include specific applications for retired individuals living at home, in retirement communities and in nursing homes in order to ensure that residents take proper precautions and respond in manners that will minimize injuries. Consistent with other research findings, the most common cause of fatal injuries was structural failure as reflected in building collapse or partial collapse (Armenian et al., “ This existing knowledge will be explained in more detail in Section 3.02. ATC-13 Methodology. 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1992; Roces et al., 1992). Those mechanisms were 8.4 times more likely to cause a fatal injury than a hospitalized injury. Only 8 of the 138 hospitalized individuals (5.8%) were hospitalized for structurally-related injuries (hit by beams, plaster, or chimneys). Other types of injuries associated with subsequent hospitalization included being hit by or caught between objects (furniture, glass, machines, crushing from trampling, and unknown objects), motor-vehicle related injuries, falls, burns, and electrocutions. The average ISS for fatally injured patients was 63.6 (range 4-76), whereas the average ISS for patients hospitalized for their injuries was 6.6. Fifty-five percent of those patients hospitalized for earthquake-related injuries had ISS scores less than 9, and 87% had ISS scores less than 16, therefore most of those hospitalized had moderate injuries. Only three hospitalized patients had ISS scores greater than 25. The body locations most commonly associated with fatal injuries were the head and thorax, but these estimates may be incorrect since complete autopsies were not conducted on all fatalities. Hospitalized patients were more commonly diagnosed with lower extremity injuries (54%) followed by upper extremity injuries (19%). Overall, 57 injuries (24 fatal and 33 hospitalized) were linked to buildings that had been inspected. Seventy-five percent of the fatally injured patients were linked to collapsed buildings. Seventeen of the 24 fatal injuries and 15 o f the 33 hospitalized injuries occurred in apartment complexes. Interestingly, one fatal injury occurred in a complex that did not reflect any damage from the City Building Inspection Report, although the medical record indicated that the injury was due to building collapse. The remaining 16 fatalities were linked to the Northridge Meadows complex that was 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. severely damaged and later demolished. Fifteen patients were hospitalized for structurally-related injuries that occurred in apartment buildings. However, only 1 of those 15 apartments (that were inspected) reflected structural damage. Six fatal and 11 hospitalized injuries occurred in single family residences o f which only 2 and 3 (respectively) were associated with inspection reports that reflected structured damage. These results bring into question the validity o f either the building inspection reports or the address information obtained through the medical documents. Five of the 30 buildings that were inspected and reported as undamaged were associated with coroner’s reports that indicated structural damage was responsible for the injury. These types of injuries included asphyxiation and body compression due to building collapse, injury due to a falling doorway, and injury due to falling structural debris. If the coroner’s reports reflected incorrect addresses for the injury scene, this might explain the disparity between the mechanism o f injury and the damage inspection report. Similarly, medical record documentation may not have reflected the correct incident address. Billing information may be the only available address information for hospitalized patients, and if this address is not current, or if it reflects an office, a rented mailbox (i.e., Mailboxes, etc.), a friend or relative, it would provide an incorrect link to the building inspection reports. Additionally, simple transcription or typographical errors are not uncommon to medical records. If Emergency Medical Services (EMS) Run-Sheets were available for patients transported by ambulance or helicopter, these would be more likely to reflect a correct incident address. However, the medical records house carbon copies of these documents, and frequently the medics must record information hastily and 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. handwriting is quite difficult to decipher. Therefore address errors based on these documents could be due to unintentional errors recorded by hasty medics, or handwriting misinterpretation. It is also possible that building inspection reports were inaccurate. If these reports were of questionable validity, concern is warranted about the usefulness of these documents for other purposes (hazard assessment, insurance reimbursement, monetary allocations, etc.). In general, the most important findings from this study include the identification of structural causes as the leading risk factor for fatal injuries, and falls and being hit by objects as the major causes of serious non-fatal injuries. The authors also speculate that a possible explanation for the unexpected finding that injuries o f lower severity resulted in hospitalization could be due to delayed care-seeking behavior, since this was apparent from review of the medical records. The authors also point out that had the event occurred 2 hours later in the morning, commuter traffic would have been building, and it is likely that many more severe and fatal injuries would have occurred. Population-Based Survey. Shoaf and colleagues also conducted a three-stage population-based random-digit dial Computer-Assisted Telephone Interview (CATI) survey between 6-24 months after the Northridge earthquake (Shoaf et al., 1999). A total of 149 of 1,830 respondents reportedly sustained an injury due to the earthquake. Ten percent of the injured patients sought care, and 5 individuals sought care at hospitals. Most of the injuries were minor (83% percent were cuts, bruises and sprains). Non- structural objects falling on the respondent caused most of the reported injuries (55%). Other mechanisms o f injury included the ‘physical force o f the earthquake’ (22%), and 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. behavioral responses (15%). Less than 1% of the injuries were caused by structural elements. The authors report on crude (unadjusted) predictors o f reported injuries and also present parameter estimates from a multivariate logistic regression. Crude predictors o f injury included socio-demographic factors (not married, apartment dwellers, non- Hispanics, and perception o f self as a victim), physical activity (whether the respondent moved or attempted to move), estimates of seismic activity and intensity, and building damage as reported in inspection results from the Los Angeles City Department of Building and Safety. In addition, a logistic regression model confirmed some of the crude associations while adjusting for other factors. Odds ratios are presented that show protective associations between increasing age (by ten-year age groups), marital status (married or living together with a partner) and ethnicity of Latino origin. Odds ratios greater than or equal to two included all categories of damage (whether the home was inspected or not), perceiving one’s self as a victim, and moving or attempting to move. The authors suggested that the distribution of characteristics observed in this sample may be applied to the population of Los Angeles to better understand how the earthquake affected residents. They estimated that 240,000 households had at least one adult injured during the earthquake. Ten percent (n=24,000) of the individuals sought care for their injuries, with approximately 8,000 presenting at health care facilities. The authors also presented a pooled comparison of the results to the three surveys conducted after the Whittier Narrows Earthquake, the Loma Prieta Earthquake, and the Northridge Earthquake. Important findings included behavioral factors such as moving 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. versus staying in place with respect to increased risk. Reporting o f injuries across the three samples was stable, so moving or staying in place during these earthquakes was consistently protective or risky. Additionally, this factor appears to be associated with or modified by demographic characteristics including gender, age, and ethnicity. This is consistent with a previous analysis of the same survey that was used for the Loma Prieta earthquake (Goltz et al., 1992). The authors also emphasize the importance of the extrapolated rates of injury to the county o f Los Angeles and how this demonstrates that many more individuals may have been injured and sought care than previously thought. Although this may be true, no limitations are discussed that might effect the study findings and subsequent generalizability (i.e., characteristics of non-respondents, lag-time between the event and administration of the survey, validation of reported injuries). An important recommendation that the authors submit based on their findings is that non-structural elements of the environment should be emphasized as a primary risk factor for injury in earthquake preparedness protocols. Traditional messages have not emphasized this, instead including it as a secondary message. Additionally, the authors point out that the traditional messages regarding movement (‘Duck, Cover, and Hold’) may not be providing optimal advice, confirming Wagner’s recommendations based on results to the case-control study of injuries subsequent to the Loma Prieta earthquake (Wagner, 1996). The time of day that the earthquake occurred is also introduced by Shoaf and colleagues as an important element to consider when making recommendations for safe responses. Lighting (daylight versus night) can effect an individual’s ability to respond appropriately therefore reducing risk for injury, and 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. general alertness (sleepiness versus being awake) probably has an impact on an individual’s ability to respond. Activity at a Disaster Application Center (DAC). A report of activity at the Northridge Disaster Application Center (DAC) indicated that 17,883 individuals sought care between January 26 and February 26, 1994 (Teeter, 1996). Information collected at the DAC included the patient age, gender and chief complaint, as well as the date of visit, and the outcome (treated versus referred). Thirty-two percent o f the 17,883 patients sought care for upper respiratory infections, 10% for headache, 9% for musculoskeletal pain, 9% for stress, 8% for blood pressure monitoring, 7% for gastrointestinal disease, 6% for medication refills, 5% for otitis media, and only 3% for lacerations and abrasions. The remainder of the patients sought care for dental problems, conjunctivitis, asthma, dermatitis, amenorrhea, dysmenorrhea, varicella, ectoparasitic infection, enterobiasis, insect bites or stings, dry skin, prenatal care, superficial fungal or bacterial infections, well-baby examinations or immunizations, to obtain baby supplies, or to obtain information regarding coccidiomycosis. The complaints of musculoskeletal pain were reportedly due to injuries sustained as a result from lifting, twisting, falling, or tripping. Ninety-eight percent of these patients were over age 20. Although not identified as such, most o f these injuries (as well as the lacerations and abrasions) would probably have been categorized as follow-up injuries or injuries indirectly related to the earthquake, since the patients presented for treatment more than 10 days after the main shock. 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.15. Generalization of Previous Iniurv Studies to Other Earthquakes. Generalization of findings from previous studies of earthquake-related injuries to other earthquakes is problematic because of variation in methodological approaches including identification and diagnostic criteria for cases and the choice of comparison groups or study controls. Other problems that complicate generalizability include geological conditions of the community, urbanization and access to medical care, construction standards, time o f day that the event occurs, variation in the age distributions of effected populations, and demographic characteristics of the community including cultural behaviors that vary between populations. Employing rates of injury and death from studies conducted in other countries may not be appropriate for Southern California, California, or the United States, since construction practices vary with cultural influences and economic conditions, and geologic conditions are highly variable and are not well- measured in most parts of the world. Additionally, some existing studies present findings as either statistically significant or non-significant. This presents a fundamental problem for epidemiologists. A school of thought exists criticizing the practicality of tests for statistical significance. Many standard epidemiologic texts address this issue, but Rothman and Greenland present a well-written chapter on this important subject that the reader may refer to for more detailed explanation (Rothman & Greenland, 1998). Significance tests and p- values are commonly misinterpreted as representing probabilities associated with test hypotheses. Rothman and Greenland state that: 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Under no circumstances, however, should one fall into the trap of thinking that a P-value for a simple test hypothesis (for example, that exposure and disease are unassociated) is a probability of that hypothesis: The P-value is usually much smaller than such a probability and so can easily mislead one into inappropriately rejecting the test hypothesis, (page 185) In general, epidemiologists prefer to conceptualize associations in terms of a range of values referred to as confidence intervals that lend some meaning to the reasonableness of point estimates based on characteristics of the data and model assumptions. The process of calculating confidence intervals is referred to as interval estimation. The width of a confidence interval is determined in part by the variability of the data, corresponding assumptions of the model, assumptions related to the data itself, and some predefined significance (alpha) level. Rothman and Greenland state that ‘I f the underlying statistical model is correct and there is no bias, a confidence interval derived from a valid test will, over unlimited repetitions of the study, contain the true parameter with a frequency no less than its confidence level’ (page 189). Wide confidence intervals can reflect more variability in the data and are more likely to include the null value of zero effect (equal to one). However, the strength of a relationship is often more apparent by providing the confidence interval surrounding the point estimate. For example, a 95% confidence interval o f 0.7-13.0 implies that the estimated effect is insignificant at an alpha level of 0.05 because the interval includes the null value of 1. However, it is clear from looking at this interval that there is a positive association between the variables being analyzed in these data. When studies fail to present confidence intervals around 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. point estimates, concern is raised regarding missed effects. Furthermore, it is difficult to compare or confirm quantitative findings from disparate data sets without some knowledge regarding the distribution of variables within individual data sets. 1.16. Sources of Injury Data. In order to conduct a population-based investigation of earthquake-related injuries, the epidemiologist must select the appropriate data source. Sources of injury data include hospital discharge data, trauma registries, emergency department logs, medical records, coroner’s reports, law enforcement reports, vital records, insurance claims, and self-reported injuries. Some of the strengths and limitations associated with some of these sources are discussed below. Hospital discharge data. Hospital discharge data are advantageous because they are usually collected on a state-wide basis and reflect the subset of the population that is admitted to hospitals for care. Usually these data contain some demographic information as well as information regarding the diagnosis of the patient at the time of discharge, services that were provided during the patient’s stay, and the length of stay. These data also exhibit five major limitations. First, they represent a biased selection of patients since only those patients warranting hospital admission are included. The fraction of injured patients admitted to hospitals is estimated to be between four and thirteen percent of the injured population that seeks medical care (Ribbeck et al., 1992; Williams et al., 1995). As previously mentioned when reviewing the injury pyramid, this is a relatively small segment of the injured population. 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Second, hospital admission rates for similar adverse health conditions may vary by facility. Admission to hospitals is dependent on availability of the necessary treatment and room to treat the patient. Therefore if the hospital does not offer a type of care (i.e., neonatal intensive care), then a patient would probably be transferred to a facility that provides the necessary care. Likewise, if a hospital offers the service required, but the necessary hospital ward has no empty beds, a patient would probably be transferred to another facility. Another factor that influences hospital admission rates is case identification or selection. This is commonly known as Berkson’s Bias and is a type of selection bias specific to hospital studies (Lilienfeld & Lilienfeld, 1980). The methodological problem is based on the knowledge that adverse health processes are frequently not uniform nor are they mutually exclusive of other biological processes. This lack of biological autonomy creates the opportunity for a spurious association between the adverse health process and other health processes or biological traits. Therefore individuals may present at facilities with not only the disease or condition of interest, but also another disease, condition, or biological trait. Although there may be no real association between the disease or condition of interest and, for example, a biological trait (like blue eyes in the general population), investigation o f patients in a hospital setting may, in fact, reveal such an association. This association is due to the different rates of admission to the hospital for people with the disease or condition of interest and (in this case) blue eyes. Conversely, this same type of selection bias may conceal an association in a study although it actually exists in the population. It is likely that 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. hospital admission rates are confounded by extraneous variables, however identification and control of all of these variables is usually not possible. The third limitation of hospital discharge data is that they are collected primarily for billing purposes, and important injury-specific information (including the cause of the injury) may be missing, incomplete, or erroneous, limiting its epidemiological usefulness (Runyan et al., 1992). Fourth, there is usually a considerable time delay involved in compilation of these data. In the State of California, there is usually a minimal 6-month lag time before annual data sets are available. Fifth, hospital admission rates are a function not only o f the diseases and conditions present in the population, but of the economic environment and insurance status of the patient. Specifically, hospital care is dictated by health maintenance organization (HMO) policy for patients who are enrolled in HMOs. Prior to admitting a patient to a hospital, the hospital must receive approval from the HMO. If this approval is not received, not only may the HMO deny payment for services rendered, but the hospital may be held liable for any problems or complications associated with the patient’s care upon admission. Therefore if the patient is not an approved admission, transferal to an approved facility may be arranged, or the patient may go untreated and may not be documented unless the condition progresses to a more serious stale that is deemed worthy of hospitalization. Trauma Registries. Trauma registries also represent a biased sample of patients. Only those patients that fit the definition of a trauma are included in these registries. Local authorities typically dictate this definition. The trauma definition is designed to 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. identify patients with the most severe injuries as well as patients in situations that might result in severe injuries, but, at the time of triage, the injuries may be undetected. For example, in Los Angeles County, patients involved in motor vehicle collisions are considered trauma patients if the space surrounding the patient was invaded as a result of the collision, regardless o f whether an injury occurs. Therefore if the passenger or driver’s door is dented and intrudes into the physical space adjacent to that individual, a trauma form is completed in order to provide documentation in the event of delayed detection of injuries. Although trauma registries are intended only for trauma patients, additional limitations have been reported that stem from difficulties differentiating patients who use the trauma facility as their primary care center from patients who are delivered to that facility for special trauma care (Payne & Waller, 1989). Emergency Department Logs o f Activity. The quality and hence the usefulness of emergency department logs vary a great deal between facilities. Although guidelines have been established by the Joint Commission on the Accreditation of Healthcare Organizations1 5 (JCAHO) regarding data collection specifications for emergency department (ED) records, these requirements do not apply to ED logs. JCAHO guidelines also do not include recommendations regarding identification of the mechanism of injury that is fundamental to injury prevention. The omission of this information is a major difference between JCAHO requirements and injury surveillance recommendations established by the Centers for Disease Control and Prevention and the b Founded in 1951. evaluates and accredits health care organizations in the United States, including hospitals, health plans, and other health care organizations. 58 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. National Academy of Sciences0 (Runyan et al., 1992). Since the mechanism of injury is usually considered anecdotal by medical personnel, and is not required by JCAHO, it is frequently missing from emergency department logs (Alberts et al., 1991; Irving et al., 1994; Ribbeck et al., 1992; Runyan et al., 1992). It is more likely to be found in medical records, but. depending on the facility and the personnel completing medical records, it may also be missing. Insurance Reports. Insurance claims are sources of information for policy makers and risk assessors regarding frequencies (counts) of losses due to disasters and costs associated with reported losses. Reports provided by insurance companies are limited because they exclude patients that are indigent, under state or federal health insurance plans, patients who do not seek care, and patients that do not file insurance claims. Another limitation with data provided by the insurance industry is that details such as patient demographics, diagnoses, and geographic locations are omitted. Instead, information regarding basic cost estimates is presented as well as frequencies of claims that represent requests for compensation of death, injury, or property loss. Little is known about the validity of the claim with respect to the dollar amount paid for compensation or whether trends observed in the insured population are applicable to the non-insured population. Self-Reported Injury. Self-reports o f injuries tend to be influenced by response bias, a systematic error due to erroneous recall or reporting. A forensic investigation of c Established by an Act of Congress in 1863. a society' of scholars dedicated to the advancement of sciences and applications for general welfare. 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. patients involved in litigation found support for bias in self-reports of baseline (pre- injury) problems (Lees-Haley, Williams, & English, 1996). Patients involved in litigation tended to report superior pre-injury functionality to that of controls. The implication is that legal proceedings influence the likelihood for self-reporting response bias. Landen and Hendricks examined the effect o f recall on occupational injury estimates from the 1988 Occupational Health Supplement of the National Health Interview Survey (Landen & Hendricks, 1995). In this survey, a 12-month reference period for injury reporting was used in order to obtain an adequately large sample size for analysis. The authors fitted a linear model to the existing survey data based on the estimated incidence rate expected if all the respondents had been interviewed within four weeks of their injuries. The effect is an adjustment for recall, resulting in a 32 percent higher injury incidence rate than before adjustment. The implications of this study suggest that accurate recall of injuries diminishes over time. This has also been observed in farm-workers that experience severe trauma, and other occupationally-based injuries (Hayden, Gerberich, & Maldonado, 1995; Landen & Hendricks, 1995; Mittleman, Maldonado, Gerberich, Smith, & Sorock, 1997). 2.00 EARTHQUAKE HAZARD ASSESSMENT. A fundamental orientation to geology, seismology, and structural engineering is necessary in order to have a basic understanding of the terms and conditions that are incorporated in hazard estimation models. The goal of the following overview is to provide a brief introduction to the geology of earthquakes, seismology, and 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. characteristics of engineered structures in order to help provide meaning to variables used in earthquake hazard methodology. 2.01. Geology Background. Various theories have been developed regarding the causes of earthquakes. It is known that as continental plates move, earthquakes occur. However earthquakes may also occur in areas that are not at the edge of a continental plate (i.e., along the New Madrid Fault in the Midwest of the United States). Nearly all theories agree on some basic principles regarding the mechanism of earthquakes, and these are outlined below. The technical material presented in the following sections has been compiled from two geology texts: Modem Global Seismology and Structural Geology: Fundamentals and Modem Developments (Ghosh, 1993; Lay & Wallace, 1995). Earthquakes are mainly attributed to an accumulation of strain due to stress along the crust of the earth. This strain is thought to accumulate along faults that are previous displacements of rock (i.e., points o f fracture along which the rocks on either side have moved in discontinuous directions) and are therefore relatively weak. The earth’s crust can withstand only a certain amount of stress. Depending on this threshold and the material properties of the rock and the fault surface, the strain must be relieved. The sliding of the planes of rock along the fault releases the strain. This sliding can be abrupt or gradual, and there is some amount of friction associated with the sliding. The material properties of the rock and the amount of stress at the fault determine the degree of friction that is produced. 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The subsequent release of energy follows the laws of physics. Much o f this energy is consumed in the heating and fracturing of rock as it slides, but a proportion is converted into seismic waves that propagate outwards from the fault source (impulsive waves). The nature of these waves can be predicted and represented mathematically, but they share features common to all waves. Although the waves do not transfer matter they do transfer energy and transmit information in the form of frequencies and amplitudes. The energy transfer follows the laws of simple harmonic motion, that is a shifting back and forth of energy from kinetic to some form of potential energy. This may result in a collapse of a building, and the collapse generally follows Newtonian laws o f motion, (e.g., Newton’s 2n d Law of Motion^: ZF = ma). The information that is transmitted by the waves is audibly perceived at the surface if the planes of rock slide abruptly. 2.02. Fault Terminology. Fault orientation varies, and the orientation effects the propagated waves. Fault terminology is rather complex, but basic nomenclature exists to describe the geographic orientation of the fault plane. A fault may be planar or curved and is a point o f fracture where slippage between two blocks of material has occurred. Two angular parameters are used to describe the orientation o f the fault surface. These are the strike and dip of the fault. In general, the strike is the direction taken by the fault plane as it intersects the earth’s (horizontal) surface.(Parker, 1994) The strike angle is formed between the projection of the fault onto the earth’s surface and the North. The dip of the fault is the d The vector sum o f a group of forces {SF) acting on an object of m ass'm ’ is equivalent to the object’s mass multiplied by the resultant acceleration 'a '. 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. angle formed between the earth’s surface and the fault plane. A graphical depiction of these angles is presented in Figure 2. Fiettre 2. Graphical Presentation o f Angular Elements o f a Fault.' dip direction dip Definition of strike, dip and dip direction o f a planar structure. ' Adapted from Ghosh. S. (1993). Structural Geology: Fundamentals and Modern De\elopments, Pcreamon Press. 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A slip vector defines the displacement or slip between the two blocks of material on either side of the fault plane. This vector can have any orientation on the fault plane. If a fault is not vertical (and they rarely are) the block of material above the fault is conventionally called the hanging wall, whereas the block of material below the fault is referred to as the foot wall. A graphical depiction of the hanging wall and foot wall is presented in Figure 3 The direction of the slip vector is measured in the fault plane and is given by the angle of slip (or rake) from the strike to the slip vector. The slip vector shows the motion of the hanging wall relative to the foot wall. Fi Pitre 3. Graphical Presentation o f Structural Elements o f a Fault/ a Fault line, hanging wall and foot wall. f Adapted from Ghosh. S. (1993). Structural Geology: Fundamentals and Modern Developments, Pcrgamon Press. 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Three basic categories of faults exist based on the slip of the material on either side of the fault. Strike-slip faults are characterized by an approximately horizontal displacement of the material on both sides of the fault. The material on either side of these types of faults moves either right or left. The San Andreas fault is an example of a strike-slip fault. Dip-slip faults are characterized by movement of the hanging wall either down (toward) or up (away) from the foot wall. A normal dip-slip fault is the term used to describe the former event, and a reverse dip-slip fault describes the latter. A reverse fault that is characterized by a dip of less than 45° is referred to as a thrust fault. The Northridge Earthquake was the result of strain along a thrust fault (Scientists of the United States Geological Survey and the Southern California Earthquake Center, 1994; Holmes & Somers, 1996). Oblique-slip faults are characterized by both vertical and horizontal displacement of the material on either side of the fault. A graphical presentation of an oblique-slip fault is presented in Figure 4. Figure 4. Graphical Presentation o f Components o f Displacement in an Oblique Fault* Net slip, strike slip and dip slip in an oblique-slip fault. * Adapted fronrGhosh. S. (1993). Structural Geology: Fundamentals and Modern De\’ elopments. Pcrgamon Press. — 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.03. Seismology Background. Seismology is the discipline that investigates the sources, generation, propagation, and recording of seismic or elastic waves. These waves characterize disturbances that expand outward from a source of stress in rock. Two fundamental types of waves are used in seismology and are referred to as body waves and surface waves. Body waves are transmitted from the underground source to the surface. Surface waves are produced at the earth’s surface through an interaction of body waves. A schematic representation of both of these types of waves is presented in Figure 5. Two types of body waves propagate through the interior of rock: P-waves (primary waves) and S-waves (secondary waves). P-waves are analogous to sound waves and can also travel through fluids. S-waves have only shearing deformation with no volume change and do not travel through fluids. Since both of these types of waves travel through the interior of a medium, they are called body waves, and Poisson first identified them in 1830. Surface waves are produced as a result of interaction between P and S waves as the waves approach the earth’s surface. As the body waves approach the surface o f the rock, the material conditions change, i.e., there is less surface stress on the rock and the earth may be less dense. These types o f changes create changes in an interactive manner that result in the generation o f new surface waves. Rayleigh waves, named after Lord Rayleigh who demonstrated their existence in 1887, are motions that occur and propagate 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 5. Graphical Presentation o f Particle Motion o f Body Wa\’ es (P-waves). and Surface IVcnvs {S-waves, I.nve waves, and Rayleigh waves). P w ave -C om pressions r U ndisturbed m ed iu m m 9B3I BBI1 B I S I BBBI tsssi ■ •B in iR B B B B N n iB S X B B rB B ae a sD s e iaa s B Q B E B IB R B B S 3C C Iiam B B E S SiX Sk7C S3E B 3B 9B B S EEsaterxEB BSB ieesiaiaK W s s a iB E a s c H B s a H ia e D ilatations S w ave 7^-rz I T “ V T . | T b ~r K !5 5 HE!S52^ P 3 Love w ave D o u b le A m p litu d e -•— W avelength — £ ■ BCnHBa2HS?»BKaW.T!r?53T'jt ! ,rE J5 ' i C L R ayleigh w ave d 3* Adapted from Lay & Wallace (1995). Modem Global Seismology, Academic Press. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. only along the surface o f the body. Love waves occur along the surface of a body that is characterized by layered material properties (most bodies have these layered properties). Both body and surface waves are influenced by various factors including the depth of the seismic event, material property changes and internal boundaries that enable wave reflection. P-waves are compressional waves that travel faster than other body and surface waves. These waves move the surface of the ground (generally) in a vertical motion. These waves also compress air and can produce loud roars like the sound of a train passing close by. The P-wave is usually not the most devastating wave of the earthquake. S-waves are distortional, moving primarily from side to side in a shearing motion. Surface waves have both vertical and horizontal components. Since most buildings are not able to withstand as much horizontal movement as vertical movement, the S-wave, Rayleigh, and Love waves are usually associated with more destruction. Body and surface waves are detected and measured by seismometers and the resulting output is called a seismograph. Crude types o f these instruments have been discovered and dated as early as 132 AD. Seismometers employ the principle of a pendulum, or freely suspended weight, in their design. The concept o f recording the time of the ground shaking does not appear to have been accomplished until 1784 by an Italian physicist named A. Cavalli. Another Italian, Filippo Cecchi, developed the first true seismograph in 1875. This instrument, although still crude, recorded the magnitude of ground shaking as a function of time. The three basic elements to a seismometer are: a supporting frame, a pendulum (inert weight) that is suspended from the frame with a pen at the tip, and a rotating drum with paper on which the earth’s movements are recorded. 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ft&tre 6. Essential Components o f a Seismograph.' Supporting frame Damping magnet crum Pen Rotating HORIZONTAL EARTH M OTION Rotating drum Supporting frame Spnng Pe„ Damping ' magnet m weignt T h e essentials o f th e seism o g rap h . The instrument at the top measures horizontal Earth motion, that at the bottom vertical Earth motion. In both cases, the supporting frame and rotating drum move during an earthquake while the inert weight does not: a pen attached to the weight therefore traces a line on the paper of the drum corresponding to the Earth’ s seismic movements. ‘ Adapted from Robinson. A. (1993). Earthshock, Thames & Hudson.(Robinson. 1993) 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Seismometers can measure both horizontal and vertical movement. In both cases (see Figure 6), the frame and the drum will move with the ground during an earthquake. However the inert weight of the pendulum remains constant. The line that is drawn on the rotating drum is therefore a recording of the earth’s seismic movement. There is a certain amount o f seismic background noise that must be considered for each site when interpreting seismographs. Continuous processes such as tides, atmospheric pressure changes, diurnal heating of the earth’s surface, and man-made vibrations and explosions effect seismic waves. Furthermore, seismic events have very large frequency and amplitude ranges resulting in a spectrum of motions and frequencies. These issues have posed special problems for seismologists. 2.04. Comparative Analyses of Seismographs. Prior to 1960, the characteristics of seismometers were not standardized and this (coupled with the previously mentioned complications) made comparative analyses difficult. After the first underground nuclear explosions occurred toward the end of the 1950s. the United States government funded research to investigate and explain seismic wave propagation in order to monitor underground nuclear testing. This initiative lead to the deployment of the World Wide Standard Seismograph Network. The network was comprised of 120 high quality, well-calibrated seismology stations. The instruments were calibrated to record specific frequency ranges as a standard function of time, and the information was pooled for analysis. This network allowed researchers to confirm previously hypothesized theories and develop new theories regarding seismic events. 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Other countries have also established well-calibrated seismic stations over the past 30-40 years, and the concept of pooled analysis o f seismic data has grown internationally. These pooled analyses have yielded much seismic information that has been used for advanced research including identification of earthquake sources, magnitudes, fault characteristics, forecasts of future earthquakes, and refinement in instrumentation. The field of seismology advanced beyond the use of crude pendulum instruments over time. During the 1970s, electromagnetic instruments were developed with an optical recording feature that recorded ground motion on photographic paper. Subsequently, force-feedback instruments have been employed that digitally record voltage from the output of seismometers. However, very strong ground shaking saturates the responses of standard seismometers, therefore accelerometers have been developed to measure and record high intensity or peak ground accelerations. Regional networks of accelerometers have been established and data has been pooled for analysis in order to arrive at estimates of high intensity ground movement at various geographic locations. These networks of accelerometers have generated useful information that has helped establish and refine construction codes for buildings in areas at risk for strong seismic activity. 2.0S Liquefaction. Ground failure has been identified with certain soil conditions in past earthquakes. These soil conditions include sensitive clays, loose unsaturated sands, and loose saturated sands. Earthquake effects including liquefaction are common in these soil 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. conditions. Factors that affect liquefaction include soil texture and density, the depth of the water table in the geographic area of interest, and ground-shaking intensity. Fine- textured soil such as fine sand, silty sand, and sandy silt liquefy more easily than coarser grained materials. Similarly, soils o f low density are more susceptible to liquefaction. Water must also be present for liquefaction to occur, hence the depth of the water table effects the likelihood of liquefaction. Since shaking intensity determines the level of shear stress on surface soil, more intense ground shaking is also associated with increased likelihood for liquefaction (Applied Technology Council, 1985). The mechanism of liquefaction is best described as it applies to loose saturated sands. Shearing deformations associated with seismic waves cause the sand particles to rearrange and become separated from one another. In the extreme case, these particles actually go into suspension. The liquefied soil-water mass has no shear (surface) strength, hence rapid catastrophic ground and soil failures occur (Applied Technology Council, 1985). Six distinct types of ground liquefaction have been described, with the most commonly observed being the lateral spread. The lateral spread is characterized by (primarily) lateral movement of surface soil over a liquefied layer of soil. Lateral spreads usually develop on gentle slopes and involve displacements that may reach several feet in length. In extreme conditions, the displacement may reach several tens of feet in length and may be accompanied by ground cracks and vertical displacements (Applied Technology Council, 1985). 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.06. Earthquake Magnitude Scales. The diversity of instruments used to record different types of waves has resulted in the development of different scales to measure the magnitude of an earthquake. The most well-known magnitude scale was developed in the early 1930s by Richter and is referred to as the Local Magnitude (M l ) scale. One of the limitations of this scale derives from the instrument that was used to develop it (Wood-Anderson torsion seismometer). That instrument is calibrated to a relatively narrow frequency range, and is no longer in general use. Additionally, this scale was developed to measure earthquakes principally in California, which are typically shallow (less than 15 km deep). The Local Magnitude scale tends to saturate at about magnitude 6.5, and therefore underestimates the magnitude of earthquakes that are greater than magnitude 6.0 (Scientists o f the United States Geological Survey and the Southern California Earthquake Center, 1994). The Local Magnitude scale is used in engineering, however, because many structures have natural periods of horizontal displacement that are similar to that of a Wood-Anderson instrument (0.8 seconds). Other scales exist including the Body Wave Magnitude (mb) and Surface Wave Magnitude (Ms). The Body Wave Magnitude measures the amplitude of the P-wave and the Surface Wave Magnitude measures the amplitude of the effect of the combined surface waves. Both of these scales are compatible with Ml. however all three scales are frequency-dependent and tend to saturate at levels that would correspond to medium sized seismic events. Therefore a fourth type of scale has been developed to represent the total size of an earthquake. This scale is mathematically interesting and considers the 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sum of the potential and kinetic energies integrated over time, and the subsequent energy that is expended in the form of work. Kanomori derived the resulting Moment Magnitude Scale (Mw) in 1977 and, although it is linked to Ms, it will not saturate because it is based on the seismic moment, which does not saturate. 2.07. Earthquake Intensity Scales. Another manner of comparing earthquakes has been based on perceived intensity. Intensity and magnitude are sometimes confused, but they are quite different. Intensity is highly subjective, whereas magnitude (although estimated) is based on quantitative measurements. Several scales of intensity have been developed and these scales are presented in Tables 2-6 and discussed in the following paragraphs. The information presented is adapted from the Applied Technology Council’s ATC-13 Earthquake Damage Evaluation Data for California (Applied Technology Council, 1985). The intensity scale that is most well known in the Western World is the Modified Mercalli Intensity (MMI) Scale (Table 2). The MMI is a twelve-point scale that is based on perceived ground-shaking and observable damage. It is obvious that population density, construction practices, and observer and reporter biases will influence this scale. However, this scale is still commonly used to present regional and building-specific damage data, and it is also often the only link to reported damage levels for seismic events prior to 1960. A graphical comparison of the values from the MMI Scale relative to the other four intensity scales is presented in Figure 7. 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2. Modified Kfercalli Intensity Scale (MNP). Adapted from Seiberp 's (1923) Mercalli-Cancani Scale. Modified and Condensed. Quoted from Wood and Neumann (195l)j MMI Description / Defining Characteristics I. Not felt, except rarely under especially favorable circumstances. Under certain conditions, at and outside the boundary of the area in which a great shock is felt: sometimes birds, animals, reported uneasy or disturbed; sometimes dizziness or nausea experienced: sometimes trees, structures, liquids, bodies of water, may sway—doors may swing, very slowly. II. Felt indoors by few people, especially on upper floors, or by sensitive or nervous persons. Also, as in MMI I, but often more noticeably: sometimes hanging objects may swing, especially when delicately suspended; sometimes trees, structures, liquids, bodies of water, may sway, doors may swing, very slowly; sometimes birds, animals, reported uneasy or disturbed; sometimes dizziness or nausea experienced. III. Felt indoors by several people, motion is usually rapid vibration. Sometimes not recognized to be an earthquake at first. Duration estimated in some cases. Vibration like that due to passing of light, or lightly loaded trucks, or heavy trucks some distance away. Hanging objects may swing slightly. Movements may be appreciable on upper levels of tall structures. Standing motor cars may be rocked slightly. 1 Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City. CA 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2. (continued)J MMI Description / Defining Characteristics IV Felt indoors by many, outdoors by few. Awakened few, especially light sleepers. Frightened no one, unless apprehensive from previous experience. Vibration similar to that of passing or heavily loaded trucks. Sensation similar to that of a heavy body striking building, or falling of heavy objects inside. Rattling of dishes, windows, doors; glassware and crockery clink and clash. Creaking of walls and house frame, especially in the upper range of this grade. Hanging objects swing in numerous instances. Liquids in open vessels slightly disturbed. Standing motor cars are rocked noticeably. V. Felt indoors by practically all people, outdoors by many or most people: outdoors direction estimated. Awakened many or most people. Frightened few— slight excitement; few people run outdoors. Buildings tremble throughout. Broken dishes, glassware, to some extent. Cracked windows—in some cases, but not generally. Overturned vases, small or unstable objects in many instances occasionally fall. Hanging objects, doors, swing generally or considerably. Pictures are knocked against walls or displaced. Doors or shutters are opened or closed abruptly. Pendulum clocks stopped, started, or ran fast or slow. Moved small objects and furnishings, the latter to a lesser extent. Spilled liquids in small amounts from well-filled open containers. Trees and bushes slightly shaken. 1 Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City. CA. 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2. (continued)! MMI Description / Defining Characteristics VI. Felt by all people, indoors and outdoors. Frightened many, excitement in general, some alarm, many people run outdoors, awakened all. Persons made to move unsteadily. Trees and bushes shaken slightly to moderately. Liquid set in strong motion. Small bells rang—church, chapel, school, etc. Damage slight in poorly built buildings. Fall o f plaster in small amounts. Cracked plaster, especially fine cracks in chimneys in some instances. Broken dishes and glassware in considerable quantity, also some window breakage. Fall o f knick-knacks, books and pictures. Overturned furniture in many instances. Moderately heavy furniture is moved. VII. Frightened all people - general alarm, all ran outdoors. Some or many found it difficult to stand. Felt by persons driving motor vehicles. Trees and bushes shaken moderately to strongly. Waves on ponds, lakes, and running water. Water turbid from mud that is stirred up. Incaving to some extent o f sand or gravel stream banks. Rang large church bells. Suspended objects made to quiver. Damage negligible in buildings of good design and construction, slight to moderate in well-built ordinary buildings, considerable in poorly built or badly designed buildings, adobe houses, old walls (especially where built without 1 Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City. CA. 77 with permission of the copyright owner. Further reproduction prohibited without permission. Table 2. (continued).1 MMI Description / Defining Characteristics VTI. (continued) mortar), spires, etc. Cracked chimneys to considerable extent, walls to some extent. Fall of plaster in considerable to large amounts, also some fallen stucco. Numerous windows broken, furniture broken to some extent. Loosened brickwork and tile shaken loose. Weak chimneys broken at the roofline. Fall o f cornices from towers and high buildings. Dislodged bricks and stones. Overturned heavy furniture with damage from breaking. Damage considerable to concrete irrigation ditches. VIII. General fright—alarm approaches panic level. Disturbed persons driving motor cars. Trees shaken strongly—branches, trunks, broken off, especially palm trees. Ejected sand and mud in small amounts. Temporary or permanent changes: in flow of springs and wells; dry wells renewed flow; in temperature of spring and well waters. Slight damage in brick structures especially built to withstand earthquakes. Partial collapse considerable in ordinary substantial buildings: racked, tumbled down, wooden houses in some cases; panel walls thrown out of frame structures. Fall of walls. Cracked or broken stone walls. Wet ground. Twisting or falling of chimneys, columns, monuments, factory stacks, or towers. Furniture moved conspicuously or overturned. 1 Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City, CA 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2. (continued).3 MMI Description / Defining Characteristics IX. General panic. Ground conspicuously cracked. Damage considerable in masonry structures built to withstand earthquakes: some buildings thrown out of plumb. Damage great in masonry buildings: some collapse in large part; wholly shifted frame buildings off foundations; racked frames. Damage serious to reservoirs: underground pipes sometimes broken. X. Cracked ground, especially when loose and wet. Cracks up to several inches in width, fissures up to a yard in width running parallel to canal and stream banks. Landslides considerable from riverbanks and steep coasts. Sand and mud shifted horizontally on beaches and flat land. Changed level of water in wells. Water thrown onto banks of canals, lakes, rivers, etc. Damage serious to dams, dikes, embankments. Damage severe to well-built wooden structures and bridges, some destroyed. Cracks developed in excellent brick walls. Destroyed most masonry and frame structures along with their foundations. Bent railroad rails slightly. Buried pipelines crushed or tom apart. Open cracks and broad wavy folds in cement pavement and asphalt road surfaces. XI. Widespread disturbances in ground, varying with composition. Broad fissures, earth slumps, and land slips in soft, wet ground. Ejected water in large amounts, 1 Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City. CA. 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2. (continued)* MMI Description / Defining Characteristics XI. (Continued) charged with sand and mud. Caused sea-waves (tidal waves) of significant magnitude. Damage severe to wood-frame structures, especially near shock centers. Damage great to dams, dikes, embankments, often for long distances. Few (if any) masonry structures remain standing. Destroyed large well-built bridges by wrecking supporting piers or pillars. Affected yielding wooden bridges to a lesser extent. Bent railroad rails greatly, and thrust them endwise. Buried pipelines completely out of service. XII. Total damage. Practically all works of construction damaged greatly or destroyed. Disturbances in ground great and varied, numerous shearing cracks. Landslides, falling rocks, slumping of river banks, etc., numerous and extensive. Large rock masses wrenched loose or tom off. Fault slips in firm rock with notable horizontal and vertical offset displacements. Water channels, surface and underground, disturbed and modified greatly. Dammed lakes produced waterfalls, deflected rivers, etc. Waves seen on ground surfaces. Distorted lines of sight and level o f sight. Objects thrown upward into the air. ■ ' Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City. C A 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 7 . Graphical Comparison o f Seismological Intensity Scales,k Modi f ied R o s s i - F o r e l H e r c a l t i JMA Geofian HSK I I 0 I I I II 11 II 11 I I I I I I I II I I I IV IV II IV IV V V I I I V V VI VI IV VI VI VII V V I I I VII VII VII VIII VIII VIII IX IX VI IX IX X X X X XI VII XI XI XII XII XII k Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data fo r California, Redwood City. CA. 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Another scale that has been used is the Japan Meteorological Agency Scale (JMA) (Table 3). This is also a qualitative scale that has seven levels of perceived intensity. JMA Level I corresponds (roughly) to MMI Levels II and HI. JMA Levels II- IV correspond closely to MMI Levels IV-VI. JMA Level V corresponds to MMI Levels VII and VIII. JMA Levels VI correspond to MMI=IX and X, and JMA=VTI correspond to MMI=XI and XII. Table 3. Japan Meterolozical Agency (JMA) ScaleJ JMA Description / Defining Characteristics 0. Not felt: too weak to be felt by humans; registered only by seismographs. 1 . Slight: felt only feebly by persons at rest or by those who are especially observant of earthquakes. II. Weak: felt by most persons; slight shaking of windows and Japanese latticed sliding doors (Shoji). III. Moderately strong: shaking of houses and buildings, heavy rattling of windows and Japanese latticed sliding doors, swinging of hanging objects, stopping of some pendulum clocks, moving of liquids in vessels; some people are so frightened that they run out of doors. IV. Strong, strong shaking of houses and buildings, overturning of unstable objects, and spilling of liquids out of vessels. 1 Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City, CA 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3. (Continued).1 JMA Description / Defining Characteristics V. Very strong: cracking brick and plaster walls, overturning stone lanterns, gravestones and similar objects, damaging chimneys and mud-and-plaster warehouses, landslides in steep mountains. VI. Disastrous: causing destruction of 1-30 percent of Japanese wooden houses; causing large landslides; fissures in flat ground and some in low fields accompanied by mud and waterspouts. VII. Ruinous: causing destruction o f more than 30 percent of the houses; causing large landslides, fissures and faults. 1 Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data fo r California. Redwood City. C A Still another scale is the Medvedev-Sponheuer-Kamik Scale (MSK) (Table 4). This is another 12-point scale with levels that correspond roughly to the same levels of perceived damage designated in the Modified Mercalli Intensity scale. Additionally, the Medvedev-Sponheuer-Karnik Scale has codes that may be incorporated into the scale to denote types of structures (i.e., stone buildings, prefabricated buildings, reinforced buildings), quantity of damage (single/few=5%, many=50%, most=75%), gradation of damage to buildings (slight, moderate, heavy, and destruction). (Applied Technology Council, 1985.) 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4. Medvedev-Sponheuer-Kantik (MSK) Intensity Scale. Quoted from Barosh (1969) m MSK Description / Defining Characteristics I. Not noticeable: The intensity of the vibration is below the limit of sensibility; the tremor is detected and recorded by seismographs only. II. Scarcely noticeable (very slight): Vibration is felt only by individual people at rest in houses, especially on upper floors of buildings. III. Weak, partially observed only. The earthquake is felt indoors by a few people, outdoors only in favorable circumstances. The vibration is like that due to the passing of a light truck. Attentive observers notice a slight swinging o f hanging objects, somewhat more heavily on upper floors. IV Largely observed: The earthquake is felt indoors by many people, outdoors by few. Here and there people awaken, but no one is frightened. The vibration is like that due to the passing of a heavily loaded truck. Windows, doors and dishes rattle. Floors and walls creak. Furniture begins to shake. Hanging objects swing slightly Liquids in open vessels are slightly disturbed. In standing motor cars, the shock is noticeable. V. Awakening: The earthquake is felt indoors by all, outdoors by many. Many sleeping people awake. A few run outdoors. Animals become uneasy. Buildings tremble throughout. Hanging objects swing considerably. Pictures knock against m Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data fo r California. Redwood City. CA. 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4. ('Continued)."' MSK Description / Defining Characteristics V. (Continued) walls or swing out of place. Occasionally pendulum clocks stop. Few unstable objects may be overturned or shifted. Open doors and windows are thrust open and slam back again. Liquids spill in small amounts from well-filled open containers. The sensation of vibration is like that due to a heavy object falling inside the buildings. Slight damage (fine cracks in plaster; fall of small pieces of plaster) in buildings made of field-stone, rural structures, adobe houses and clay houses. Sometimes change in the flow of springs. VI. Frightening: Felt by most indoors and outdoors. Many people in buildings are frightened and run outdoors. A few persons lose their balance. Domestic animals run out of their stalls. In a few instances, dishes and glassware may break, books fall down. Heavy furniture may possibly move and small steeple bells may ring. Slight damage (fine cracks in plaster; fall o f small pieces of plaster) is sustained in single buildings of ordinary brick, large block, prefabricated buildings, half- timbered structures and buildings of natural hewn stone. Damage in buildings made of field-stone, rural structures, adobe houses and clay is moderate (small cracks in walls; fall of fairly large pieces of plaster; pantiles slip off; cracks in chimneys; parts of chimneys fall down). In a few cases, cracks up to 3 1 Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data fo r California. Redwood City. CA 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4. (Contitmed).m MSK Description / Defining Characteristics VI. (Continued) widths o f 1 centimeter are possible in wet ground; in mountains occasional landslips; changes in flow of springs and in level of well water are observed. VII. Damage in buildings: Most people are frightened and run outdoors. Many find it difficult to stand. The vibration is noticed by persons driving motor cars. Large bells ring. In many reinforced buildings and well-built wooden structures, slight damage (fine cracks in plaster; fall o f small pieces of plaster) is sustained. In many buildings of ordinary brick, large block, prefabricated buildings, half- timbered structures and buildings of natural hewn stone, damage is moderate (small cracks in walls; fall o f fairly large pieces of plaster; pantiles slip off; cracks in chimneys; parts of chimneys fall down). Many buildings made of field-stone, rural structures, adobe and clay sustain heavy damage (large and deep cracks in walls; fall of chimneys), with few sustaining destruction (gaps in walls; parts of buildings may collapse; separate parts of the building lose their cohesion; inner walls and filled-in walls of the frame collapse). In single instances landslips of roadway on steep sloes; cracks in roads; seams of pipelines damaged; cracks in stone walls. Waves are formed on water, and water is made turbid by mud stirred up. Water levels in wells change, and the flow of springs changes. In few cases m Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data fo r California. Redwood City. CA. 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4 (Continued).m MSK Description / Defining Characteristics VII. (Continued) dry springs have their flow restored and existing springs stop flowing. In isolated instances parts of sandy or gravelly banks slip off. VIII. Destruction of buildings: Fright and panic; also persons driving motor cars are disturbed. Here and there branches of trees break off. Even heavy furniture moves and partly overturns. Hanging lamps are in part damaged. Many reinforced buildings and well-built wooden structures sustain moderate damage (small cracks in walls; fall o f fairly large pieces of plaster; pantiles slip off; cracks in chimneys; parts o f chimneys fall down); few sustain heavy damage (large and deep cracks in walls; fall o f chimneys). Many buildings of ordinary brick, large block, prefabricated buildings, half-timbered structures and buildings of natural hewn stone sustain moderate damage (large and deep cracks in walls; fall of chimneys). Many buildings made of field-stone, rural structures, adobe and clay sustain destruction (gaps in walls; parts of buildings may collapse; separate parts of the building lose their cohesion; inner walls and filled-in walls of the frame collapse). Occasional breakage of pipe seams. Memorials and monuments move and twist. Tombstones overturn. Stone walls collapse. Small landslips in ” Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data fo r California. Redwood City. CA. 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4. (Continued).m MSK Description / Defining Characteristics VIII. (Continued) hollows and on banked roads on steep slopes; cracks in ground up to widths of several centimeters. Water in lakes becomes turbid. New reservoirs come into existence. Dry wells refill and existing wells become dry. In many cases change in flow and level o f water. IX. General damage to buildings: General panic; considerable damage to furniture. Animals run to and fro in confusion and cry. Many reinforced buildings and well- built wooden structures sustain heavy damage, few sustain destruction. Many buildings of ordinary brick, large block, prefabricated buildings, half-timbered structures and buildings of natural hewn stone sustain destruction; a few sustain total damage. Many buildings made of field-stone, rural structures, adobe and clay sustain total damage. Monuments and columns fall. Considerable damage to reservoirs; underground pipes partly broken. In individual cases railway lines are bent and roadways damaged. On flat land overflow of water, sand and mud is often observed. Ground cracks to widths of up to 10 centimeters, on slopes and river banks more than 10 centimeters, furthermore a large number of slight cracks in ground; falls of rock, many landslides and earthflows; large waves on water. Dry wells renew their flow and existing wells dry up. m Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City. CA. 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4. (Continued).m MSK Description / Defining Characteristics X. General destruction of buildings: Many reinforced buildings and well-built wooden structures suffer destruction, a few are totally damaged. Many buildings of ordinary brick, large block, prefabricated buildings, half-timbered structures and buildings of natural hewn stone are totally damaged; most buildings made of field-stone, rural structures, adobe and clay are totally damaged; critical damage to dams and dykes and severe damage to bridges. Railway lines are bent slightly. Underground pipes are broken or bent. Road paving and asphalt show waves. In ground, cracks up to widths of several decimeters, sometimes up to 1 meter. Parallel to water courses occur broad fissures. Loose ground slides from steep slopes. From river banks and steep coasts considerable landslides are possible. In coastal areas displacement of sand and mud; change of water level in wells; water from canals, lakes, rivers, etc., thrown on land. New lakes occur. XI. Catastrophe: Severe damage even to well-built buildings, bridges, water dams, and railway lines; highways become useless; underground pipes destroyed. Ground considerably distorted by broad cracks and fissures, as well as by movement in horizontal and vertical directions; numerous landslips and falls of rock. The intensity of the earthquake requires to be investigated specially. m Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data fo r California. Redwood City'. CA. 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4. (Continued).m MSK Description / Defining Characteristics XII. Landscape changes: Practically all structures above and below ground are greatly damaged or destroyed. The surface of the ground is radically changed. Considerable ground cracks with extensive vertical and horizontal movements are observed. Falls of rock and slumping of river banks over wide areas; lakes are dammed; waterfalls appear, and rivers are deflected. The intensity of the earthquake requires to be investigated specially. ^ Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data fo r California. Redwood City. CA. The Rossi-Forrel Scale (RF) (Table 5) is another type o f intensity scale. This scale is a ten-level scale (again) based on perceived intensity o f the seismic event, and includes seismograph impact at the lowest levels. The highest level of intensity (RF=X) is comparable to MMI=X or greater, and RF=III is also comparable to MMI=III. No other similar cut-points are noted between the two scales. 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5. Rossi-Forel (RF) Intensity Scale. Quoted from Richter ('1958)” RF Description I Defining Characteristics I. Microseismic shock: Recorded by a single seismograph or by seismographs of the same model, but not by several seismographs of different kinds: the shock felt by an experienced observer. II. Extremely feeble shock: Recorded by several seismographs of different kinds; felt by a small number of persons at rest. III. Very feeble shock: Felt by several persons at rest; strong enough for the direction or duration to be appreciable. IV. Feeble shock: Felt by persons in motion; disturbance of movable objects, doors, windows, cracking of ceilings. V. Shock of moderate intensity: Felt generally by everyone; disturbance of furniture, beds, etc., ringing of some bells. VI. Fairly strong shock: General awakening of those asleep; general ringing of bells; oscillation of chandeliers; stopping of clocks; visible agitation of trees and shrubs; some startled persons leaving their dwellings. VII. Strong shock: Overthrow of movable objects; fall of plaster; ringing of church bells; general panic, without damage to buildings. VIII. Very strong shock. Fall of chimneys; cracks in the walls o f buildings. IX. Extremely strong shock: Partial or total destruction of some buildings. " Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City. CA. 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5. (Continued)." RF Description / Defining Characteristics X. Shock of extreme intensity: Great disaster; ruins; disturbance of the strata, fissures in the ground, rock falls from mountains. " Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City. CA. The Geofian Intensity Scale is another 12-point perceived intensity scale (Table 6). This scale has brief descriptions of the earthquake, as well as brief descriptions of aftereffects of the earthquake. Included in the aftereffects are buildings and structures, residual phenomena in the ground with a change in the status of ground and surface waters, and other characteristics that are described as ‘symptoms’. The categories are somewhat comparable to those used in the MMI, but there is apparently some overlap. (Applied Technology Council, 1985.) Table 6. Geofian Intensity Scale. Quoted from Barosh (1969).° Geofian Description / Defining Characteristics I. Earthquakes not felt by persons. The oscillations of the earth are registered with instruments. II. Noticed by individual persons who are very sensitive and how are perfectly at rest. ° Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City'. CA. 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6. (Continued).0 Geofian Description / Defining Characteristics III. Oscillations noted by a few persons who are at rest inside buildings. Careful observers note only a slight swinging of hanging objects. IV. Light swaying of hanging objects and o f standing automobiles. Slight vibration of liquids in vessels. Slight ringing of densely stacked unstable dishes. Earthquake perceived by most people located indoors. In rare cases sleepers are awakened. Felt by individual people outdoors. V Hanging objects swing noticeably. In rare cases pendulums of wall clocks stop. Water splashes sometimes from filled vessels. Unstable dishes and ornaments on shelves sometimes topple over. Felt by all persons inside buildings and by majority of persons in the outdoors; all wake up. Animals are restless. VI. Hanging objects swing. Sometimes books fall off shelves and pictures shift. Many pendulums of wall clocks stop. Light furniture shifts. Dishes fall. Many persons run out of the houses. Movement of persons unstable. Animals run out of shelter. VII. Chandeliers swing strongly. Light furniture shifts. Books, vessels, and vases fall down. All persons run out of the buildings and in individual cases jump out of windows. It is difficult to move without support. ° Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data fo r California. Redwood City. CA 93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6. (Continued).0 Geofian Description / Defining Characteristics VIII. Some hanging lamps are damaged. Furniture shifts and frequently tilts over. Light objects jump and tilt over. Persons can stand on their feet with difficulty. All run out o f buildings. IX. Furniture topples over and breaks. Animals very panicky. X. Numerous damages to household goods. Animals cry and howl. XI. Loss of life, animals, and property under fragments from buildings. XII. Great catastrophe. A considerable part o f the population is killed by collapse of the buildings. Vegetation and animals destroyed by avalanches and landslides in mountainous regions. ° Excerpted from Applied Technology Council (1985). ATC-13 Earthquake Damage Evaluation Data for California. Redwood City. CA. An example of an isoseismal map of MMI is presented in Figure 8 (Dewey, Reagor, Dengler, & Moley, 1995; EQE, International & California Governor’ s Office of Emergency Services, 1995) This particular map was developed by the United States Geological Survey (USGS) and compiled by the California Governor’s Office of Emergency Services after the 1994 Northridge earthquake. The geographical contours of estimates of MMI were compiled some six months after the event based on reports submitted by postmasters that were on duty during the earthquake. 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 8 . Isoseismal M a p o f Modified Mercalli Intensity Estimates (MMI) d u e t o th e Northridee Earthquake. January 17.1994. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. An isoseismal map of estimates of peak ground acceleration (PGA) due to the Northridge earthquake is presented in Figure 9 (EQE, International & California Governor's Office of Emergency Services, 1995). The geographic distribution of MMI that was reported after the Northridge earthquake is similar to the geographic distribution of PGA. It has been suggested that the energy released in thrust fault earthquakes may be better represented by MMI than PGA (EQE, International & California Governor's Office of Emergency Services, 1995). However, it is also suggested that the apparent correlation between reported MMI and reported building damage in the Northridge and Whittier Narrows earthquakes (both thrust faults) may also be due to a systematic bias towards observing less damage. This bias may be due to improved construction practices and retrofitting programs (EQE, International & California Governor’ s Office of Emergency Services, 1995). 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 9 . IsoseismaI M a p o f L o g P eak Ground Acceleration (PGA) Estimates d u e t o th e Northridee Earthquake, 1/17/94. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.00 STRUCTURAL ENGINEERING BACKGROUND The most pertinent aspects of structural engineering that should be considered in the topic of earthquake hazard assessment address building types and the methodological approach to assessing post-earthquake damage. This section will provide broad descriptions on building types, with a more critical explanation of the various methodologies available for earthquake hazard researchers. Based on the Los Angeles County Assessor’s Office, structures can be divided into five major categories determined by the material used to construct the building and the presence or absence of reinforcement. These structural categories are steel-framed buildings, reinforced concrete-frame buildings, brick or concrete-walled buildings, wood framed buildings, and miscellaneous buildings. Steel-frame buildings are characterized by fireproofed, noncombustible steel frames for all weight-bearing walls, floors and roofs. Concrete-frame buildings are characterized by fireproofed, noncombustible reinforced concrete frames for all weight bearing walls, floors, and roofs. Brick or concrete structures are characterized by exterior walls built of a noncombustible material such as brick, concrete block, or poured-in-place concrete. However, the interior partitions, walls, and roofs o f these types o f structures are built of combustible material. Floors of these structures may be either concrete or wood-frame. Tilt-up wall structures are also included in this category, although they are sometimes categorized in concrete frame structures. The walls in tilt-up buildings are pre-poured on the ground slab and then raised or tilted into their final vertical position 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. after the concrete cures. These types of buildings are usually one story high and are used primarily for industrial purposes (Holmes & Somers, 1996). Wood-frame buildings have either a wood frame or a wood and steel frame. Any buildings that do not fit in the above categories are grouped in the miscellaneous building category (EQE, International & California Governor's Office of Emergency Services, 1995). 3.01 Earthquake Hazard Estimation Methodology. The National Oceanic and Atmospheric Agency (NOAA) was one of the first entities to fund earthquake hazard (or loss) estimation in 1972. The first funded study focused on losses in San Francisco, and it was followed by over thirty major loss studies. In 1989, the Federal Emergency Management Association (FEMA) published a report listing guidelines for conducting earthquake loss estimation studies and channeled funding toward development of this methodology. By 1992, the National Institute of Building Sciences (NIBS) joined with FEMA to help standardize the loss estimation methodology and focus on regional loss studies. In 1993 Risk Management Solutions, Inc. (RMS), a private company that had provided the insurance industry with estimates of losses due to causes other than earthquakes, and the California Universities for Research in Earthquake Engineering (CUREe) were contracted by NIBS to assist in the development of standardized methodology. In 1994, these experts developed the FEMA- NTBS earthquake loss estimation methodology under contract to NIBS (Whitman et al., 1997). The FEMA-NTBS experts developed theories regarding how structures, lifelines (essential utilities including water, natural gas, and electric power systems), emergency 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. medical systems, and communities would respond during specific types of seismic events. The approach was to forecast hazards and loss due to earthquakes through simulations (including Monte Carlo) of multivariate, structural (deterministic) equations, prototyped by civil engineers (Cobum, Spence, & Pomonis, 1992; Durkin & Thiel, 1991; Durkin & Thiel, 1992; Emmi & Horton, 1993; Litan, 1993; Murakami, 1992; Okada, 1993; Samardjieva & Oike, 1992; Shiono, 1992; Shiono, Krimgold, & Ohta, 1991; Steinbrugge, 1990; Stojanovski & Dong, 1992; Yin, 1993). Noticeably missing from the hazard assessment research consortium was any epidemiologic or public health expertise. Earthquake hazard assessment models are typically based on hypothetical data due to the paucity o f actual data for building inventories, structural damage due to earthquakes, and injuries due to earthquakes. Seismic events are relatively rare and there are even fewer events that have meaningful seismic measurements coupled with damage and loss data. Hence, researchers have been forced to rely on hypothetical data. The hypothetical data are usually generated based on the opinions of experts from the field of engineering. The models that have been developed are then applied to a geographical area and are made available to users through a Geographic Information System (GIS). The GIS will maintain matrices of data in a manner so that multiple layers of information may be presented graphically in either the form of a map, a graph, or a table. The GIS allows the user to define specific characteristics o f the earthquake (i.e., magnitude, epicentral location, time of day). Once the specifications of the event are defined, the GIS will estimate losses by applying the data (hypothetical or real) to the specified model. 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. GIS interfaces are particularly useful in the event of a disaster. Government officials used maps in the Northridge earthquake in order to address different needs. An initial map was used to define the scope of the disaster during the first ten hours after the earthquake. This map was based on combined data from seismic sources, seismic activity, geo-technical conditions, and ground motion characteristics in order to estimate ground motion due to the earthquake. Additionally, estimates of MMI (based on aftershocks and improved estimates of ground shaking) were mapped and provided to government officials four days after the event. That map was used to focus recovery efforts. One of the most important applications of that map was its use by state and federal officials to justify expedited funding from the Federal Disaster Housing Assistance Program (FDHAP) (Eguchi et al., 1997). The FDHAP provides funds for alternative housing costs if dwellings are made uninhabitable due to a disaster. The usual process for receiving this funding is to complete an application and wait 2-5 weeks for an inspection in order to verify that the structure is uninhabitable. The maximum allowable amount is $3450 over three months for homeowners and $2300 over two months for renters. However, due to the wide geographic range and relatively dense population associated with the effected area, the requirement for inspection was waived. Instead, the zip code of each applicant was matched with the estimate of earthquake intensity. Homes and apartments located within an MMI of VIII or greater were considered eligible for funding. There were 66 zip codes in these MMI areas. In a “rather bold step, a total o f49,000 checks were sent amounting to $138 million in grants” (Eguchi et al., 1997, EERI Vol. 13, No. 4, p. 826). 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.02 ATC-13 Methodology. Early studies investigating potential damage and economic loss associated with a large earthquake in a California urban environment indicated that this loss might be severe (National Oceanic and Atmospheric Administration, 1972; National Oceanic and Atmospheric Administration, 1973; Steinbrugge, 1990). Therefore, in the early 1980s, FTEMA channeled funding into a comprehensive analysis of this problem. Moore, et al. (1985) developed a computer-based program that modeled damage and loss estimation utilizing simulated data (Moore, Okamoto, Russo, Wilson, & Rojahn, 1985). This computer program is known as the FEMA Earthquake Damage and Loss Estimation System (FEDLOSS) (Applied Technology Council, 1985). In order to estimate the economic impacts of damage and loss, an additional computer simulation methodology was developed by Engineering-Economics Associates (EEA) of Berkeley, California (Applied Technology Council, 1985). This model is known as the FEMA Earthquake Impacts Modeling System (FEIMS) (Applied Technology Council, 1985). In order to incorporate real data in these methodologies, an additional component was required that would allow cross matching of economic sector facility types with actual structural inventories. This methodology was developed by the Applied Technology Council (ATC), a nonprofit corporation, and has been regarded as the standard for earthquake damage and loss estimates that are based on existing (inventoried) facilities (Applied Technology Council, 1985). The advisory panel that was formed by ATC was comprised of senior-level experts in earthquake engineering. The methodology was developed specifically to 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. provide the engineering community with methodologies and data to be used in the FEDLOSS model. Output from the FEDLOSS model was then used as input for the FEIMS model. The ATC project addressed four major issues: 1) establish earthquake shaking characterization; 2) develop an appropriate facility classification scheme; 3) develop methodology to estimate earthquake-related damage and loss; and 4) develop an inventory of data consistent with the adopted facility classification scheme.(Council, 1985) A TC-13 Methodology Part I: Earthquake Shaking Characterization. Although several earthquake intensity scales exist, the MMI was selected as the most appropriate scale. This decision was based on the preponderance of knowledge and existing data associated with earthquakes in the United States (Applied Technology Council, 1985). A TC-13 Methodology Part II: Facility Classification Scheme. Two types of facility classification schemes were used in the ATC-13 methodology. The Earthquake Engineering Facility Classification scheme categorizes structures based on size, structural system, and structure type. The 78 classes o f structures (40 buildings and 38 other structure types) can be grouped into thirteen different structural groups, and these are summarized in Table 7 (Applied Technology Council, 1985). The Social Function Classification system categorizes structures by their economic function. The 35 classes of facilities can be grouped into twelve economic types, and these are summarized in Table 8 (Applied Technology Council, 1985). 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 7. ATC-13 methodology: Earthquake Engineering Facility Classification. Structure Type Structure Class Facility Number Building Wood Frame (Low Rise*) 1 Building Light Metal (Low Rise*) 2 Building Unreinforced Masonry: Bearing Wall (Low Rise*) 75 Building Unreinforced Masonry: Bearing Wall (Medium Rise**) 76 Building Unreinforced Masonry: Load Bearing Frame (Low Rise*) 78 Building Unreinforced Masonry: Load Bearing Frame (Medium Rise**) 79 Building Unreinforced Masonry: Load Bearing Frame (High Rise***) 80 Building Reinforced Concrete Shear Wall with Moment-Resisting Frame (Low Rise*) 3 Building Reinforced Concrete Shear Wall with Moment-Resisting Frame (Medium Rise**) 4 Building Reinforced Concrete Shear Wall with Moment-Resisting Frame (High Rise***) 5 Building Reinforced Concrete Shear Wall without Moment-Resisting Frame (Low Rise*) 6 Building Reinforced Concrete Shear Wall without Moment-Resisting Frame (Medium Rise**) 7 Building Reinforced Concrete Shear Wall without Moment-Resisting Frame (High Rise***) 8 Building Reinforced Masonry Shear Wall without Moment-Resisting Frame (Low Rise*) 9 Building Reinforced Masonry Shear Wall without Moment-Resisting Frame (Medium Rise**) 10 Building Reinforced Masonry Shear Wall without Moment-Resisting Frame (High Rise***) 11 Building Reinforced Masonry Shear Wall with Moment-Resisting Frame (Low Rise*) 84 Building Reinforced Masonry Shear Wall with Moment-Resisting Frame (Medium Rise**) 85 Building Reinforced Masonry Shear Wall with Moment-Resisting Frame (High Rise***) 86 Building Braced Steel Frame (Low Rise*) 12 Building Braced Steel Frame (Medium Rise**) 13 Building Braced Steel Frame (High Rise***) 14 Building Moment-Resisting Steel Frame: Perimeter Frame (Low Rise*) 15 Building Moment-Resisting Steel Frame: Perimeter Frame (Medium Rise**) 16 Building Moment-Resisting Steel Frame: Perimeter Frame (High Rise***) 17 Building Moment-Resisting Steel Frame: Distributed Frame (Low Rise*) 72 Building Moment-Resisting Steel Frame: Distributed Frame (Medium Rise**) 73 Building Moment-Resisting Steel Frame: Distributed Frame (High Rise***) 74 Building Moment-Resisting Ductile Concrete Frame: Distributed Frame (Low Rise*) 18 Building Moment-Resisting Ductile Concrete Frame: Distributed Frame (Medium Rise**) 19 ^ Building Moment-Resisting Ductile Concrete Frame: Distributed Frame (High Rise***) 20 2 Building Moment-Resisting Non-Ductile Concrete Frame: Distributed Frame (Low Rise*) 87 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 7. (Continued). Structure Type Structure Class Facility Number Building Building Building Building Building Building Building Building Bridges Bridges o Pipelines Pipelines __ Dams D a m s _______________ Tunnels Tunnels Tunnels_____ Storage Tanks Storage Tanks Storage Tanks Storage Tanks Storage Tanks Storage Tanks _ _ _ _ _ ___ Roadways and Pavements Roadways and Pavements Roadways and Pavements Chimneys (High Industrial) Chimneys (High Industrial) Chimneys (High Industrial) C ranes_______________ Conveyor Systems______ Moment-Resisting Non-Ductile Concrete Frame: Distributed Frame (Medium Rise**) Moment-Resisting Non-Ductile Concrete Frame: Distributed Frame (High Rise***) Precast Concrete other than Tilt-Up (Low Rise*) Precast Concrete other than Tilt-Up (Medium Rise**) Precast Concrete other than Tilt-Up (High Rise***) Long-span (Low Rise*) Tilt-Up (Low Rise*) Mobile Homes Conventional (less than 500-foot spans): Multiple Simple Spans Conventional (less than 500-foot spans): Continuous/Monolithic Major (greater than 500-foot spans) Underground A t Grade Concrete EarthfiHand Rockfill ______ ___ Alluvium Rock Cut and Cover __ Underground: Liquid Underground: Solid On Ground: Liquid On Ground: Solid Elevated: Liquid Elevated: Solid Railroad Highways Runways Masonry Concrete Steel _ _ _ (one class) _ ___________________ _ __________________ (one class) _ ______________________________ ____________________ 88 89 81 82 83 91 21 23 24 25 30 31 32 35 36 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 7. IContinuedI. Structure Type Structure Class Facility Number Towers Electrical Transmission Line: Conventional (less than 100-ft high) 55 Towers Electrical Transmission Line: M ajor: (more than 100-ft high) 56 Towers Broadcast 57 Towers Observation 58 Towers Offshore 59 Other Canals 61 Other Earth Retaining Structures (over 20-ft high) 62 Other Waterfront Structures 63 Equipment Residential 64 Equipment Office (furniture, computers, etc.) 65 Equipment Electrical 66 Equipment Mechanical 68 Equipment High Technology & Laboratory 70 Equipment Trains, Trucks, Airplanes, & Other Vehicles 90 * 1-3 Stories “ 4-7 Stories * ” 8+ Stories o 0\ Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 8. ATC-13 Methodology: Economic Social Function Classification. Economic Type Structural Use Reference Number Residential Permanent Dwelling 1 Residential Temporary Lodging 2 Residential Group Institutional M ousing 3 Commercial Retail Trade 4 Commercial Wholesale T rade 5 Commercial Personal and Repair Services 6 Commercial Professional, Technical & Business Services 7 Commercial Health Care Services 8 Commercial Entertainment and Recreation 9 Commercial Parking 10 industrial Heavy Fabrication and Assembly 1 1 Industrial Light Fabrication and Assembly 12 Industrial Food and Drugs Processing 13 Industrial Chemicals Processing 14 Industrial M etal and Minerals Processing 15 Industrial H igh Technology 16 Industrial Construction 17 Industrial Petroleum 18 Agriculture Agriculture 19 M ining M ining 2 0 ............ Religion & Nonprofit Religion & Nonprofit 21 Government General Services 22 Government Emergency Response Services 23 Education Education 24 Transportation (Freight & Passenger) Highway 25...... " " Transportation (Freight & Passenger) Railroad 26 Transportation (Freight & Passenger) A ir 27 Transportation (Freight & Passenger) Sea/Water 28 Utilities Electrical 29 Utilities Water 30 Utilities Sanitary Sewer 31 Utilities Natural Gas 32 Utilities____________________ Telephone & Telegraph 33 Communication (Radio & T V ) Communication .....3 4 ......... Flood Control Flood Control 35 .... ' A TC-13 Methodoloev: Part III: Earthquake Damage and Loss Estimation. Four categories of loss are estimated in FEDLOSS: 1) structural damage caused by ground shaking; 2) losses due to collateral hazards; 3) loss o f function; and 4) estimates of death and injury. Estimates of the percentage of physical (structural) damage caused by ground shaking were arrived at via a questionnaire process and corresponding damage probability matrices (DPM). A questionnaire was administered at three different times to the members of the advisory board as well as 58 other selected earthquake engineering experts. The questionnaire required the respondent to provide low, best, and high estimates of the damage factor (percent damage) to selected facility classes for MMI levels VI through XII. During the first administration of the questionnaire, the respondent was also required to provide a self-evaluated degree o f confidence in the various estimates of damage. During the second administration, the respondents were provided summary graphs of all the other respondents3 answers as well as their own responses. They were then asked to re-evaluate their first-round responses after considering the patterns of responses from other experts. The third administration of the questionnaire was identical to the second round, but the re-evaluations were based on each respondent’s second estimate of damage. After the third administration of the questionnaire, the data were examined to determine the most appropriate probability distribution (Beta, normal or log-normal) associated with the estimates of damage. The Beta distribution was used to develop damage probability matrices (DPM) for the 78 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. engineering facility classes. An example of a DPM is abbreviated in Table 9 (Applied Technology Council, 1985). Table 9. A TC-13 Methodology: Example o f a Damage Probability Matrix Based on Expert Opinion. CENTRAL MODIFIED MERCALU INTENSITY DAMAGE ------------------------------------------------------------------------------------------------------------- FACTOR_______ Vj________ vjj________VIII_______ |X________ X_________ X I______ Xll_ 0.00 18.1 0.50 69.8 17.8 0.6 — — — "* 5.00 12.1 82.2 97.7 71.8 14.6 0.3 *** 20.00 *** 1.7 28.2 83.2 68.8 29.4 45.00 — * * * " * • “ 2.2 30.9 70.4 80.00 *** « «« .« .«* « q 2 1 0 0 0 0 * * * • * * * t « * * * • * « M t **• Very small probability____________________________________________________________________________________________ Collateral hazards are potentially dangerous environments that are not defined solely by ground shaking. These types of environments include areas of ground failure, fault rupture, water inundation, and fire. Most of the estimates of probability of damage due to collateral hazards that are generated through ATC-13 Methodology are based on advisory board consensus opinions, since little or no data exists for these specific earthquake hazard sources. Estimates are provided for damage caused by four general environments: 1) ground failure or liquefaction hazards; 2) landslides; 3) fault rupture; and 4) inundation. No methodology is included in ATC-13 Methodology for estimating damage due to fire because this hazard was considered too difficult to estimate (Applied Technology Council, 1985). 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Loss of functionality is based on two premises in ATC-13 Methodology. The first assumption is that direct damage to individual facilities impacts the loss of function at the facility and the time required to restore function. The second assumption is that direct damage to lifelines on which individual facilities depend impacts the loss of function and the time required to restore function. Since very little data exists on which estimates of functionality and time to restore function can be based, expert opinion was (again) solicited. Estimates of functional loss and restoration time were accomplished using a questionnaire administered in a manner similar to that used to arrive at estimates for damage caused by ground-motion. Members of the advisory board and 29 additional experts were asked to estimate the time required to restore function at facilities for three different functionality levels (30%, 60%, and 100%). The questionnaire was administered three times, with the second and third administrations associated with revisions of estimates based on comparative standings with other respondents. Weighted-mean restoration times for each social class (structure use) category were calculated for each damage level (MMI VI-XII) and each functionality level (30%, 60%, 100%). An example of these weighted estimates for some structural use categories is presented in Table 10. Note that not all 29 expert opinions were included in each of the various structural classification categories. For example, the social function class o f ‘16’ incorporated opinions from only three experts for calculated functionality loss estimates. In fact no more than nine expert opinions were included in any one social class category (Applied Technology Council, 1985). 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 10. A TC-13 Metodoloav: Estimates of Weighted Loss o f Function Restoration Time (In Davsl bv Social Function Classification. SOCIAL FUNCTIONAL CLASSES 1. 2, AND 3: OS NEXP MEAN 30 MIN 30 MAX 30 SDEV 30 MEAN S O MIN 60 MAX SO SDEV 60 MEAN 100 MIN 100 MAX 100 SDEV 100 2 8 0.2 0 0 10 0.4 02 0.0 10 0.4 0 8 0.0 3 0 1.1 3 8 0 3 0 0 20 07 15 0 0 6 0 21 3 3 0 0 100 3 6 4 8 19 0.0 5 0 1.7 5 4 0 0 150 49 105 3 0 300 9 0 5 8 152 20 600 16.7 30.5 5 0 1200 332 719 110 2400 61.7 6 8 572 300 1800 46.7 938 550 2400 542 1466 90.0 3650 810 7 8 a • * 4 a 4 4 a 2119 1500 365.0 62.7 SOCIAL FUNCTIONAL CLASS IS : DS NEXP MEAN 30 MIN 30 MAX 30 SDEV 30 MEAN SO M/M SO MAX SO SDEV 60 MEAN 100 MIN 100 MAX 100 SDEV 100 2 3 0.0 00 0 0 0.0 0.0 0.0 0 0 0 0 1 1 0 0 3 0 1.4 3 3 47 00 100 42 5 5 0 0 100 40 165 140 200 2.7 4 3 368 150 600 185 559 450 600 67 1118 600 1800 52.9 5 3 1364 600 2400 792 198 2 900 3660 126 7 258 2 900 4850 1751 6 3 198.2 900 3660 126 7 281 1 900 5480 2049 4291 1800 7300 238 3 7 3 • a a • • 4 a a 6120 5400 7300 893 N q M w : DS Damage state (DS 2 = MMI VI, DS 3 = MMI VII; DS 4 = MMI VIII; etc.) NEXP Number of experts MEAN_30 Mean time (days) to restore 30% of usability MIN_30 Minimum time (days) to restore 30% of usability MAXJ30 Maximum time (days) to restore 30% of usability SDEV_30 Standard deviation of time to restore 30% of usability MEAN_60 Mean time (days) to restore 60% of usability MINJ30 Minimum time (days) to restore 60% of usability MAXJ30 Maximum time (days) to restore 60% of usability SDEV_80 Standard deviation of time to restore 60% of usability MEAN_100 Mean time (days) to restore 100% of usability MIN_100 Minimum time (days) to restore 100% of usability MAX_100 Maximum time (days) to restore 100% of usability SDEV_100 Standard deviation of time to restore 100% of usability ’ __________Statistics not provided because experts did not provide numerical responses for these levels Deaths and injuries are considered collateral losses in the ATC-13 methodology. These types of losses, in the event of a severe earthquake, are assumed to be due to structural failure of man-made facilities (i.e., buildings and dams). Injury and death rates were therefore based on the estimate of total damage to a structure, and are provided for two categories of construction: 1) light steel and wood-frame construction; and 2) all other types of construction (Applied Technology Council, 1985). Earthquakes that occurred in the United States for a 100-year period (1872-1972) were reviewed by the authors o f the ATC-Methodology, and a summary table was compiled and is presented in Table 11. The authors o f the ATC-13 Methodology acknowledge that death rates from historical events may not be applicable for current California construction standards. However, the results of these historical earthquakes are referred to as a “useful guideline when used with judgment and in the context of the time of day, comparative construction, and appropriate Modified Mercalli Intensities” (Applied Technology Council, 1985, p. 257). Based on the information presented in Table 11, NOAA proposed a death rate of 12/100,000 for wood-frame residences and 50/100,000 for all other persons in the event of a (Richter) magnitude 8.3 earthquake on the San Andreas fault in the bay area of San Francisco. They additionally proposed a ratio of serious injuries (those requiring hospitalization) to deaths as 4:1, and a ratio of minor injuries to deaths as 30:1 (Applied Technology Council, 1985). 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 11. ATC-13 Methodology: National Oceanic and Atmospheric Association Review o f Historical Earthquakes. Earthquake Location and Damage Location Richter Magnitude MMI (Max) Date Serious Death Injury Time Rate* Rate* Owens Valley, CA we ee March 26,1872 ee ee ee Lone Pine *e ee March 26, 1872 e* 8,000 ** Charleston, SC ee X August 31, 1886 9.51 PM 45 ee San Francisco, CA 8.3 X I April 18,1906 5:12 AM ee ee San Francisco 8.3 X I April 18, 1906 5:12 AM 124 104 Santa Rosa 8.3 ee April 18,1906 5:12 AM 116 69 San Jose 8.3 V III April 18, 1906 5:12 AM 80 38 Santa Barbara, CA 6.3 IX June 29,1925 6:42 AM 45 119 Long Beach, CA 6.3 IX March 10,1933 5:54 PM 26 1,300 Imperial Valley, CA 7.1 X May 18,1940 8:37 PM 18 40 Puget Sound, WA 7.1 V III April 13,1949 11:56 PM 1 ee Kern County, CA 7.7 X I July 21, 1952 4:52 AM ee ee Tehachapi 7.7 X I July 21,1952 4:52 AM 500 ee Bakersfield, CA 5.8 V III August 22,1952 3:41 PM 3 47 Anchorage, Alaska 8 4 X I March 27,1964 5:36 PM 9 315 Seattle-Tacoma, WA 6.5 V III April 29,1965 7:29 AM 1.5 ee San Fernando, CA 6.4 X I February 9,1971 6:01 AM 64 180 Excluding VA Hospital 12 * Events per 100,0(30 population “ Not Available Notes: Table does not Include events with fewer than 8 deaths. Table Is adapted from NOAA (1972) and Anagnostopoulos and Whitman (1977). These reported rates of death and injury are questionable from an epidemiological point of view. Although one might expect that deaths and death rates could be retrieved in the event of an earthquake, Peek-Asa, et al., found that upon epidemiologic review of coroner’s records in Los Angeles County, 33 people died as a result of the Northridge Earthquake (Peek-Asa et al., 1998). Previous reports o f deaths ranged from 57 (reported by the Earthquake Engineering Research Institute) to 72 (reported by M. Durkin, an independent researcher), but these estimates were not based on standard epidemiologic methodology (Holmes & Somers, 1996; Marquis, 1998). A similar problem was evident with the reports of serious (hospitalized) injuries. Peek-Asa reported 138 hospitalizations while other estimates ranged from 1,032 (Red Cross Survey reported by Earthquake Engineering Research Institute) to 1,525 (Los Angeles County Emergency Medical Services Agency Survey reported by Earthquake Engineering Research Institute) (Holmes & Somers, 1996; Peek-Asa et al., 1998). Problems acknowledged in the reporting of the latter two estimates include an inconsistent time-frame for the reporting periods, inconsistencies in case definition, missing or incomplete information needed in order to attribute the injury to the earthquake, and lack of participation from facilities. The ratio of deaths to hospitalizations from Peek-Asa’s study (4.2 : 1) is in basic agreement with NOAA guidelines (4:1), however the death and injury rates are not reproducible, particularly by building type. Based on the non-epidemiological survey administered through the Red Cross after the Northridge earthquake, the ratio of hospitalized patients to those who were 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. treated and released was 1:10, rather different than the NOAA estimate of 1:30 (Applied Technology Council, 1985; Holmes & Somers, 1996; Peek-Asa et al., 1998). Whitman, et al. (1975) proposed another scheme for estimating deaths and injuries based on a six-category, qualitative assessment of structural damage, and central damage ratio for each damage level (Whitman, Cornell, & Taleb-Agha, 1975). The fraction injured and dead is then assumed to be constant for each damage level. This scheme is presented in Table 12. Again, validation o f the assumption of constant fractions of injured and dead people at each damage level is problematic and hinges on difficulties obtaining reliable injury data by structural damage level. Table 12. ATC-13 Methodology: Whitman. Cornell, et aL (1975). Death and Injury Proportions. D am age State Central D am age Ratio (CDRr Injury Rate Death Rate None 0.0 0 0 Light 0.3 0 0 Moderate 5.0 1 /1 0 0 0 Heavy 30.0 1 / 50 1 /4 0 0 Total 100.0 1 /1 0 1 /1 0 0 Collapse 100.0 1 1 / 5 • E xpressed in P ercen t 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Based on the NOAA guidelines, the results of the Whitman, et al. study, and judgmental evaluation and consensus from the ATC-13 advisory board, death and injury rates were developed for the ATC-13 methodology. These rates are based on the seven damage states of MMI VI-XII, central damage factors (based on structure and contents) that correspond to each damage-state, and associated rates of injury and death. The rates are calculated for two construction types: 1) light steel and wood-frame; and 2) all other construction types. The results are summarized in Table 13 (Applied Technology Council, 1985). Table 13. ATC-13 Methodolonrv: Irtiurv and Death Rates oer Damage State. CDF2 Minor Injury S e rio u s Injury Damage S ta te 1 Rate R ate Death Rate Non-Light Steel or Non-Wood-Frame Construction: 1 0.0 0 0 0 2 0.5 3 /1 0 0 ,0 0 0 1 / 250,000 1 / 1,000,000 3 5.0 3 /1 0 ,0 0 0 1 / 25,000 1 /100,000 4 20.0 3 /1 ,0 0 0 1 / 2,500 1 / 10,000 5 45.0 3 /1 0 0 1 / 250 1 / 1,000 6 80.0 3 /1 0 1 / 25 1 /1 0 0 7 100.0 2 / 5 2 / 5 1 / 5 Light Steel or Wood-Frame Construction: 1 0.0 0 0 0 2 0.5 3/1,000.000 1 / 2,500,000 1 /10,000,000 3 5.0 3 /1 0 0 ,0 0 0 1 / 250,000 1 / 1,000,000 4 20.0 3 /10,000 1 / 25,000 1 /100,000 5 45.0 3 /1 ,0 0 0 1 / 2,500 1 /1 0 ,0 0 0 6 80.0 3 /1 0 0 1 / 250 1 /1 ,0 0 0 7 100.0 2/50 2 / 5 0 1/50 ' Dam age S tate (D S) 1 = MMI VI: DS 2 = MMI VII: DS 3 = MMI VIII; etc. * Central D am age Factor, including d am ag e to the structure and contents Note: Death and injury rates are estimates based on consensus of ATC-13 advisory board. NOAA (1972) estimates, and estimates from Whitman. Comal, at al. (197S). 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A TC-I3 Methodology Part I I : Data Inventory for Facility Classifications. The inventory methodology was developed to use data that is available to FEMA. Ideally, the inventory would itemize man-made facilities and provide an inventory of facility contents, facility use, structural classification, facility replacement value, and the number of occupants or users of each facility. Since few facility inventory databases exist, most of the structural inventory used in the ATC-13 methodology is synthesized from economic data. This synthesis has been based on existing data available to FEMA or Engineering-Economics Associates (EE A). If no databases were available to those agencies, the synthesis would be based on either population data or any other available data. The ATC-13 guidelines emphasize that the data inventory methodology has not been tested and that it could be improved by a scientific approach to sampling actual facilities (Applied Technology Council, 1985). 3.03 Correlation between Ground Motion and Structural Damage. Engineers are still refining methodology that will more accurately reflect structural damage due to earthquakes. Controversy exists regarding the correlation between peak ground acceleration and structural damage. Specifically, elastic acceleration response spectra (elastic spectra), are said to “provide better insight into the demands imposed by ground motion on structures than does peak ground acceleration” (Holmes & Somers, 1996, p. 2). Although peak ground acceleration influences structural responses, predicting individual structural responses based on ground acceleration is not as accurate as actually measuring the structural elastic spectra during an earthquake 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Kircher. Reitherman, Whitman, & Arnold, 1997a). However, the feasibility of obtaining these spectra for each structure in a community is not realistic. It is also questionable whether a sample of given structure types can provide elastic spectra that are representative of all responses during a seismic event for that structure type. This question regarding generalizability is due to variation in soil types, construction code compliance, and other environmental variables. An example of elastic spectral readings for specific structures is presented in Figure 10, along with peak ground accelerations at the structure sites (Holmes & Somers, 1996) Elastic spectra are structure-specific, and are referred to as ffee-field (FF) or structure-based recordings of ground motion. Notable in this figure is the difference in PGA relative to elastic spectra. The reported peak ground acceleration for Sylmar (0.84 g) was less than the PGA reported for Santa Monica (0.88 g). However, the strength demand spectra for the structure in Sylmar (Olive View Medical Center) peaked at a higher level than the strength demand spectra for the structure in Santa Monica (Santa Monica City Hall). (Holmes & Somers, 1996.) An additional component in the elastic spectra that helps explain damage to structures is the amplitude period (reported in seconds). The longer period of amplification reported at Santa Monica City Hall has been associated with damage to taller structures in Santa Monica including the 15-story Champaign Tower on Ocean Avenue (Holmes & Somers, 1996). As previously mentioned, however, one cannot obtain elastic spectra on all structures in a community, particularly an urban area. There are also legal ramifications to estimating and reporting damage to certain types of 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Finn re 10. (iraphical Comparison o f Elastic Strength Demand Spectra Recorded during the Northridtie Earthquake, January 17, 1994.p Damping = 5% 3 - Santa Monica FF ( PGA = 0.88 g, r= 24 km) - Sylmar FF ( PGA = 0.84 g, r = 15 km ) - Newhall (PGA ■ 0.59 g, r = 19km) - Arleta FF ( PGA = 0.34 g, r * 9 km) • ATC3-S2 ( PGA = 0.4 g) 2 > > U . 0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Period (sec) Elastic acceleration response spectra (strength dem and) lor Tour frec-field or structure base ground m otions, as com pared with the present code design spectrum . V V = weight of oscillator mass; Fy - required lateral strength for elastic response. p Excerpted from Eanhquakc Engineering Research Institute (19%). Northridge Eanhquakc of January 17,1994 Rcconnaisancc Report, Volume 2. — Earthquake Spectra. vo structures (i.e., military installations, nuclear power facilities, large stadiums, hi-rise housing, and unique skyscrapers) (Whitman et al., 1997). Therefore, this type of information is not likely to be readily available in the near future. 4.00 RECENT ADVANCES IN EARTHQUAKE HAZARD ESTIMATION. There are two main schools of research regarding earthquake hazard estimation. The older of these schools is the theoretical estimation process that has been used by risk management (usually on a statewide or national level) in the long-term preparatory period before an earthquake. Models developed through this school are typically focused on a single, large event, and are based on ATC-13 guidelines and existing knowledge of regional geology and seismology, and hypothetical data (Eguchi et al., 1997; Whitman et al., 1997). These models have been geared toward preparing municipal governments and utilities for damage from a single, large event in a metropolitan area (Eguchi et al., 1997). The most noteworthy methodology that falls into this category is the previously mentioned FEMA-NIBS loss estimation methodology. 4.01 FEMA-NIBS Hazard Assessment Methodology. The FEMA-NIBS methodology is designed to work through a software package that was developed specifically for earthquake loss estimation. The software is known as HAZUS and operates through mapping software (Maplnfo, a GIS application), C++ computer language, and a relational database management program. Graphical depictions of the modules that form HAZUS are presented in Figure 11 (Whitman et al., 1997). This modular approach to the methodology allows the user to determine the 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Fixiire II. HAZUS Methodology Modules.*1 Direct Losses Indirect Losses Direct Damage Structural Inventory Indirect Damage PESH (Ground Motion Site Effects) degree of complexity o f the model. Estimates of loss may be based on simple models and limited inventory, or they may be more refined based on extensive inventories and detailed analyses of existing data (Whitman et al., 1997). The starting point for the estimation process is the Potential Earth Science Hazard (PESH) module. A scenario earthquake is chosen by the user and may be one of three approaches: 1) a deterministic model for a scenario event; 2) a model based on probabilistic spectral response contour maps developed by the United States Geological Survey (USGS); or 3) a model based on user-supplied digitized maps of ground motion. q Adapted from Whitman RV. Anagnos T. Kircher CA et al (1997): Development of a national earthquake loss estimation methodology. Earthquake Spectra. 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The last option was created so that users could apply well-recorded motion from historical events to investigate the level of agreement or disagreement between predicted and observed losses (Whitman et al., 1997). After the user chooses the earthquake scenario, input ground motion levels are determined based on soil conditions. These ground motion levels determine site-specific response spectra through attenuation equations that have been adopted and employed by the USGS. Additionally, user-supplied data regarding soil types and ground characteristics may be incorporated into the model. It is recommended that this is done on a local basis since the default settings for soil conditions will assume that all soil for the region is of medium stiffness with a specific shear wave velocity between 180 and 360 m/sec. Prior experience with earthquake-related losses has shown that regional variation in soil conditions effects the likelihood of losses (Whitman et al., 1997). It is also advised that users develop an enhanced inventory o f structures that reflects regional characteristics regarding building stock, and incorporate this information into the model (Whitman et al., 1997). However, if this is not possible, the methodology includes default data that is based on the most recent census (Whitman et al., 1997). The inventory of structures is classified by occupancy use (residential, commercial, etc.,), and by building types (36 types based on compositional material and height) (Whitman et al., 1997). Default data is admittedly limited, especially regarding utility lifelines (Whitman et al., 1997). This type of limitation has important implications for numbers and severities of injuries, as will be explained in Section 4.03, Interdisciplinary Advances o f Hazard Estimation. 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A notable improvement in the current HAZUS methodology compared to earlier methods, especially for estimating severity of building damage, involves incorporation of spectral response patterns for specific sites. Earlier methodologies relied on non engineering, subjective parameters such as MMI. As mentioned earlier, spectral response is generally acknowledged in the engineering community to be more closely correlated to building damage than either MMI or PGA (Holmes & Somers, 1996). The damage states associated with structural loss are broken into five qualitative categories for HAZUS: no damage, slight damage, moderate damage, extensive damage, and complete damage or structural collapse. HAZUS has incorporated structural fragility curves based on site-specific spectral responses of the 36 building types. These fragility curves are lognormal functions that describe the probability of reaching or exceeding the given qualitative level of destruction (structural damage), based on the median spectral response for that structure type and damage state. An example of fragility curves at different levels of damage is presented in Figure 12. Although these curves consider the variability and uncertainty associated with the expected curve properties at maximum capacity, damage state, and ground movement, the curves are essentially theoretical with no mention of validation or calibration in either the lab or the field (Kircher et al., 1997b). 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 12. Examples o f Building Fragility Curves for Slight. Moderate. Extensive, and Complete Damage.r 1.0 15 0.5 0.0 Slieht M oderate C om plete m m m m Weak Medium Strong Shaking Shaking Shaking Spectral R esponse ’ Excerpted from Kirchcr CA. Reitherman RK. Whitman RV. Arnold C (1997); Estimation of -earthquake losses to buildings; Earthquake Spectra. 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In the general formulation of fragility curves, the lognormal standard deviation (P) has been broken down to two components: a) randomness (Pr), that inherent variability in structural responses: and b) uncertainty (Pu), variability that could be reduced by improving knowledge of structural responses (Kennedy, Cornell, Campbell, Kaplan, & Perla, 1980). In terms of seismic probability, these two components are not easily distinguished, therefore the combined variability (P) is employed in fragility curve estimation. The conditional probability of being in (or exceeding) a given damage state given the spectral displacement is defined by Equation 1, taken from Kircher, Nassar, et al„ (1997): Three sources contribute to the total variability of fragility curve at any damage state. These sources include the variability associated with the capacity (maximum) Equation 1 . Where: ds = damage state Sd = spectral displacement X d.ds = median value of spectral displacement at which the building reaches the damage state (ds) threshold. Pds = standard deviation of the natural logarithm o f spectral displacement for damage state (ds) < t > = standard normal cumulative distribution function 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. curve, the demand (elastic) spectrum, and the discrete threshold of each damage state (Kircher et al., 1997b). It has been pointed out in the literature that reducing damage- state variability would have a significant effect on the probability of being in that damage state if that probability is small (Kircher et al., 1997b). For example, the two highest damage states (extensive and complete) typically exhibit a low probability of occurrence (less than 0.10), even for events that exhibit strong ground shaking (Kircher et al., 1997b). Since subsequent injuries are dependent on these damage states, attenuation of damage state variation should also help reduce variation in injury estimates. However, if structure-specific injuries can be linked with structural damage states, perhaps the variation in these fragility curves may be indirectly attenuated. However, the basic differentiation between structural mechanisms o f injury compared to non-structural mechanisms of injury must be distinguishable. In contrast, the literature indicates that economic loss is less sensitive to damage- state variability because all damage states are thought to contribute significantly to economic loss (Kircher et al., 1997b). The same rationale may be applicable regarding non-severe injuries if all damage states are assumed to contribute to non-severe injuries. However, this assumption has not been investigated in the literature. Problematic with the HAZUS approach of loss estimation based on spectral response is the generalization of specific spectral responses to groups of structures. It is general knowledge that there is much variation between structure types (and consequently spectral responses) depending on contractors’ practices, date of construction, initiation of legislation to improve construction practices, and stability o f 126 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the ground on which the structure stands (Kircher et ai., 1997b; Lay & Wallace, 1995). It is also acknowledged that there are always surprises in structural responses during earthquakes (Scientists of the United States Geological Survey and the Southern California Earthquake Center, 1994; Holmes & Somers, 1996). 4.02 EPEPAT Methodology. Networks for providing ongoing surveillance of different levels of ground motion have advanced with the development of high speed computing and satellite telemetry. The Southern California Seismic Network, operated jointly by the California Institute of Technology and the United States Geological Survey, has provided data on real-time (rapid post-event) broadcasts of earthquakes greater than magnitude 3.0 on an experimental basis since the early 1990s. This information dissemination system is known as the Caltech-USGS Broadcast-of-Earthquakes (CUBE) system (California Institute of Technology & United States Geological Survey Broadcast of Earthquakes (CUBE), 1992). The CUBE system is comprised of approximately 200 seismometers and accelerometers distributed at various sites around Southern California. This network has located more than 10,000 earthquakes every year. The CUBE system identifies earthquakes after compilation of seismograms is completed (from remote instruments) at the Seismological Laboratory at Caltech. These seismograms are transmitted via telephone lines or radio waves immediately after registering on the seismometer and the verification and compilation of these data is very fast, usually within a few minutes. Once the data are compiled and verified, the information is transmitted via modem or 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. radio to a commercial paging system that contacts subscribers with pertinent information, (i.e., magnitude, latitude, longitude, depth, and time of occurrence) (California Institute of Technology & United States Geological Survey Broadcast of Earthquakes (CUBE), 1992). Concurrent with the CUBE system, EQE International, Inc., an independent engineering and research group, was retained by the Governor’s Office o f Emergency Services (OES) for the State of California. EQE International was contracted by OES to design a GIS-based information system in order to model hazards and losses rapidly after earthquakes were identified through CUBE. This has lead to the development of the first earthquake hazard model that attempts to provide a real-time estimate of building and lifeline damage as well as estimates of injuries and deaths based on ground motion surveillance. This hazard assessment system is named the Early Post-Earthquake Damage Assessment Tool (EPEDAT) (Eguchi et al., 1997). The EPEDAT methodology has been specific in detailing steps to calibrate estimates of predicted ground motion (i.e., additional earthquakes or aftershocks o f the initial event) with CUBE information as it is broadcast. The methodology has also detailed the calibration of estimates of structural damage to actual field assessment of damage including satellite and aerial photographs as these become available after a seismic event. However, the ability to calibrate estimates of injuries is hampered by the limitations that have been addressed in previous sections of this dissertation regarding injury data. Specifically, there is no ongoing surveillance of injuries or centralized injury registry in Los Angeles (Bourque, 1997; Mahue & Weiss, 1996a). Furthermore, as will 128 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. be detailed more in Chapter 3, record-keeping, even at institutions that are accustomed to maintaining records (i.e., emergency departments), is in a state o f disarray immediately after a disaster (Bourque et al., 1997; Mahue & Weiss, 1996b). Emergency department logs, for example, may be incomplete as long as two weeks after the event (Mahue & Weiss, 1996b). Therefore, if officials are interested in more accurate and timely estimates of injured people, modifications regarding data collection and standardization need to be implemented particularly for institutions that are likely to be treating or triaging injured people (i.e., outpatient clinics, Red Cross shelters, and emergency departments). The casualty aspect of EPEDAT (and other hazard assessment methodologies) is not the only weak piece. Both structural engineering components and geologic measurements also exhibit limitations. Regarding structure type and generalizability within broad groupings of structures, existing data reflecting damage from actual earthquakes is limited. The only construction type that has been studied well in past earthquakes is for single-family, wood-frame structures (Eguchi et al., 1997). In Southern California, however, as emphasized in the Northridge earthquake, the largest contributor to structurally related hazards is more likely to be associated with multiple- unit residences rather than single-family homes. Regarding geologic components, the EPEDAT model employs peak ground acceleration rather than site-specific spectral responses. As mentioned earlier, controversy exists regarding the usefulness of peak ground acceleration with respect to estimates of structural damage (Holmes & Somers, 1996). 129 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The EPEDAT methodology does have some major advantages over competing methodologies. In addition to the use of ongoing ground motion surveillance information, the EPEDAT system is distinguished by methodology that is not based exclusively on ATC-13 guidelines. Specifically, EPEDAT allows calibration of model parameters by incorporation of actual data from seismic events in an iterative manner. .Another advantage to the EPEDAT system is the estimate of MMI that is based on measured ground acceleration rather than subjective perceptions and reports of ground motion. MMI is frequently reported in the media and non-scientific journals because of its ease of interpretation. However, from a scientific point of view, instrumental measurements are more likely to be reliable and valid than subjective assessment of ground shaking. Additionally, there is a circularity of reasoning when perceived intensity is used to predict intensity. 4.03 Interdisciplinary Advances in Hazard Estim ation. Recent advances in earthquake hazard estimation have emphasized the importance of individual research not only from engineers and seismologists, but also from socioeconomic researchers (Shinozuka et al., 1997). Two of these studies have focused on loss estimation simulations for Memphis, Tennessee, which is at risk for earthquakes stemming from the New Madrid Seismic Zone. These studies were both conducted through the National Center for Earthquake Engineering Research (NCEER) and are noteworthy because the estimation process is approached using detailed local data. Furthermore, advances gained through these studies represent innovative multi disciplinary collaborations in the field of hazard estimation (Shinozuka et al., 1997). 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The Loss Assessment of Memphis Buildings (LAMB) Project investigated theories surrounding expected seismic performance and associated loss in reinforced concrete and unreinforced masonry buildings. This estimation procedure is implemented in a GIS environment, and incorporates an actual building inventory of reinforced concrete and unreinforced masonry buildings that was obtained from the Shelby County Tax Assessor’s records. Fragility models are presented in the form of fragility curves representing the probability of structural failure based on various damage states at specific levels of ground shaking intensity. These models were then calibrated on both PGA and spectral acceleration. Four qualitative damage states were defined and direct economic loss was estimated using models that had been calibrated to actual data observed for structural damage after the Northridge Earthquake. A total of 24 loss estimations were simulated based on a Mw 7.5 earthquake centered at Marked Tree, Tennessee. (Shinozuka et al., 1997.) An important finding in the LAMB loss estimation methodology is that these results are not always comparable to results based on ATC-13 methodology. The LAMB methodology does not attempt to estimate casualties or injuries. Recalling that justifications for assumptions regarding injuries are weaker in the ATC-13 methodology than justifications for assumptions made in the development of ATC-13 fragility curves, estimates of injuries based on ATC-13 may be equally difficult to validate based on real data. (Shinozuka et al., 1997.) The Memphis Light, Gas, and Water Division (MLGW) Case Study focused on development of a methodology to estimate economic loss due to lifeline system damage 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. from three major utility systems (water, natural gas, and electric power) in Memphis Tennessee. This study is novel because the economic loss estimate is applied to the local level. Again, casualties are not included in this estimate, however the results have implications for casualty estimation. (Shinozuka et al., 1997.) The study showed that the overall economic impacts (including repair costs, revenue loss, and economic impact) of system damage differed substantially between the three utility systems. The study also simulated results that were in agreement with observations after the Northridge earthquake regarding damage resulting in high repair costs in the electric power system. Additionally, it was found that although repair costs were high, indirect damages due to service disruption in some sectors outweigh the direct damage. The service disruption results in a general bottleneck of service delivery and is dependent on the recovery time frame. These indirect losses impact both qualitative and quantitative aspects of injuries. For example, there may be an increase in the expected number of severe casualties due to delayed or incomplete treatment because of electric power disruption. Alternately, there may be more automobile collisions and subsequent injuries if traffic signals are not functional at intersections for extended periods of time. (Shinozuka et al., 1997.) 4.04 Summary o f Limitations o f Current M ethodologies. Many assumptions applied to current models do not consider the demographic characteristics of the population. For example, hazard assessment that has been supported by the Federal Emergency Management Association (FEMA) invokes assumptions regarding entrapment rates (the risk of being trapped in various types of 132 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. structures) based on international data. However, housing and business structures vary widely across the world, and applying international rates to developed countries may result in artificially high entrapment rates, whereas applying the same rates to impoverished countries may result in artificially low entrapment rates. Another limitation o f these assumptions is that they do not consider factors that are known to influence an individual’s ability to leave a structure and individual likelihood for risk-taking activities. For example, very young and very old individuals are less mobile than twenty-year-olds. Additionally, factors such as age and gender have been associated with risk-taking activities that increase the likelihood o f injuries. Consequently, a male, aged 24, may be more likely to walk into the path o f a tumbling bookcase and be trapped or injured than a baby in a crib. Pre-existing medical conditions (e.g., paralysis, compromised vision, and chronic diseases) may also modify behaviors, but such conditions are not applied to hypothetical populations in current models. Other factors that are not considered in current models include shifting structural contents, mechanism of injury, and varying severity of injuries. Estimates of cost of care related to morbidity and mortality from simulated models is subject to a high degree o f variation based (in part) on default assumptions that are not standardized and therefore may be misleading. Benefit-cost models have assumed that the economic statistical value of a life is worth an average of 1.74 million dollars in 1987 and an average of 2.2 million dollars in 1994 (Olson & Eidinger, 1995; Risk Management Software, 1993). The corresponding economic value of a minor injury has been assumed to cost an average of one thousand dollars in 1987 and two hundred dollars in 1995 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Olson & Eidinger, 1995; Risk Management Software, 1993). The fact that this estimate was reduced over the seven-year period is contrary to the inflationary performance of most health care costs and warrants further explanation. Perhaps the definition of a minor injury has changed over time, but standardized terminology and measures would ameliorate this type of confusion. The confidence surrounding estimates of economic loss based on any of these assumptions is unknown and ignored (Olson & Eidinger, 1995). In fact, it is widely acknowledged that the current state of knowledge regarding health-related casualties is not sufficient to establish uncertainty around such estimates (Cheu. 1989; Couison, 1989; Litan, 1993; Noji, 1994b; Olson & Eidinger, 1995; Smith, 1990; Software, 1993; Steinbrugge, 1990). 4-05 Summary of Recent Advances in Earthquake Hazard Estim ation Hypothetical data have inherent limitations. It is thought that injuries due to structural collapse are most difficult to estimate (Risk Management Software, 1993). However, simulated models do not incorporate sufficient details of injuries to provide accurate injury estimates. Numerous assumptions regarding survival are invoked in simulated models so that numbers of injuries and severities of injuries can be generated, but these assumptions have never been tested or compared to actual data (Risk Management Software, 1993). Therefore, conclusions based on models from simulated data may be misleading. This type of modeling is a starting-point for estimating casualties, but only by defining the mechanisms and severities of earthquake-related injuries, and describing how and where people were injured during the quake, can 134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. recommendations be made regarding safer building designs, disaster preparedness, and injury prevention (Noji, 1994b). Indeed, despite reports o f sizable mortality attributed to earthquakes during the last century (1.3 million people)(Smith, 1990), earthquake injury research has noticeably lacked epidemiological expertise (Wagner et al., 1994). Methodology (to date) has relied heavily on data that is collected for reasons other than the study of earthquake-related injuries. Reports o f injuries that have been incorporated into simulation models do not distinguish between various mechanisms of injury, particularly structural causes versus all other (Risk Management Software, 1993). Application of standard epidemiologic techniques to these data may enhance their usefulness. Such techniques include application of a clear case definition, identification of an appropriate comparison group, collection of data for cases and controls by adhering to a standard protocol, coding and grouping data in a uniform manner that is clinically meaningful, and investigating potential risk factors, effect modifications, and confounding factors related to injuries. 135 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C H A P T E R III: P I L O T S T U D Y The 1994 Northridge Earthquake created an environment that introduced a variety of public health concerns. A group of health care providers and researchers from the Los Angeles County Department of Health Services (LAC-DHS) and the University of California Los Angeles (UCLA) proposed collaborative and complementary studies to assess the impact of the earthquake on the population of Los Angeles County. Research was conducted to investigate acute infectious disease outbreaks, mortality attributed to the earthquake, hospitalized injuries, injuries presenting at emergency departments after the earthquake, and population-based surveys to inquire about preparation, household damage, responses, illness, injuries, information and care-seeking behavior. Pertinent details of the pilot study of injured patients that were documented at a sample of emergency departments in Los Angeles County after the earthquake are presented in this chapter. 1.00. PfLOT STUDY BACK GROUND. The Los Angeles County Department of Health Servicesr Injury (LAC-DHS) and Violence Prevention Program (IVPP) was funded by the Centers for Disease Control and Prevention (CDC, contract #CM 67983) to conduct a descriptive epidemiological study r Under the auspices o f the Health Officer of Los Angeles County. Title 17. Chapter 4. Article 1 of the State of California Code of Regulations. 136 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of injuries resulting from the 1994 Northridge earthquake. The objectives of the study were: 1. To determine the availability of data necessary to assess any increase in injuries that might be attributable to the earthquake. 2. To assess any increase in injuries that might be attributable to the earthquake from January 17, 1994 through January 31, 1994 at emergency departments in areas of Los Angeles County that were impacted by the earthquake and compare these injuries with baseline data from January 17-31, 1992 and January 17-31, 1993. 3. To determine the geographic location of patients identified with injuries. 4. To describe demographic characteristics of injured patients for the time periods January 17-31, 1992, 1993 and 1994. 5. To determine the mechanisms or causes of all identified injuries. A summary of protocols for data abstraction and management, definition of terms, definition of data elements, quality control procedures, and results from the pilot study are presented below. 2.00. PILOT STUDY M ETH O DO LO G Y. This study was limited to nine hospitals that agreed to participate and had necessary records and documentation to be included in the study. The study was conducted over a ten-month period (May 15, 1995 to March 15, 1996), and was not 137 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. intended to be comprehensive due to the short time-line of the study and funding restrictions. Ail of the facilities were located in areas of high impact5 and included one Level I Trauma Center, one Level II Trauma Center, four Basic Emergency Departments, and three Health Maintenance Organization Basic Emergency Departments. The general differences between Trauma Centers and Basic Emergency Departments included the requirement that Trauma Centers have a helipad, a minimum annual patient case load of at least 350 traumas, and at least 18 different specialists that are available for service on an ‘On-Call’ basis. Differences between Level I Trauma Centers and Level II Trauma Centers are presented in Table 14 and denote differences in annual trauma caseloads and facility size with respect to the diversity and availability o f ‘On-Call’ specialists. Table 14. Comparison between Levels o f Trauma Care Centers. TRAUMA CENTER CHARACTERISTIC LEVEL I LEVEL II Commitment to treat any patient X X Helipad X X Immediately available emergency medical physician, X X anesthesiologist, and surgeon Annual trauma patient case load 750 350 Number o f ‘On-Call’ specialists____________ 28__________ 18 * Defined by preliminary estimates of ground shaking reported to the State of California Governor’s Office of Emergency Services 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.01. Case Identification through Em ergency Departm ent (EDI Log Data. To estimate increases in injuries due to the Northridge earthquake, baseline data were abstracted from each of the nine participating facilities for the months of January 1993 and January 1992. Data were also abstracted for the entire month of January, 1994. Facility Selection. Documentation was maintained regarding problems that were encountered in requesting and obtaining data from various facilities. Data were requested from 29 facilities, but only nine facilities were willing to participate and able to provide data for this study. Documented problems included communication difficulties with medical record administrators, information management supervisors and emergency department contacts. For example, one administrator lost his office because of the earthquake and had no mode of communication except a pager. One administrator was non-responsive via telephone. Another designated contact was a part-time worker with an irregular schedule. Other problems with facility inclusion were due to indirect complications from the earthquake. For example, one facility lost baseline data due to flooding from sprinklers activated during the earthquake One hospital was so severely damaged that it was still closed from the earthquake at the time of this pilot study. One facility that suffered severe damage was evacuated due to the earthquake. The emergency department at that facility never reopened, and no staff contact for emergency department records was ever established. Some hospitals were not included because o f their distance from areas highly impacted by the earthquake or outside the usual range for patient transfer. Two large hospitals refused requests for emergency department log review until 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the last month of the study, and this did not provide sufficient time to complete the data abstraction One facility was not included due to multiple emergency departments and points of entry into the hospital. Specifically, the pediatric and adult emergency departments were in separate facilities. To complicate matters, the pediatric ED was abruptly relocated when the building was red-tagged after the earthquake. This relocation raised some concern about retrieval o f logs and records. It was also thought that it would be inappropriate to ask staff to perform extra work in light of the relocation. One facility in a high-impact area was willing to cooperate; however a third party managed their data. Obtaining permission to receive downloads of ED logs from this facility was time-consuming and required much follow-up. Eventually, the data were provided, well after the funding period had ended. However numeric fields were packed, and coding schemes were not provided with the data. Therefore additional follow-up was required to import and interpret the data. This particular facility was one that was evacuating units (including the ED) and was more likely to be transferring patients to other facilities than to be admitting patients for care. The ED was actually evacuated to the parking lot, and this arrangement limited access to diagnostic equipment and special services for which the ED was usually equipped. Therefore, this facility may have compromised record keeping and documentation during the immediate aftermath of the earthquake. One facility was heavily impacted and ED staff responded in disaster mode. Specifically, the means o f documentation became the half-page Disaster Tag, and these 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. tags could not be located when requested a year later. The tags were used for approximately 24 hours, and included fields to record the patient’s name, the services provided, and the physician’s diagnosis. However, since these tags were not available, the facility could not be included. This particular facility had no power, the ED was functioning with flashlights because the backup generator had failed, and hospital units were evacuating rather than admitting patients. Patients were transferred elsewhere with their charts, and no tracking system was in place to follow the patients and their records. Additionally, this facility had been ‘downsized’ by County Administrators after the earthquake, and morale at the hospital was quite low. The staff had been reduced in all areas (including management), and it seemed inappropriate to aggressively demand the disaster tags. Therefore, this facility was also not included. Case Definition. Injuries were defined according to standard ICD-9 guidelines, capturing any insult to the body (including poisonings) caused by an environmental event, circumstance (including late effects) or a specific agent (International Classification of Diseases, 9th Revision, Clinical Modification, 1992). Non-specific back pain was considered an injury since many patients present at emergency departments with back injuries and back pain as their only symptom. Non-specific pain for more than one body location was also considered a potential injury because many falls and motor- vehicle collisions result in this type of complaint. Drug and alcohol overdoses or intoxications were considered poisonings, but drug withdrawals were not. Non-specific visits for depression were also categorized as injuries since the potential for self-harm exists and may not have been documented in the brevity of the emergency department 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (ED) log. Myocardial infarctions and non-muscular chest pains were not considered injuries. Follow-up visits for previous injuries and probable or possible injuries were collected, but were not included in the final analysis. An earthquake-related injury was identified by explicit statement as such in the ED log, or if the time of the injury was recorded to be at the approximate time of the earthquake (4:30 am) on January 17, 1994. 2.02. Data Management and Quality Control. Data elements were housed in a relational data management software package (FoxPro®©, Microsoft Corporation). Data were difficult to concatenate due to variation in data management practices between and within facilities. Specifically, not all facilities collected the same data elements. Additionally, variation within facilities was sometimes noted in the data elements collected from year to year. To further complicate the issue, some facilities did not record the date that these changes went into effect. Logical checks for inconsistencies of data were performed to ensure that data were within appropriate ranges. Variables that were checked included date, time, gender, disposition of the patient from the emergency department, and the mode of arrival of the patient to the emergency department. Certain variables (date, billing number, and medical record number) were investigated for missing data, since these would be required to retrieve medical records. Data were also reviewed for duplicate entries of events with identical medical record numbers and dates of service, and identical patient names and dates of service. Discrepancies surrounding all inconsistencies were resolved prior to coding and may have involved a repeat review o f the emergency department log. 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Text-based coding of narrative descriptions of patient complaints and physician’s diagnoses was accomplished using FoxPro®© (Microsoft Corporation). Further coding and global concatenation of files was performed using Statistical Analysis System software (SAS®, Cary, North Carolina). Body region was coded according to the nine Abbreviated Injury Scale (AIS) designations: head, face, neck, thorax, abdomen, spine, upper extremity, lower extremity, and unspecified regions. External cause or mechanism of injury followed International Classification o f Diseases, Revision 9 (ICD-9) designations. These were reduced to fifteen groups: motor vehicle-related injury, fall, firearm-related injury, non-firearm-related assault, overdose/poisoning/drug reaction, force of nature, struck by or caught between object(s), drowning/near drowning/submersion, machine-related injury, fire/burn/electrocution, suicide attempt/self-inflicted injury, injury caused by an animal/amphibian/reptile/insect/plant, cutting or piercing injury, miscellaneous (other) injury, and unspecified cause o f injury. A sample of emergency department log data were validated through medical record review, which was considered the gold standard. Random ten-percent samples of injuries recorded at each of three facility’s emergency department logs were generated. Medical records were then requested and abstracted to validate the data obtained from the ED logs. The instrument that was used for medical record abstraction may be found in Appendix 2. This tool was based on two previously used instruments. One of those instruments was used by the Centers for Disease Control during the investigation of injuries presenting at emergency departments after the 1992 Los Angeles riots. The other instrument was used by the State of Louisiana to monitor disaster-related injuries and 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. deaths following an episode of tornadoes. The current form was developed and pre tested prior to use in this Pilot Study, and is divided into three sections: demographic information, emergency department/hospital visit information, and injury incident details. A validity study has been completed for data obtained at one of the study facilities. Results are presented in Section 3.02. Validity Study Comparing ED Logs aiuiMedical Records at Hospital 2.03. Statistical M ethods. Analyses were performed using Statistical Analysis System Software (SAS®, Cary, North Carolina). Injury data from 1992 and 1993 served as a comparison for injury data recorded following the 1994 earthquake. Similarly, demographic characteristics of injured patients from 1992 and 1993 served as a baseline for comparison with those characteristics of patients injured after the earthquake. Pearson Chi-Square statistics were used to summarize differences between the baseline and post-event injury groups. 3.00 PILOT STUDY RESULTS Descriptive results for nine facilities are presented below. Although the data are presented in aggregate form, caution is warranted regarding interpretation and generalization of these results. It is not known whether results are generalizable to other facilities or are representative of the population in general. The sample is one of convenience because: a) time and staff limitations prohibited inclusion o f all facilities, and b) four facilities that lost data due to the catastrophe were not included in the study. 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.01 Proportion of Em ergency Department Visits for Iniurv Treatment. For the months of January 1992. January 1993, and January 1994, a total of 66,211 visits were reviewed from nine emergency departments in the San Fernando Valley, West Los Angeles, and Santa Monica areas. Table 15 shows the level of emergency care for which each facility was licensed, and the size of the facilities (reflected by the number of licensed beds for each facility). Table 16 details proportions of injury-related visits by facility and in aggregate form. Relative to 1992 and 1993, the frequency of injury visits increased for the 1994 time-period for all facilities in this study except for one managed care facility. That facility (Hospital 'F') showed a decrease in the total number of visits during January 1994 (for unknown reasons). Table 15. Size o f Study Facilities and Licensed Level o f Emergency Medical Care. Facility Code Licensed Number o f Beds Level o f Licensed Emergency M edical Care Hospital ‘A’ 201 Basic Emergency Department Hospital ‘B’ 236 Basic Emergency Department Hospital ‘C’ 268 Basic Emergency Department Hospital ‘D’ 181 Basic Emergency Department Hospital £ E’ 365 Basic Emergency Department Hospital ‘F’ 331 Basic Emergency Department Hospital ‘G’ 212 Basic Emergency Department Hospital ‘H’ 251 Level II Emergency Department Hospital T 1121 Level I Emergency Department 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table IS. Northridge Earthquake Study: Proportions o l Injuries fr o m N ine Emergency Departments. January 1994. 3 a > o c — C — » 3 « - — « • 10 M N t> ‘ «l f l ( « N N N Q f t • 10 M f j n N « « A ^ A A « A N # ANA ^ ^ S NNO \ n o > n - <5 r - <n r> n - <© ■ * - < n o > ^ ^ cm S N < 6 04 5 O - < 0 . . ^ « A # I/) C M C M C M < * > O) N N V N (O ■ A C D N jjj N A A T A N < 0 «- «- m I 8 ^ 9 C M C M I C M t n *“ (D ' C M •“ * - * - O N J ^ m ® CM 0 4 0 - 0 0 r- * - < o s I # A ? ! o § 8 S A N N m A N - 0 ) ^ O 0> I A ( 0 « P > » « A C M C M O 3 1 0 A O 0 > ^ * - ▼ O CM S A O A C D 5 P4 A O A O A A O O O C M «- C M ^ s i? I N- A ! 8 In A o» - o > > w M O — A — > — A — — • AAA A hJ C M < •» ^ tf> N N A A A A Z O C — C — ► 3 w — « A fv O N S O C O 35 ^ si! « < n ^ n . o> •#• ^ C M C M _ in in «r n - cm co m o ^ v r-» « a 5 » • $ a ^ CM CM ‘ ~ A OBJ- J JN A N A * A # K A A A A A A • s i i ^ ^ r4 ^ l o N. N » N > Q i A « « 0 O O I D A CM A A — e — . 3 h- — © a 8 (p Q - O ' * - A A # A A A O I 0 I A C M ^ CD A J N (M O f ?> 35 *“ O *“ A A K A N A N A »* A N N f A A CM A A C M A A A B O A »- A A A •— O # # N A A « 8 8 § § 5 3 SS S 8 S A f f N A A n A « A A A C M # A CO ^ C M CM C M © A A A § m i“ o < CM * — cm m ib a A A n n a # o > m o o f 0 4 *“ • » £ » 2 s s > » « o c — C —* 3 w — « «s o> in m m < co r> i a A o a a o cdcm^ mmo O f ? o in V- CM C M O A f A A N N 7 A A i n a ^ cm cm c o n . c o o> t n * • a a a — C — 3 * - — A A in a n a a a o m m b o n o w n a i O f C O 04 C O in B A N A A O •- *- O J A I n ^ N A A N- «D © * - *- B A N *- r t N - © t n • * « ■ m «- A N r > i n ^ NBA X 5 P * Q A A • A r -* C O 5 r " C O 5 _ C O 5 C O 5> . C O A p p Ol U. O • Ti • *5 • *5 • ”< 5 • *5 • N. % N. f N- s. N- f . N- *“■ as Q. o o o o O X X X X X S -.lt •I i i £ 146 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ^ T a b le 16. (Continued). 3 o > 2 o c — c —. 3 *. — « — c —. 3 ^ SS5S 0 ^ ^ ^ CM CM f > e S o < o 2 rt | - ? a . a 8 § S N ^ M ^ CM CM O C O O co <d co co «- «- o a > W V N ^ r * ^ C O N » a > ii 1 8 5? — C — » 3 w — « « — o — a — > — ■ o — •• * 1 3 w — « * — C — * 3 «- — C *» a> tn fs- C M C M 0 tn m 8 C O 5 * ■ » tn <0 p«- co P -. <* CO ^ N N tN « 3 1 f» - r - ■ |A C M ! S £ 5 s Q J co cO O O P-* 2 « - O O C M 8X 5 , 8 8 CM — O O r - — « £ ? S <0 co <0 a <0 S S - - < 5 £ a j* A r * • * & i g I % s 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In total, 34% of the 66.211 visits in the three time periods were recorded as injury-related. This proportion coincides well with the national estimate that approximately 37% of all emergency department visits are for treatment of injury (Bonnie, Fulco, & Liverman, 1999). The proportion of injury-related visits for the month of January remained fairly constant over the 3-year period. Since the earthquake occurred on January 17, 1994, data were also stratified by time period (January 1-16 and January 17-31). Chi-square statistics were tested for the significance of the association between time period and injury versus non-injury visits. The results are not presented here, however, because the cell frequencies were so large that even minimal changes resulted in significant findings. A small reduction in the proportion of injury-related visits can be seen over the three-year period during the first half of the month (January 1- 16). The decline could be due to random variation, since the observation period is relatively short, or changes in recording or reporting practices. There was an increase in the proportion of injury-related visits for the second half of January, 1994 (40%), compared to corresponding time-periods from 1992 (35%) and 1993 (33%). This could also be due to random variation, changes in recording or reporting practices, or a true increase in injuries due to the earthquake. Demographic and Injury Characteristics of Injured Patients. Patients visiting the emergency departments for injuries after the earthquake were slightly older than patients presenting with injuries at baseline (January 17-31, 1992; January 17-31, 1993). The modal 10-year age group shifted from the 20-29 year age group during the baseline timeframes to the 30-39 year age group after the earthquake. 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 17 presents statistically significant differences in demographic and injury characteristics between injured patients presenting at emergency departments after the earthquake and injured patients from comparable baseline periods. A slight increase in the proportion of injured patients that were female was apparent after the earthquake compared to the baseline timeframe (50% of all injuries January 17-31, 1994, versus 47% of all injuries, January 17-31, 1992, and 48%, January 17-31, 1993). Patients also exhibited proportionately more injuries to lower extremities compared with all other body regions after the earthquake compared to baseline data (31% of all injuries January 17-31, 1994. versus 22% of all injuries January 17-31, 1992 and 1993). Injuries to upper extremities compared with all others showed a statistically significant proportionate decrease after the earthquake (26% o f all injuries January 17-31, 1994, versus 29% of ali injuries January 17-31, 1992, and 27% January 17-31, 1993). Mechanism of injury was obtained from about 35% of the injury data. The large fraction of missing data might be due to the intentional brevity of the ED logs or due to the difficulties maintaining documentation in emergency situations. More patients were injured due to objects striking them and cutting or piercing injuries compared to all other known mechanisms after the earthquake than during baseline. Objects that struck a person caused 4% of all injuries January 17-31, 1994, compared with 2% of all injuries January 17-31, 1992 and 1993. Objects that cut or pierced a person caused 9% of all injuries January 17-31, 1994, compared with 4% of all injuries January 17-31, 1992 and 1993. This is expected in an earthquake since objects that are not anchored to stable frames will become projectiles. Additionally, glass and ceramic (windows, mirrors, 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 17. Comparison o f Injury and Demographic Characteristics o f Patients Presenting to Nine Emergency Departments after the 1994 Northridge Earthquake and at Baseline. 1992 1/17-1/31 1993 1/17-1/31 1994 1/17-1/31 Pearson X21 n < % ) n < % ) n < % ) (p Value2) Gender Male Female 1853 1651 (53.4) (46.6) 1777 1639 (52.0) (48.0) 2551 2506 (50.4) I (49.6) J 7.5 (< 0.05) Body Location Lower Extremity All Other Body Parts 764 2663 (22.3) (77.7) 756 2637 (22.3) (77.7) 1550 3518 (30.6) - 1 (69.4) ; 104.5 (< 0.001) Upper Extremity All Other Body Parts 975 2452 (28.5) (71.5) 906 2487 (26.7) (73.3) 1312 3756 (25.9) 'I (74.1) I 6.9 (< 0.05) Mechanism of ‘ ’ v Plant / Animal / Reptile / Amphibian / Insect 73 (5.8) 59 (4.3) 113 (7.2) \ 11.3 All Other Mechanisms 1168 (94.1) 1315 (95.7) 1453 (92.8) J (< 0.005) Cut By 48 (3.9) 48 (3.5) 139 (8.9) 1 50.2 All Other Mechanisms 1193 (96.1) 1326 (96.5) 1427 (91.1) J (< 0.001) Struck By or Against 29 (2.3) 23 (17) 63 (4.0) 'I 16.2 All Other Mechanisms 1212 (97.7) 1351 (98.3) 1503 (96.0) J (< 0.001) 1 Pearson Chi-Square on 2 Degrees of Freedom 2 Statistically significant differences in demographic & ’ ; , characteristics between patients presenting after the earthquake and those from comparable baseline time periods.________________________________________________________________ O 0 8 dishes, beverage vessels, etc.) are fragile and break, leaving sharp edges that can puncture the skin. People wakened from sleep during the Northridge earthquake may not have donned slippers or shoes prior to walking through areas covered with shards of glass. A higher proportion of patients appeared at emergency departments for injuries or poisonings resulting from an animal, insect, or plant compared to all other known mechanisms after the earthquake than during baseline. These types o f injuries caused 7% of all injuries January 17-31, 1994, compared to 6% January 17-31, 1992, and 4% January 17-31, 1993. This might be due to problems associated with pets being frightened and running away from home, being approached by strangers, and subsequently biting the stranger. Additionally, owners may have put themselves in harm’s way by trying to control their own frightened pets. Individuals that were displaced from their homes and living outside may also have been at increased risk for exposure to injuries or poisonings from insects, plants and animals. 3.02 Validation Study Comparing ED Logs and Medical Records at Hospital ‘P*. Of 642 injured patients recorded in the ED log at Hospital ‘D ’ between 1/17 and 1/31 of 1992, 1993, and 1994, 70 randomly selected entries were compared to data abstracted from corresponding medical records. 68 of the 70 patient records were retrieved. Fourteen variables were compared between the two data sources. Discrepancies were classified as major, moderate and minor. Variables that were omitted in both sources (ED log and the medical record) were also tallied. Entries that 151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. differed greatly between sources were considered major (i.e., the name, gender, medical record number, mode o f arrival, date of service or location of injury on the body of the patient were not the same from both sources). Discrepancies were categorized as moderate if data were specified in one source but not specified in another source (i.e., the payment type was unspecified in the ED log, but identified as workman’s compensation in the medical record). Discrepancies between left and right appendages were also considered moderate. If the description of the injury did not match the body location, the discrepancy was also considered moderate. For example, if the injury was described in the ED log as a concussion from being hit in the head, but the medical record showed a diagnosis of wrist fracture due to a fall with an associated head abrasion, but no mention of concussion, this type o f discrepancy would be categorized as moderate. Minor discrepancies included spelling errors or transpositions. Minor errors might cause problems filing or retrieving patient medical histories, or might reflect inaccurate demographic characteristics (age, gender, ethnicity). Validity study results for Hospital ‘D’ are presented in Table 18. No major discrepancies were found regarding the patient complaint, primary diagnosis, or external cause of injury. Major discrepancies were low, but they were found in the medical record identification number nine times (13% of the records) and in the patient’s name once (1% of the records). These fields are often used to identify patients and to retrieve medical records; therefore they are important fields to maintain accurately. Major discrepancies in the gender of the patient could be resolved for two patients by assuming common names of patients as either male or female, but two other major 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 18. Validity Study for Hospital P*: Agreement between Medical Record* and Emergency Department Logs. Variable Major2 Discrepancies1 Moderate* Minor* Omission* M edical R ecord No. 9 0 0 2 P a tie n t S u rn am e 0 0 2 3 0 P a tie n t First N am e 1 0 7 0 G e n d e r 4 0 0 0 A g e 0 0 0 1 Prim ary P ay er 3 2 0 0 D ate of ED Arrival 1 0 0 1 T im e of ED Arrival 0 0 0 1 M o d e of ED Arrival 1 5 0 1 D isposition of Patient from ED 2 9 0 2 M ajor P atien t C om plaint 0 0 4 0 Prim ary D iagnosis 0 1 8 1 L ocation of Injury on Body 2 3 0 5 E xternal C a u se of Injury 0 19 0 8 T o ta ls 23 39 42 22 O verall D iscrepancy P e rc e n ta g e 5 2 % 4% 4% 2 % Frequencies of total entries checked (N»68). • Name, gender, medical record number, mode of arrival, or location of injury on body did not agree. or service date preceded incident date. 3 Unspecified in one source, s pecified in the other, or leftTnght disagreement in appendages. or injury description did not match body location. * Typographical or spelling discrepencies. 5 Omitted from both the EO log and the medical record. ° Based on 9S2 fields (66 records. 14 variables eech) 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. discrepancies on gender could not be resolved in this manner. The disposition o f the patient from the emergency department was different for two patients. This is of interest since studies may use emergency department logs to determine frequencies o f patients admitted with various complaints, particularly since considerable time delays are associated with available hospital discharge data. Major discrepancies were also noted in the primary payer. This information is important for reimbursement, especially in times of disaster If the payer can be identified through ED logs, this source can be used to complete forms for reimbursement for services rendered in times of disaster. Major discrepancies were also detected in the body location of the injury for two patients. These discrepancies are important since injury severity scoring is based on body location. Additionally, future medical treatment may be based on patient histories and record keeping from previous visits to the emergency department; therefore care should be taken to record the body location accurately. 4.00. DISCUSSION OF PILOT STUDY Findings of this pilot study apply to the general methodology used in disaster epidemiology. Rates are not presented because of difficulties identifying the population at risk for earthquake-related injuries. This issue will be discussed in the following paragraphs. Various estimates of injuries that were generated through different methodological approaches will also be discussed. In addition, implications based on the validity study between emergency department logs and medical records will be discussed. 154 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.01. Problems Associated with Calculation of Rates. Rates are not calculated because the population at risk for earthquake-related injuries (i.e., the denominator) cannot be identified, and ascertainment of all injuries (i.e., the numerator) is also not possible. The denominator might be defined by including the entire population in any seismically active geographic location. However, population information is available by political boundary, and those boundaries do not correspond to boundaries of estimated seismic activity. If injuries were collected for all emergency departments and housed in a global registry (similar to a cancer or trauma registry), injury incidence rates and cause-specific injury rates would be calculable for populations based on political boundaries. No such registry exists in Los Angeles County, and there are no estimates of injury morbidity available. Rates may be calculated for each facility based on designated catchment areas (i.e., 5-mile radius around a facility). However, this may be misleading since patients appearing at specific facilities during the earthquake may not have resided in the catchment area and might not have presented there under usual circumstances. Patient loads at individual facilities during and after the earthquake were highly dependent on facility functionality. Although it is known which facilities were functional and which facilities were evacuated (and at what levels and times the evacuations occurred), usual facility utilization by individual patients is not available. Specifically, it is not known which patients presented for treatment at facilities that they would not have visited under normal conditions, and which patients utilized their usual facility for emergency care. Therefore rates based on designated catchment areas may be artificially high or low. 155 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.02. Comparison o f Study Results Based on Different M ethodologies. Results of rapid assessments performed within the County of Los Angeles’ Department of Health Services (LAC-DHS) immediately after the earthquake were quite different than results obtained in this study (see Table 19). One rapid assessment conducted by the LAC-DHS Emergency Medical Services was performed during the first 72 hours after the earthquake (a ‘real-time’ assessment) and again two-months after the earthquake (a retrospective assessment). The real-time assessment was administered via telephone, and a copy of this form is found in Appendix 2. The form shows that tallies were taken for patients treated and released (‘T/R’) on January 21 and 22, 1994, as well as tallies for patients admitted (‘A’). Note that there are few hospitals with any entries on these particular days, and this was the trend throughout the paper-work associated with this survey. It is unclear whether all the listed facilities were telephoned and had no patients for earthquake-related injuries, if the facilities refused to respond, or if the facilities were simply not contacted. Limitations with this methodology are fundamental since not all facilities had this mode of communication available during the first 24 hours after the earthquake. Furthermore, individual responses to the survey were subject to high variability, since the response could have been an actual tally, a number that had been committed to memory, or a guess regarding how many patients had been seen with earthquake-related injuries. The two-month retrospective assessment was administered via fax. A copy of the form that was used in this survey is also included in Appendix 2. The questionnaire is not 156 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 19. Example of Discrepancies in Reported Injuries due to Different Study Methods. Facility Numbers of Reported Injuries: January 17 & 18,1994 Real Time Survey Response1 Retrospective Survey Response2 _____________Pilot Study3 Hospital #1 10 27 Not Included Hospital #2 329 197 152 Hospital #3 No Response 95 Not Included Hospital #4 160 No Response Not Included Hospital #5 No Response 325 Too Late for Inclusion Hospital #6 No Response 368 227 Hospital #7 180 No Response 109 Hospital #8 No Response 465 232 Hospital #9 304 No Response 43 Hospital #10 No Response No Response 71 R e a l T i m e S u r v e y A d m i n i s t e r e d b y T e l e p h o n e , L o s A n g e l e s C o u n t y E m e r g e n c y M e d i c a l S y s t e m 2 R e t r o s p e c t i v e S u r v e y A d m i n i s t e r e d b y F a x , L o s A n g e l e s C o u n t y E m e r g e n c y M e d i c a l S y s t e m R e t r o s p e c t i v e S u r v e i l l a n c e C o n d u c t e d b y L o s A n g e l e s C o u n t y I n j u r y a n d V i o l e n c e P r e v e n t i o n P r o g r a m E p i d e m i o l o g i s t s - J specific regarding who should complete the response and how the cases should be identified. Furthermore earthquake-related injuries and illnesses are counted together. Varying responses were (again) observed for this methodology compared to the real-time assessment Neither the real-time assessment nor the 2-month retrospective survey resulted in findings that correspond well with the results obtained in the pilot study. The LAC-DHS Acute Communicable Disease (ACD) unit conducted a separate rapid assessment during the first week of February 1994. two weeks after the earthquake. Epidemiology staff from this unit visited ten facilities. Emergency department logs were reviewed and injuries that were treated on the day of the earthquake were tallied. An example of the disparity in injury counts between the ACD survey and this pilot study are presented in Table 20. The total number o f injury visits tallied at that facility for the day of the earthquake on February 1, 1994 (n=6) was dramatically lower than results obtained in the current study on July 1, 1996 (n=167). The most likely explanation for this huge discrepancy is probably due to the delay in updating and maintaining records during and after a disaster. It is likely that the usual documentation at emergency departments is not in place after a disaster. In normal circumstances, many emergency departments re create the log of daily activity retrospectively (i.e., the next day). This results in a minimum lag-time of one day for documenting activity under usual circumstances. However, during a catastrophe, this process of retrospective documentation is laborious and is a low priority for emergency department function. In fact, these results suggest that documentation may not be in place two weeks after the disaster, and that reports of 158 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. emergency department activity immediately after an earthquake grossly underestimate actual activity. Table 20. Example of Discrepancies in Reported Injuries due to Date of Data Review: Reported Injuries at Hospital #6 on January 17. 1994 Injury Type Date of Data Review February 1,1994 July 1. 1996 Burns 1 2 Cuts and Lacerations 2 127 Head and Trauma Injuries 1 4 Orthopoedic Injuries 1 22 Other Injuries 1 12 Totals 6 167 Another possible explanation for the undercount could be due to differences in case definition. For example, terms such as ‘trauma’, and ‘other injuries’ need fairly specific inclusion criteria from an epidemiologic point of view. Additionally, the injuries noted in Table 20 are grouped according to the ACD survey. These groups include injury mechanisms (bum, cuts, lacerations), body location of injury (head), and nature or type of injury (trauma, orthopoedic) as if they were of the same category. Injury epidemiologists usually assess these categories separately. Furthermore, injuries are better described mechanistically. For example, since many injuries may be orthopoedic in nature, the external cause of injury is a more useful epidemiologic category than the nature of the injury. The differentiation may seem subtle, but with respect to prevention and public 159 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. health practice, the difference is important. Without mechanism of injury, prevention programs and interventions cannot be focused. This also articulates a basic difference behind epidemiologic areas of expertise. Since epidemiologists with experience primarily in acute infectious diseases completed this survey, the presentation of injuries was not in the usual injury epidemiology format. Grouping injuries in such a manner may lead experts to be concerned about the basic methods and definitions behind case selection and identification. The frequencies reported in the LAC-DHS Emergency Medical Services’ survey were subject to the same type of misclassification since no clear case definition for injury was established. Still another (although unlikely) explanation for the disparity in counts of injuries could be due to the source of the data. For example, the patient may complain of a certain problem, but be diagnosed with a different problem. The epidemiologic standard is the physician’s diagnosis, and in some cases (particularly when calculating Injury Severity Score) verification from other sources is required (radiology reports for fractures, witnesses to verify length o f time for loss of consciousness, or Computed Tomography (CT) scans to assess head injuries characterized by loss of consciousness). Since the standard case definition in acute infectious diseases is also based on the physician’s diagnosis (and may require laboratory confirmation), these standards were likely followed for selection of injury cases. In short, discrepancies in reported injuries may be due to methodological issues such as vague or imprecise case definitions, inconsistency in protocols, and variations in sources of data. Additional complications occur when data from different facilities are 160 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. merged. This is because the availability of data varies widely between facilities and sometimes within facilities. Data may also vary within a facility across years, between facilities by type of personnel responsible for maintaining, coding and reporting emergency department activity, and by the level of training and amount of supervision those personnel receive. 4.03. Discussion of the Validity Studv between Emergency Department Logs and Medical Records. The overall proportion of observed discrepancies is low for this facility. Out of 952 fields that were compared (68 records were compared on 14 variables), only 12% showed discrepancies or omissions. This leads one to believe that the information that was available through this facility's emergency department logs was consistent with information documented in the medical record. It was the policy of the facility involved in this validity study not to record the time of exit from the ED. Therefore, time spent in the ED could not be evaluated. Comparisons of the length of time that is spent in the emergency department and disposition o f the patient may show differences in a disaster when compared with usual activity. This type of information is useful for policy and planning purposes because staffing recommendations may need to be modified, and identification of well-ventilated emergency backup waiting space for patients might be prudent. In order to plan for changes in case-load distributions and processing of patients after disasters, it is important to review patterns of patient-flow by time. 161 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.00. LIMITATIONS TO THE PILOT STUDY A limitation to this pilot study is the sampling methodology. Due to the short time frame over which data collection and analysis occurred, included facilities were those that had data readily available and agreed to participate in a timely manner. Although other facilities subsequently agreed to participate, the process took too long, and the study had ended by the time they agreed. The sample is basically one of convenience. Another major study limitation stems from non-documented injuries. Documentation was not established at most facilities until at least 6-8 hours after the earthquake. It is possible that the injuries presenting immediately after the event may have been different (more or less severe) than those that presented later that day. Severe injuries may have appeared earlier due to rapid response from search and rescue teams or family members. If patients were subsequently admitted to a facility, a retrospective record of entry through the ED would have been created. However, patients that were stabilized and transferred to other facilities may not have been documented at all. Hospitals that were in high-impact areas were more likely to discharge or transfer patients than to admit patients in the midst o f equipment failures and structural compromises. Since documentation varied by time period relative to the earthquake and by facility, the reported comparisons of proportions o f injury visits is likely biased. A large proportion o f injuries recorded in the emergency department logs did not have any information regarding the external cause of injury. This is worrisome and limits 162 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the usefulness of emergency department logs for surveillance of injuries. It is possible that the mechanism and severity of injury were associated with reporting or non-reporting of this information. For example, minor cuts or scratches might not have been reported. It is also likely that this selective non-reporting varied by time-period. For example, after the earthquake, personnel were probably less likely to document mechanisms associated with minor injuries compared to serious injuries. This type of systematic error leads to biased study results (overestimates or underestimates o f relative risk associated with specific mechanisms of injuries). Since mechanism was rarely recorded in emergency department logs, it was subsequently difficult to attribute injuries to the earthquake. Therefore, it is probable that earthquake-related injuries were misclassified as non-earthquake-related. This reduces the sensitivity (categorization of individuals who are truly cases) of the study. 6.00. CONCLUSIONS BASED ON THE PILOT STUDY. Results of this pilot study suggest that it is difficult to accurately attribute an injury to the Northridge Earthquake since that information was rarely recorded in the emergency department logs. Indeed, most of the data had no information regarding mechanism of injury. This limits the usefulness of emergency department logs regarding rapid assessments that are the basis for disaster response recommendations made to health officials. Administrators cannot appropriately assess the impact of a catastrophe based on these data, and a community cannot respond appropriately if information is not available or is inaccurate. 163 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. It was also found that much of the emergency department log data did not contain address information and, therefore, could not be linked to structures. _To obtain more complete information about the injury scene and circumstances of the injury, a medical record review was necessary. Since this pilot study was not intended to include medical record review, the data limitation was acknowledged and additional funding was subsequently justified to pursue medical record review. That grant proposal was funded through the Department of the Interior, United States Geological Survey, and is included in Appendix I. 164 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C H A P T E R IV: S T U D Y M E T H O D O L O G Y In this chapter, the methodology used to study the associations between injury characteristics, seismic activity, and the built environment for injured patients that were treated at a sample of emergency departments after the Northridge Earthquake is described. The intended study design will be presented, followed by the actual study design and methodology. 1.00 INTENDED STUDY DESIGN. The intended design of this study was to obtain a cross-sectional sample of injured patients that presented at emergency departments during the month of January, 1994. Specific hypotheses to be tested were: a. The number of injuries reported in emergency departments after the Northridge earthquake that occurred in residences that were damaged (as assessed by either a red or yellow tag) is statistically significantly different than those injuries reported in emergency departments that occurred in residences that were not damaged or inspected (as assessed by a green tag or no tag). b. Reported variation in type and severity of injuries detected in emergency departments after the Northridge earthquake can be explained in part by increased 165 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. risk associated with specific structure types, seismic activity at the injury location, and demographic characteristics (i.e., age, gender ethnicity) of the patient. Emergency departments that were eligible for inclusion in this study were those that participated in the Pilot Study. The study proposal is available for reference in Appendix I. Progress reports have also been submitted to the funding agency and are available at the USGS external research programs web-site (www.usgs.gov). The original proposal goal was to obtain medical record data from 12 facilities. The requested funding was considerably diminished and resulted in necessary reductions in the labor-intensive goals related to medical record review. Delays in approval and acceptance of the grant award by the Los Angeles County Board of Supervisors resulted in a late start for the study and less time to complete the goals. The Department of Health Services was ‘re-engineered’, and the staff of this grant were required to vacate their office space and work remotely for the final six-months. This disruption created additional problems primarily with access to equipment. Consequently, four facilities were approached for medical record review and data abstraction was completed for all four facilities. The data entry and analysis were completed for the entire month of January, 1994, for three facilities. However, the data entry and analysis were completed only for the time-period following the earthquake (January 17-31, 1994) for one facility (Hospital ‘D’) due to exhaustion o f funding. 166 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.00 FACILITY SELECTIO N Facility selection was a difficult aspect of this research. Facilities were included in this study based on: 1) Participation in the pilot study. 2) Facility size. 3) Level of care provided at the emergency department. 4) Level of functionality after the Northridge earthquake. The first criterion (participation in the pilot study) was necessary since emergency department logs were used to identify cases. Even if the facility submitted data after the pilot study had technically ended, it was eligible for inclusion in this study. The second criterion (facility size) was used because during the pilot study we found that very small hospitals had unusual characteristics and may represent only a select segment of the general population. For example, one facility with 88 beds did not have a maternal or child health practitioner on staff. Therefore, most of the patients seen at that emergency department were over the age of 40, and more likely to be male. An arbitrary cut-off point of 100 beds was used, and only facilities that were licensed for at least 100 beds were considered for inclusion in the study. The third criterion (level of care provided at the emergency department) was used to ensure that a range of injuries would be sampled. If the study facilities included only Basic Emergency Departments, few (if any) traumas or serious injuries would be seen. Similarly, if only Trauma Centers were included in this study, minor injuries may be 167 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. under-represented. Since only a sample of emergency departments could be included, facilities were chosen to sample all levels of care (Level I Trauma Center, Level II Trauma Center, and Basic Emergency Department1 ) were represented. The final criterion (level o f functionality after the Northridge earthquake) was used because facilities or emergency departments that were closed or evacuated were believed to have diverted patients to other facilities. Diversions were not documented, but practitioners made this point during interviews conducted in the pilot study. Therefore, the final selection included those facilities that were fully functional and were thought to have received patients diverted from facilities that were evacuated or non functional. Table 21 lists hospitals that were included in the pilot and current study, as well as those that were in zip codes o f average MMI=VTII.U All levels o f care (Basic Emergency Department, Level I Trauma Center, Level II Trauma Center) were represented in this study. Two Level I Trauma Centers were selected because: a) they met the previously stated criteria; b) the patient volume at those facilities is typically much higher than at other facilities; and c) they were more likely to receive patients diverted through emergency medical transportation. The basic emergency department was chosen because it met the previously stated criteria and because of its catchment area; it provides service to patients from the South Bay, Inglewood, and South-Central areas. These areas were densely populated and reported to have been impacted by the earthquake. ' See Chapter 3, Section 2.00 for definitions of these levels of care. u No facilities were located in MMI higher than VIII. 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 21. Hospitals In Los Angeles County Included In Pilot Study of Earthquake-Related Injuries and Hospitals in Average MMI1 Areas of VIII due to the 1994 Northridge Earthquake. Facility Number Ave M M I1 Ave PGA2 # Beds EDJ Analyzed in Pilot Study USGS4 Study Notes Hospital #1 Unavailable Unavailable 345 Yes No No Received data from ED Log after preliminary study ended Hospital #2 V II 0.42 236 Yes Yes No Hospital #3 V II 0.42 331 Yes Yes No No increase In visits post earthquake Hospital #4 V II 0.42 647 Yes No No Hostile, denied access Hospital #5 V II 0.48 Unavailable Yes No No Lost data for 1/17/94 Hospital #6 V II 0.30 1121 Yes Yes Yes Hospital #7 V II 0.42 268 Yes Yes Yes Hospital #8 V III 0.24 196 No No No No ED Hospital #9 V III 0.41 711 Yes No Yes Received data from ED log after pilot study ended. Hospital #10 V III 0.42 61 No No No No ED Hospital #11 V III 0.42 156 Yes No No Attempted to obtain data, unsuccessful. Hospital #12 V III 0.42 181 Yes Yes No Small - fewer than 181 beds being used. Hospital #13 V III 0.42 182 Yes No No Not able to contact medical record administrator Hospital #14 V III 0.42 212 Yes No No Obtained data from nearby facility. Hospital #15 V III 0.42 212 Yes Yes No No increase in visits post earthquake Hospital #16 V III 0.42 213 Yes No No Attempted to obtain data from the parent facility. Hospital #17 V III 0.42 362 Yes No No Attempted to obtain data, unsuccessful. Hospital #18 V III 0.48 251 Yes Yes Yes Lost 25% EQ-related medical records Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 21. IContinued). Facility Number Ave M M I1 Ave PGA2 # Beds ED3 Analyzed in Pilot Study USGS4 Study Notes Hospital #19 V III 0.48 365 Yes Yes No Actual number of beds being used much fewer than 365. Hospital #20 V III 0.48 501 Yes No No Under construction - closed for preliminary study Hospital #21 V III 0.63 537 Yes No No ED Closed, never reopened, no contact person Hospital #22 V III 0.73 201 Yes Yes No Hostile Hospital #23 V III 0.75 14 No No No No ED Hospital #24 V III 0.75 69 Yes No No Fewer than 100 beds Hospital #25 V III 0.75 259 Yes No No Lost baseline data ' Modified Mercalli Intensity 2 Peak Ground Acceleration 3 Emergency Department 4 United States Geological Survey - j o 2.01. Facilities in Areas o f 4High Impact*. Another difficulty in facility selection pertains to the assessment of ‘heavily impacted’ areas. Furthermore, the issue of where the patients were when they were injured and where they most likely sought treatment needed to be considered in conjunction with facility selection. There were 90 emergency departments in Los Angeles County in facilities with at least 100 beds at the time of the earthquake. MMI levels corresponding to zip codes in Los Angeles County were provided by EQE Center for Advanced Planning and Research.' Based on average MMI levels for census tracts where these 90 hospitals were located, 40% were in areas of average MMI < VI, 14% were in areas of average MMI=VI, 31% were in areas of average MMI=VII, and 14% were in areas of average MMI=VTII. No facilities were located in zip codes with average MMI of IX. Misclassification is inherent in this summarization, since zip codes may cover large geographic areas and the degree of ground-shaking is not uniform throughout each zip code. One facility that was close to the epicenter was in a zip code for which MMI and PGA estimates were unavailable. Since no facilities were located in MMI=IX. MMI=VIII represents the area with the highest estimated ground-shaking intensity. For comparative purposes, the average PGA for each zip code is presented in Table 21 along with the MMI. The range of PGA associated with MMI=VTII is quite wide: 0.24 - 0.75 g. This range also covers the range of PGA associated with MMI=VTI (0.30-0.42). Although MMI is the scale that is usually used to assess damage and hazard after an earthquake, it is clear that high MMI may ' This is the most appropriate boundary for MMI since it is reported by zip code. 171 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. occur in areas of lower ground motion. This is additional justification for sampling in areas other than the highest MMI. Notes regarding problems encountered at various facilities are also presented in Table 21. Five of nine facilities in areas of MML=VTII were ineligible for this study because of the following problems noted during the pilot study: a) inability to contact the medical record director, b) loss of data at two facilities, c) closure at two facilities. The other four facilities were approached, but no personal contact had been established. Two of the four facilities that were included in this study were in areas of MMI=VTI. One of these was a Level I Trauma Center and the other was a licensed Basic Emergency Department. The two facilities in areas of MMT=VTII included a Level I and Level II Trauma Center. Three of the facilities (a Level I and Level H Trauma Center as well as the Basic Emergency Department) were in areas of PGA=0.41-0.48, the other Level I Trauma Center was in PGA=0.30. 2.02. Summary of Facility Selection. Although activity and injuries seen at this sample of emergency departments may not represent the general population at risk, the sample is based on emergency departments that were not evacuated and were able to treat patients injured in the earthquake. Additionally, all levels and types o f care available at emergency departments are included in this sample. It is also thought that a relatively broad sample of the population was obtained, since the associated injury scenes ranged from the Santa Clarita Valley to South-West Los Angeles and South Central Los Angeles. 172 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.00. CASE IDENTIFICATION. This study is based on injured patients that were detected through emergency department logs. Once potential cases were identified, corresponding medical records were requested and information was abstracted onto a standard form. Data were then entered into a computer. An estimate o f the error due to data entry transcription was also included in this study. The following paragraphs detail the methods behind these processes. 3.01. Data Abstraction. Medical records were requested at participating facilities for all injuries or suspected injuries that were identified through review of emergency department logs for January 1-31, 1994. Data were abstracted and recorded on the form used in the pilot study (see Appendix 2) with the addition o f a field to record a description o f the circumstances of injury. Data were entered into a relational database using FoxPro©® (Microsoft Corporation). FoxPro©® was chosen since it allows identification of embedded character strings, which was an important feature in automating the coding process. The availability of requested medical records is summarized by hospital and time period (pre-earthquake versus post-earthquake) in Table 22. Baseline data were not entered into a database from Hospital ‘A’ (due to funding exhaustion), therefore a comparison cannot be made for this facility between the proportion of data retrieved that originated prior to the earthquake versus after the earthquake. Two facilities were relatively consistent regarding the proportion of records retrieved before and after the 173 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 22. Comparison o f Numbers o f Medical Records Requested and Retrieved at 3 Facilities In Los Angeles County. January. 1994. Pre-Earthquake1 Post-Earthquake' > Facility Number Requested Number Retrieved % Retrieved Number Requested Number Retrieved % Retrieved Hospital ’ A' naJ naJ naJ 644 628 74% Hospital 'B ' 403 341 84% 707 521 74% Hospital 'C ' 1022 861 84% 1055 895 85% Hospital 'D ' 579 448 77% 671 530 79% 1 January 1-16,1984 2 January 17-31,1994 3 not available due to exhaustion of funding -4 earthquake. This proportion was around 85% at one facility and 80% at another. The proportions may seem low. but under the circumstances (4-year time lapse and the number of records requested) they may be acceptable. However, one facility that was located in an area of more intense ground-shaking was able to locate a smaller proportion of records after the earthquake (74%) compared to before the earthquake (84%). There were problems retrieving records that had been stored at the warehouse for this facility, particularly for the first few days immediately after the earthquake when (presumably) most patients would have arrived. Many requested records were reported as lost at that facility after the earthquake. 3.02. Quality Control. Although some restrictions were placed on data entry fields, logical checks for inconsistencies of data and range checks were also performed. Variables that were checked included the date of abstraction, gender, age, date of birth, ethnicity, payment type, date of arrival, time of arrival, time of exit, mode of arrival, disposition from the emergency department, incident date, incident address, use of the property where the incident occurred (i.e., residence, public building, retail store, street, freeway) special circumstances of the incident (work-related, earthquake-related, entrapment or extrication-related, sports-related), body location of injury, intention o f injury (whether the act was a deliberate attempt to injure or if the injury was an unintentional consequence of an act), and external cause of injury (i.e., struck by an object, falls, cut by an object, caught in or between object(s), firearm injury, etc.). Discrepancies 175 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. surrounding all inconsistencies were resolved prior to analysis, and may have involved a second review of the medical record. Data were excluded due to missing or invalid address information, visits not relating to initial injury incidents (follow-up, duplicates, non-injuries), or invalid data (incident date out of range). Examples of invalid addresses included post office boxes, rural routes, addresses that were not within Los Angeles County, or incidents that lacked complete addresses. If the injury occurred prior to January 1, 1994, or if the injury occurred after the date of arrival at the emergency department, the observation was deleted. Additionally, if the emergency department visit was a follow-up visit, a surgical complication, or a non-injury, it was not included in the analysis. Table 23 shows injury and address eligibility and exclusion criteria for patients based on the data abstracted from the medical records. Of 4,190 records from the four facilities, 92% contained valid information. Approximately 83% o f the records represented injuries that were reported to have occurred at the patient’s residence. Seven percent of the observations represented injuries that occurred at locations other than the patient’s residence. Two percent o f the injury incidents occurred on sidewalks, streets, highways, or freeways. Eight percent o f the records were excluded due to missing or invalid address information, or if the incident date was out of range. Records with valid address information were further reviewed to confirm that those with mechanisms of injury classified as ‘Other’ or ‘Unknown’ were truly associated with injuries. For example, a diagnosis o f ‘Cardiac Arrhythmia’ associated with a 176 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 23. Eligibility and Exclusion Criteria for Injury Incidents Treated at 4 Emergency Departments. Los Angeles County. January 17-31.1994. Categories for Eligible Injury Incident Scenes Frequency Incident Address Differed from Residence 288 Incident Address was Same as Residence 3488 No Address for Incident, Census Tract of Injury Scene Used as Surrogate1 74 Sub-Total 3850 Categories for Ineligible Incidents No Incident or Demograpic Information Available in Medical Record 134 Unusable Incident or Residence Address2 151 Duplicates, Follow-up Visits 19 Follow-up Visits to Different Facilities 8 Incident Date out of Range 9 Non-Injuries 19 Sub-Total 340 Total Eligible 4190 Non-Specific Complaints Deleted Joint Pain, Muscle Pain, Open Wound (not further specified) 254 Total Eligible Patients with Identifiable Mechanisms o f Injury 3595 ^ Incident occurred on a street, sidewalk, highway, or freeway. 2 Post office box, rural route, address outside Los Angeles County. complaint of ‘Chest Pain’ might have been classified erroneously as an injury (‘Other’). Since this type of diagnosis is not associated with an injury, it would have been deleted. An additional 254 observations were subsequently deleted due to non-specific complaints that were not associated with a mechanism of injury (i.e., non-specific joint or muscle pain; open wound, not further specified), leaving 3,596 valid observations. Frequencies of valid injuries treated at each of the four facilities are presented by day in Table 24. All of the facilities show a marked increase in frequency o f injury visits on the day o f the earthquake. January 17, 1994, compared to the rest o f the month. Additionally, there is an increase in frequency of injury visits after the earthquake compared with baseline data for each facility for which baseline data was collected. Certain variables (internal identification number, incident date, medical record number, and patient name) were checked for duplicate observations. Eight patients appeared at more than one facility for the same injury incident and were only considered once in the analysis. The earliest visit was retained for analysis; subsequent visits were considered follow-up visits and were not analyzed. If patients presented at different facilities for different injury incidents, each of these incidents was counted in the analysis. Only two patients were identified as sustaining more than one unique injury for the study time period. After deleting records based on injury or address exclusion criteria, 3,850 records remained in the database (see Table 23). A random 10% sample of abstraction forms was compared to the data that were entered in the database to estimate a transcription error 178 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 24. F requencies o f Valid Injuries Treated a t 4 Facilities in L os A n g ele s Countv. January 1994. DATE__________Hospital 'A' Hospital V H ospital 'C Hospital 'D' 1/1/94 35 37 23 1/2/94 22 42 14 1/3/94 13 49 11 1/4/94 14 47 17 1/5/94 15 41 12 1/6/94 21 41 11 1/7/94 21 37 24 1/8/94 22 47 24 1/9/94 19 31 25 1/10/94 16 46 21 1/11/94 17 53 22 1/12/94 13 44 25 1/13/94 11 32 22 1/14/94 26 44 29 1/15/94 31 38 22 1/16/94 23 46 27 1/17/94 124 207 76 96 1/18/94 42 38 69 32 1/19/94 30 29 50 33 1/20/94 28 9 54 25 1/21/94 29 12 39 28 1/22/94 46 16 50 26 1/23/94 32 17 39 32 1/24/94 26 12 52 19 1/25/94 28 20 50 26 1/26/94 33 9 49 18 1/27/94 22 14 54 21 1/28/94 41 21 49 22 1/29/94 47 17 46 26 1/30/94 46 11 25 37 1/31/94 38 10 55 21 Sub-total (B aseline) ------ 319 6 75 329 Sub-total (P ost E arthquake) 612 442 7 5 7 4 6 2 TOTAL 612 761 1432 791 Grand Total = 3,596 179 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. rate. Of 385 abstraction forms that were selected for comparison, 348 (90%) had no errors after the data had been cleaned. Eight percent had only one error, and two percent had two errors. No more than two errors were found for any record included in the sample used to estimate the transcription error rate. 3.03. Identifying Cases of Earthquake-Related Injuries. Injuries were considered earthquake-related if specifically recorded as such in the medical record, or if the injury was reported to have occurred at 4:30 or 4:31 A.M. on January 17, 1994. A major problem with identifying earthquake-related injuries was due to the fact that this information was rarely documented in the medical record. Therefore, an algorithm was developed to decide whether certain injuries were likely to be earthquake-related based on the day of the week that the patient presented after the event. Injuries were assumed to be earthquake-related if there was an excess frequency any day after the earthquake (from January 17-24, 1994) compared to weekday averages based on frequencies observed for the rest of the month. The methods and rationale used to make these assumptions are presented in the following paragraphs. Associated results of statistical tests are included in Appendix 3. The initial step in determining whether injuries might be assumed to be earthquake-related involved a crude comparison of frequencies of all injuries seen before the earthquake compared to after the earthquake. The day of the earthquake showed obvious increases in frequencies of injuries, so the hypothesis of interest was: 180 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Individual facilities treated an increased number of injuries during a 2-week period alter the earthquake compared with average baseline activity for comparable weekdays. Results of the modified 2-sample t-test (assuming unequal variances) for pre-earthquake weekday averages compared to post-earthquake weekday averages by facility are presented in Appendix 3, Table A3-J. Recall that since Hospital ‘A’ did not include baseline data, no assumptions were made for earthquake-related injuries from that facility. These differences were not remarkable except for Saturdays. One facility showed a decrease in visits on Saturdays, while the other two showed an increase. Since only one day (Saturday) in the two-week period after the earthquake was associated with an increase compared to the two-week period prior to the earthquake, the seven-day period after the earthquake was investigated to address the following hypothesis: Individual facilities treated an increased number of injuries during the 7-day period after the earthquake compared with average activity for comparable weekdays from the rest of the month. Injuries were stratified by facility and day of week (Monday, Tuesday, Wednesday, etc.). The number of injury visits for each day during the week after the earthquake was compared to the weekday averages for the rest of the month using a 1-sample t-test. 181 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Statistical significance was set at p < 0.05. These results are also presented in Appendix 3, Table A 3-2. Increases in the number of injury visits were noted at all facilities for the day after the earthquake (Tuesday, January 18, 1994), and at two facilities for Wednesday, January 19, 1994. Two facilities also showed increases on the Saturday after the earthquake, while one facility in an area of high impact showed decreased activity on the Friday and Saturday after the earthquake. Injuries were then stratified by mechanism (i.e., cuts, falls, motor vehicle-related injuries, etc.) in addition to stratification by facility and day o f the week. The days that showed increases in injury visits during the week after the earthquake compared to the rest of the month were investigated by facility for increases in specific mechanisms of injury. The day of the earthquake was included in this process. The results of 1-sample t-tests by weekday (January 17, 18, 19, 21,22,1994) for specific mechanisms of injury are presented by facility in Appendix 3 (January 17: Table A 3-3, January 18: Table A 3-4, January 19: Table A3-5, January 21: Table A 3-6, January 22: Table A 3-7). On January 17, 1994, an increase in the frequency o f motor vehicle collisions was noted at one facility; increases in the number of falls were noted at two facilities; an increase in the number of environmental poisonings and injuries was noted at one facility; increases in frequencies o f injuries caused by cutting or piercing objects and being struck by objects were noted at all three facilities; an increase in the number of 182 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. injuries caused by being caught in or between objects was noted at one facility, an increase in the number of injuries caused by overexertion was noted at one facility. On January 18. 1994. an increase in the frequency of injuries caused by falling, being cut by or pierced by an object, and being struck by an object persisted at two facilities. On January 19. 1994, am increase in the number of injuries caused by falls and being struck by or against objects persisted in two hospitals, while one hospital showed an increase in the number o f injuries caused by cutting or piercing objects. No increases were noted again until Saturday, January 22, 1994. An increase in the number o f injuries caused by being struck by or against an object was noted at two facilities on that day. Also, a maximum o f one facility showed an increase in the number o f injuries caused by motor vehicle collisions, firearm-related injuries, falls, and being cut by or pierced by an object on Saturday, January 27, 1994. Once increases in the numbers of injuries for specific external causes were detected during the week after the earthquake, injury incidents for those mechanisms were examined individually. A subjective assessment was made to determine if the injury was likely to be earthquake-related. This assessment was based on available details including the patient complaint, the circumstances o f the injury, the physician’s diagnosis, and the date and time of the injury. Examples o f injuries that were included in this group are a motor vehicle-related injury from a crash caused by freeway buckling, falls from stepping onto a porch that had collapsed, carbon monoxide poisoning, cuts caused by stepping on glass from shattered windows, pictures, or mirrors, loss of 183 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. consciousness from falling or being struck by objects, sprains or strains from trying to catch large pieces of furniture that were falling. Injuries were also categorized as indirectly related to the earthquake. Examples of indirect earthquake-related injuries include falling from a ladder while cleaning broken ceiling plaster, being cut by broken glass while replacing windows, sprains or strains from moving furniture back into place or moving boxes, and attempted suicide due to depression exacerbated by loss of housing in the earthquake. If insufficient information was provided, the injury was not assumed to be earthquake-related. After reviewing all possible injury records (for the week after the earthquake) to determine whether the mechanism of injury was related to the earthquake, 684 injuries were attributed or assumed to be attributable to the earthquake. Proportions of injuries attributed to the earthquake by hospital are presented in Table 25. The following general definitions were used to relate earthquake-attributability: 1) Clearly Earthquake-Related Injuries were specifically recorded as such in the medical record, or the injury was reported to have occurred at 4:30 or 4:31 A.M. on January 17, 1994. 2) Assumed Earthquake-Related Injuries were caused primarily from the force of the earthquake and were detected through an excess in the frequency o f injury (by mechanism and day of the week) after the earthquake (from January 17-24, 1994) compared to weekday averages for the rest of the month. 184 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 25. Proportions of Injuries Attributed to the 1994 Northrldae Earthquake at 4 Facilities In Los Anaeles County. January 1994. Not A ssum ed Clearly A ssum ed to be EQ- EQ- EQ- A ssum ed EQ- % EQ-Related Hospital & Time Period R elated R elated’ R elated2 Indirect3 Totals Injuries Hospital 'A '4 Baseline5 0 0 0 0 0 0.00% Post-Event6 528 84 0 0 612 13.73% Hospital '0' Baseline5 319 0 0 0 319 0.00% Post-E vent8 178 34 226 4 442 5 9 7 3 % Hospital 'C' Baseline5 672 3 0 0 675 0 4 4 % Post-Event8 565 85 101 6 757 2 5 3 6 % Hospital *D ' Baseline5 329 0 0 0 329 0.00% Post-Event6 318 47 96 1 462 31.17% A ggregate Baseline5 1320 3 0 0 1323 0.23% Post-Event8 1589 250 423 11 2273 30.09% Totals 2909 253 423 11 3596 11njuries that w ere specifically recorded a s such in th e medical record, or if the injury w as reported to have occurred at 4.30 or 4.31 A.M. on January 17,1 9 9 4 . Injuries cau sed primarily from the force of the earthquake, detected through an ex cess after the earthquake (from January 17 -2 4 ,1 9 9 4 ) com pared to w eekday averages for th e rest of th e month. Injuries sustained during clean-up activities after the earthquake, detected through an excess after the earthquake (from January 1 7 -2 4,1994) com pared to w eekday averages for th e rest of the month. 4 Incom plete Data: No assu m ed earthquake-related injuries, no baseline data. January 1-1 6 ,1 9 9 4 8 January 1 7 -31,1994 3) Assumed Indirect Earthquake-Related Injuries were sustained during clean-up activities after the earthquake and were detected through an excess in the frequency o f injury (by mechanism and day of the week) after the earthquake (from January 17-24. 1994) compared to weekday averages for the rest of the month. Direct earthquake-related injuries were identified clearly in 250 medical records during the 2-week period after the earthquake. Three direct earthquake-related injuries were identified prior to the earthquake. These patients presented for care at the health maintenance organization facility on the 9th and 10th of January, 1994, about one week before the Northridge earthquake. The records were reviewed twice by different individuals to verify the information. Although the medical record documentation reflected that these patients sustained earthquake-related injuries, and a small seismic event was recorded off the coast of Santa Monica at about the same time as the injuries were reported to have occurred, these events were not considered in the analysis. However, these observations are important and will be discussed in Chapter VI. It was assumed that 423 additional observations were also directly earthquake- related. Eleven injuries were assumed to be indirectly attributable to the earthquake. As previously stated, since baseline data were not obtained for Hospital ‘A’ (due to exhaustion of funding), no assumptions were made regarding earthquake-related injuries at that facility. Therefore Hospital ‘A’ reflects only earthquake-related injuries that were 186 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. clearly reported in the medical records as being earthquake-related or were reported to have occurred at 4:30 or 4:31 on January 17, 1994. 4.00. INJURY CODING AND ASSOCIATED ASSUMPTIONS. The injury scenes, body location of injury, and external cause of injury were categorized based on available data. Some assumptions were made in this process. The coding methods are presented along with corresponding assumptions in the following paragraphs. 4.01. Injury Scene Categorization. It was assumed that the address recorded in the demographic section of the medical record was the patient’s residence. This may be a false assumption, since the address may reflect a billing address or the address of a friend or relative. However, it is the best available surrogate for patient residence when specific documentation is missing. Injury incident scenes were categorized as: a) the patient’s residence, b) addresses other than the patient’s residence, or c) non-address locations (sidewalks, streets, highways, or freeways). Non-address locations were coded by census block, based on the 1993 Thomas Guide (Thomas Brothers, 1993). Injury incidents that occurred at public buildings, retail stores, office buildings or other buildings without an address recorded in the medical record were investigated individually to obtain the structure address either through the Thomas Guide (Thomas Brothers, 1993) or telephone directory assistance. 187 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.02. Coding of Body Location and External Cause of Iniurv. Up to four fields were available to denote the body location of the injury. The first field was used for descriptive statistics and for modeling purposes since this indicated the primary body area affected (or the body location most severely impacted). Body locations were grouped as: head and neck (including face), upper extremity, lower extremity, spine, thorax, abdomen (including pelvic contents), and unspecified. Similarly, up to four fields for identifying mechanisms of injury were available for a given incident. For example, one injury incident could involve a person (1) tripping over a cat, (2) falling down the stairs, (3) hitting their head on the banister, and (4) twisting an ankle upon landing. The first field was used for summary statistics and modeling because it was coded with the primary mechanism o f the most severe injury. For example, if a person lost consciousness from hitting their head on the banister and sprained an ankle upon landing, the mechanism associated with the loss of consciousness (struck head on banister) would be coded first. Non-specific injury mechanisms that were described as earthquake-related were retained as ‘Unknown’ earthquake-related mechanisms. Other injuries of non-specific mechanism were deleted. Categories o f mechanism of injury included motor vehicle collisions, firearm-related injuries, falls, poisonings, exposure to environmental elements, being injured by an animal/amphibian/reptile/insect/plant, cutting or piercing mechanisms, being struck by or against object(s), being caught in or between object(s), drowning or submersion, suffocation/strangulation/hanging, fire/bum, fight/brawl, machinery-related injury, homicide/assault, suicide/suicide attempt, child abuse/assault, 188 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. rape/sexual assault, domestic violence, overexertion, other external causes, and unknown or missing mechanisms. Vague descriptions, particularly those that focused only on intent (i.e., assault), were used only when additional details were missing. 4.03. Iniurv Severity Scoring The circumstances of the injury were reviewed along with diagnoses and all available fields denoting body location and mechanism of injury to score injury severity for all earthquake-related injuries (assumed and documented). Standard ISS coding guidelines were followed using the recommended Abbreviated Injury Severity (AIS) rules. These rules include a dictionary of scores that covers over 2,000 injury descriptions. General AIS and ISS coding guidelines and rules are presented in Appendix 4. The AIS dictionary is organized by body location of injury, and the rules specify whether an injury description may be used or whether clinical diagnostic confirmation is required. If insufficient information was available to score injury severity, the ISS score was coded as ‘99’, and the observation was not included in the analysis. Since the range for usable ISS has possible values o f 1-76, (l=minimal injury, 76=unsurvivable injury), and most injuries are minor, this distribution is positively skewed " In this study data-set, the ISS range was 1-25 with 1% (n=8) characterized by an ISS greater than or equal to 10. Rather than deleting observations that were outliers, Injury Severity Scores were grouped for analysis into three categories based on AIS severity levels in conjunction with ISS rules: ISS ^ 3 were considered minor; ISS 4-8 w Positive skewness is characterized by values above the median being farther from the median in absolute value than values below the median.(Rosncr. 1990). 189 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. were considered moderate; ISS > 9 were considered severe. Since the ISS is the sum of the highest AIS code in the three most severely injured body regions (Baker et al., 1974), and AIS scores of 1, 2, and 3 correspond to minor, moderate, and serious injuries (Association for the Advancement of Automotive Medicine, 1990), it was ssumed that an individual with: a) only minor wounds would have a maximum ISS of 3 ( l 2 + I2 + l 2 = 3). b) a moderate injury would have a minimum ISS o f 4 (22 = 4). c) a serious injury would have a minimum ISS of 9 (32 = 9). Injury severity scores were calculated for 641 of the 684 post-event earthquake- related injuries. Fifteen observations reflected mechanisms that were not included in the AIS dictionary (drug or alcohol-related events), and consequently could not be scored. Twenty-eight observations did not include sufficient information in the injury description to be scored for injury severity and were assigned an ISS o f ‘99’. Of the 644 earthquake-related injuries that were scored, 81% were minor (ISS s 3), 14% were moderate (ISS 4-8), and 5% were serious (ISS £ 9). Summaries o f injury severity scores by earthquake-relatedness (clearly identified as earthquake-related, assumed directly related to the earthquake, assumed indirectly related to the earthquake) are presented in Table 26 for post-event injuries. A smaller proportion of all clearly identified earthquake-related injuries was minor (75%) compared to the proportion of all assumed earthquake-related minor injuries (83%). Similarly, a 190 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. larger proportion of all clearly earthquake-related injuries was serious (7%) compared to the proportion of all assumed earthquake-related serious injuries (4%). There were few assumed indirectly earthquake-related injuries, and all of these were minor. Table 26. Iniurv Severin' Scores for Earthquake-Related Injuries from 4 Emergency Departments in Los Anseles County. January 17-31. 1994. Injury Severity Score Group Clearly Earthquake- Related Injuries* Assumed Earthquake- Related Injuries* Assumed Indirect Earthquake-Related Injuries* Totals* Minor Injury (ISS=l-3) 170 (75) 337 (83) 10(100) 517(81) Moderate Injury (ISS=4-8) 40(18) 51(13) 0 91 (14) Serious Injury (ISS*9) 17(7) 16(4) 0 33 (5) Totals 227 404 10 641 ♦Frequency (Column Percent) Summaries of ISS by facility are presented in Table 27. The distributions of injuries by severity differ between the facilities. This is expected since the level of care 191 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. varies at each of the facilities. Ninety-two percent of the injuries treated at the basic emergency department were minor, and only 1% were serious. In contrast, the Trauma Centers treated a larger proportion of serious injuries (5-8%) and a smaller proportion of minor injuries (63-79%). Table 27. Iniurv Severity Scores b v Facility for Earthauake-Retated Injuries Treated at 4 Emergency Departments in Los Angelas Countv. January 17-31.1994. Minor injuries1 Moderate Serious Injuries1 Facility (ISS 1-3) Injuries1 (ISS 4-8) (ISS > 8) Totals Hospital 'A' 2 48 22 6 76 (63) (29) (8) Hospital 'B' 200 36 18 254 (79) (14) (7) Hospital 'C' 172 14 1 187 (92) (7) (1) Hospital 'D' 97 19 8 124 (78) (15) (6) Totals 517 91 33 641 (81) (14) (5) 1 Frequency (Percent) ~ Incomplete: No assumed earthquake-related injuries, no baseline data. 192 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.00 LINKING INJURY DATA TO ENGINEERING DATA. Injury incident locations were provided to EQE, International, to be linked to existing databases of building structures and geologic conditions. The structural characteristics of the buildings were obtained from the Los Angeles County Assessor’s Tax Roll for the year 1993. Definitions for terms used to categorize structural elements are included in Appendix 5, Table 2, of the Memorandum from EQE dated October 30, 1998. Injury incident locations were also linked to Los Angeles City Building and Safety Reports of Inspection that were conducted after the earthquake to record structural damage and associated hazards. The matching process between injury incident scene and the structural engineering databases is described with associated results in Section 5.01. Once an injury incident location was linked to the County Assessor’s database, the parcel number used to identify that property was linked to a geologic database. The geologic database included estimates o f ground movement, soil conditions, and perceived intensity by property parcel number. The matching process between the County Assessor’s property parcel number from injury incident scenes and the geologic databases is described with associated results in Section 5.02. 5.01. Linkage of Iniurv Characteristics to Building Databases. Not all data elements were uniformly recorded throughout the Assessor’s Tax Roll database. Therefore, not all of the building structure elements were available for each address. Since taxes are based primarily on location, size, and use of the structure, these elements were most likely to be available. Additionally, since governmental 193 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. facilities are not taxed, they are not likely to be re-inspected over time. Therefore, information relating to these structures is considered less accurate than other structures. Details regarding the results of the matching process are presented in the final report submitted by EQE, and are included in Appendix 6. Overall, 65% (n=427 of 641) of the earthquake-related injury records were matched to the Assessor’s database. Los Angeles City Building and Safety Inspection reports were also linked to the address recorded in the medical record. These reports reflect inspections conducted after the earthquake and include information regarding tagging placards that summarized building damage. Three categories of damage were recorded (Holmes & Somers, 1996). Red Tagged buildings were unsafe: extreme hazard was found, the structure was in imminent danger of collapse, and entry was prohibited except by authorities. Yellow Tagged buildings permitted limited entry: dangerous conditions were believed to be present, owners were allowed to enter only for emergency purposes, no continuous usage was allowed, and no public entry was permitted. Green Tagged buildings were considered safe, although repairs may have been required: no restrictions were placed on occupancy or use. Categories and frequencies of structural characteristics linked to injury scenes are displayed in Table 26. Available structural characteristics included square footage o f the building and the unit, structure construction type (wood-framed, steel-framed, concrete- framed, un-reinforced masonry, and a special category that included anything not fitting into the other 4 categories), year of construction, and structure use (single family housing, duplex, multifamily housing, mobile home, commercial, and governmental use). 194 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 28. C haracteristics o f Structures Linked to Iniurv Scenes for P atien ts M u red from the Northridam Earthquake and Treated at 4 Emergency Departments In Los Angelas Countv. January 17-31. 1994. Structural Characteristics Number (%) of Earthquake-Related Injuries by Level of Injury Severity Minor1 Moderate2 Serious3 Total S iz e o f B uildina L e s s th a n 1 4 6 8 S q u a re F e e t 8 0 (7 7 ) 1 6 (1 5 ) 8 (8 ) 104 1 4 6 8 - 2 2 1 7 S q u a re F e e t 8 2 (7 9 ) 1 8 (1 7 ) 4 ( 4 ) 103 2 2 1 8 - 8 4 7 9 S q u a re F e e t 8 4 (8 1 ) 1 3 (1 3 ) 6 (6 ) 104 G r e a te r th a n 8 4 7 9 S q u a re F e e t 8 5 (8 2 ) 1 6 (1 5 ) 3 ( 3 ) 104 M issing 1 8 6 (8 3 ) 2 8 (12) 1 2 ( 5 ) 226 S iz e o f Unit L e s s th a n 1351 S q u a re F e e t 8 2 (7 9 ) 1 4 (1 3 ) 8 (8) 104 1351 - 1 8 3 6 S q u a re F e e t 8 3 (8 0 ) 1 7 (1 6 ) 4 ( 4 ) 104 1 8 3 7 - 4 9 9 1 S q u a re F e e t 86 (8 3 ) 12(12) 5 ( 5 ) 104 G r e a te r th a n 4 9 9 1 S q u a re F e e t 8 0 (7 7 ) 2 0 (19) 4 ( 4 ) 104 M issing 1 8 6 (8 2 ) 2 8 (12) 12(6) 226 S tru c tu re U s e S in g le/D u p lex H ousing 1 8 0 (8 3 ) 2 9 (13) 8 ( 4 ) 217 Multi-family H ou sin g 1 4 4 ( 7 7 ) 3 3 (18) 11 (5 ) 18 8 M obile H o m e 4 ( 8 0 ) 0 1 (2 0 ) 5 C o m m e rc ia l U se 7 (1 0 0 ) 0 0 7 Institutional / G o v e rn m e n ta l U se 9 ( 9 0 ) 1 (10) 0 10 M issing 1 7 3 (8 1 ) 2 8 (13) 1 3 ( 6 ) 214 Y e a r Built P re -1 9 4 3 8 4 (7 9 ) 1 8 (1 7 ) 5 ( 4 ) 107 1 9 4 3 -1 9 6 0 61 (7 7 ) 1 4 (1 8 ) 4 ( 5 ) 79 1 9 6 1 -1 9 7 5 91 (7 8 ) 1 8 (1 5 ) 8 ( 7 ) 117 P o st-1 9 7 5 9 4 (8 5 ) 1 3 (1 2 ) 3 ( 3 ) 110 M issing 1 8 7 (8 2 ) 2 8 (12) 1 3 ( 6 ) 228 S tru c tu re C o n stru c tio n T voe S te e l F ra m e 3 (1 0 0 ) 0 0 3 C o n c re te F ra m e 1 (5 0 ) 1 (50) 0 2 U n rein fo rced M aso n ry 5 (1 0 0 ) 0 0 5 W o o d -F ra m e 3 0 4 (7 9 ) 6 0 (15) 2 0 (5) 384 S p ecial 2 ( 6 7 ) 0 1 (3 3 ) 3 M issing 2 0 2 (8 3 ) 3 0 (12) 1 2 ( 5 ) 244 C itv B uildina In sD ec to r's P o s t-E a rth a u a k e I R eD ort G re e n T ag (S a fe ) 6 5 (8 1 ) 1 2 (1 5 ) 3 ( 4 ) 80 Y ellow T ag (L im ited Entry) 8 ( 7 3 ) 2 ( 1 8 ) 1 (9) 11 R e d T a g (U n safe) 8 ( 8 9 ) 0 1 (1 1 ) 9 U nknow n (In sp e c te d , b u t n o t ta g g e d ) 1 (1 0 0 ) 0 0 1 No R e c o rd o f In spection 4 3 5 (8 1 ) 7 7 ( 1 4 ) 2 8 (5) 540 T otal 517(81) 91 (14) 33(5) 641 ' Injury S e v e rity S c o r e s 1-3 2 Injury S e v e rity S c o r e s 4 -8 3 Injury S e v e rity S c o r e s s 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Structure use was recorded most (n=427, 67%), followed by square footage (n=415, 65%), year built (n=413, 64%), structure construction type (n=397, 62%) and tagging information (n=101, 16%). Ninety-four percent of the injury location addresses linked to structures were used as single/duplex housing (n=217) and multi-family housing (n=188). Five structures were identified as mobile homes, seven as commercial buildings, and ten as governmental (institutional) buildings. Square footage of both the building and units were categorized by quartile after deleting missing data. Twenty-six percent of the buildings were built prior to 1943, 19% were built between 1943-1960, 28% were built between 1961-1975, 16% were built between 1976-1987, and 11% were built after 1987. Two categories (1976-1987 and post-1987) were collapsed into one group for subsequent analyses. Nearly all of the available structure types (96%) were wood-framed houses. Eighty percent of the inspected structures were safe (green tag, n=80), 11% were limited entry (yellow tag, n=l 1), 9% were unsafe (red tag, n=9), and one building was reported as inspected, but no damage (tagging) information was recorded. 5.02. Linkage of Iniurv and Building Characteristics to Geologic Database. Linkage of geologic characteristics to structures was accomplished through the parcel number from the County Assessor’s database for injury incidents that were successfully matched to that database. Parcel numbers were then linked to the EQE/Office of Emergency Services (OES) Damage Database that was developed during the Northridge Earthquake response. Details regarding this database are also discussed in 196 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the final report submitted by EQE (see Appendix 6). A limitation to this methodology is the exclusion of injuries that did not occur in a structure or property associated with a structure. Notably absent are any vehicular events unless they occurred in a driveway or parking structure. Data available from the EQE/OES Damage Database included estimates of geologic activity including Modified Mercalli Intensity (MMI), peak ground acceleration (PGA), and Evemden 2.5” Soil Code (site geology definitions). Definitions for these terms are included in Appendix 6, Table 2, of the Memorandum from EQE dated October 30, 1998. PGA estimates were based on initial reports of seismic readings immediately after the earthquake. Over time, more precise estimates have been generated. However, the improved estimates have not been linked to the County Assessor’s database, therefore the original PGA estimates were used in this analysis. Linkage of the improved estimates to the County Assessor’s database requires additional funding not included in this study. Frequencies and categories of geologic groupings are shown in Table 29. Available geologic characteristics included estimates of soil type (n=419), MMI (n=415), and PGA (n=420). Approximately one-third of the data were missing, as expected since geologic data were linked through parcel number which was not identified for 427 of the 641 earthquake-related injuries. More than one-third of all earthquake-related injury scenes occurred in areas that were characterized by sedimentary soil that was not thought to be liquefiable. The range for MMI levels associated with injury scenes was VI — DC Sixty percent of all earthquake-related injury scenes occurred in areas of MMI=VII and VTII. The range for PGA was 0.30 g to 0.94 g. 197 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 29. Characteristics o f Estimates o f Geologic Conditions at Sites Linked to Iniurv Scenes for Patients Inlured from the Northrldae Earthquake and Treated at 4 Emergency Departments In Los Angeles County. January 17-31.1994. Number (%) o f Earthquake-Related Injuries Geologic Characteristics Minor1 Moderate2 Serious3 Total Evernden Soil Code Paleozoic / Mesozoic Rock 4(80) 1(20) 0 5 Oligocene / Pliocene Rock 12(63) 5(26) 2(1) 19 Pliocene / Pleistocene Rock 72 (85) 7(8) 6(7) 85 Liquefiable Sedimentary Soil 42 (90) 3(6) 2(4) 47 Other Sedimentary Soil 205 (78) 48 (18) 10(4) 263 Missing 182 (82) 27 (12) 13(6) 222 Modified Mercalli Intensity (M M I) V I 29 (91) 3(9) 0 32 V II 157 (76) 36 (18) 12(6) 205 V III 139 (81) 24 (14) 8(5) 171 IX 5(72) 1(14) 1(14) 7 Missing 187 (83) 27 (12) 12(5) 226 Peak Ground Acceleration (PGA) Less than 0.62 g 77(78) 18(18) 4(4) 99 0.62 - 0.67 g 91 (85) 14 (13) 2(2) 107 0.68 - 0.79 g 80 (83) 11 (12) 5(5) 96 Greater than 0.79 g 87 (74) 21 (18) 10(8) 118 Missing 182 (83) 27 (12) 12(5) 221 V O 00 11njury Severity Scores 1-3 *' I , Severity Scores 4-8 3 Injury Severity Scores ^ 9 9 Sixty-three percent of the injury scenes that matched the geologic database were characterized by sedimentary soil that was not thought to be liquefiable. Forty-nine percent of the injury scenes that matched the geologic database were identified in areas of MMI=VTL and 41% occurred in areas of MMI=VTH. Twenty-eight percent of the injury scenes that matched the geologic database were identified in areas that exhibited peak ground acceleration less than 0.62 g. Twenty-four percent were in areas of peak ground acceleration greater than 0.79 g 6.00. STATISTICAL METHODS. Methods for analysis included establishing cut-points for quantitative variables, grouping ordinal and qualitative variables, statistical modeling techniques, and associated computer software. These processes will be detailed in the following paragraphs. 6.01. Grouping of Demographic Variables. Available demographic characteristics included gender (n=641), ethnicity (n=440), and age (n=634). Ethnicity was grouped into 4 categories: Caucasian, African- American, Hispanic, and Other (Egyptian, Iranian, Persian, Armenian, Ethiopian, and not able to categorize). Age was grouped into 10-year categories. However, due to sparse observations in the extreme age-groups, the top two categories (age 60-69 and 70 and greater) and the lowest two categories (ages 0-9 and 10-19) were collapsed for analysis. 199 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6.02. Establishing Cut-Points for Year of Building Construction. Three different approaches were used to establish categorical cut-points for the year that the structure was built. The first approach was to categorize structures by the years of major building code revisions (Holmes & Somers, 1996). These building codes are based on local (city) codes as well as Uniform Building Codes (UBC) which are national standards set for engineers and architects for structural construction. A . limitation to this approach is that not ail building contractors make changes at the same time. There may be a lag between enactment of a code and enforcement, and some locales may enforce a change prior to enactment of a code. Therefore these cut-points may be imprecise measures of actual changes in building practices. Four major code revisions were noted: 1) in 1943, a maximum of 13 stories for buildings was established for the city of Los Angeles, 2) in 1959-1961, the height limit was removed and ductile framing was required when the building was greater than 13 stories, 3) in 1976, a major revision of the Uniform Building Code (UBC) was introduced, including a significant increase in the amount of base shear motion that short and mid-height buildings were built to withstand, and 4) in 1988, another major revision of the UBC was introduced, including a change in the equation used to calculate base shear motion. Shoaf & Bourque (1999) used slightly different cut-points (1928, 1960, 1975) to analyze a population-based survey of community response to the Northridge Earthquake. This survey included questions regarding structural damage from the earthquake. Although two cut-points in that analysis coincided well with the previously identified housing code revisions, the lowest cut-point (1928) reflected the first recorded building 200 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. codes (Shoaf, 1999). The analysis of this dissertation is based on three cut-points: 1942, I960, and 1975. Categories of older houses were investigated in reference to houses built after 1975. 6.03. Grouping of Geologic Variables. Due to the sparse number of observed severe injuries in MMI levels VI and EX, these categories were collapsed into adjacent categories (MMI VI and VII were combined. MMI VTII and IX were combined). Unless testing for a trend, all further analyses grouped MMI as a dichotomous variable, comparing risk for more severe injuries in MMI VIII and IX relative to MMI VI and VII. PGA estimates were categorized by quartile. Soil types were categorized as non-sedimentary soils (rocks from the Pleistocene Era or older, n=109), liquefiable sedimentary soil (n=47), and other sedimentary soil (n=263). 6.04. Statistical Modeling. Analytic Methods and Computer Software. Proportional polytomous logistic regression and dichotomous logistic regression were used to model estimated relative risk for injury severity. The logistic model is based on an outcome variable that follows the binomial distribution.x This outcome is commonly denoted as either a success or a failure, or the presence or absence of a characteristic or condition. Specifically, the number of success in a sample of n independent observations with a probability o f success equal to p is characterized by the binomial distribution. x All explanations and equations of logistic modeling are taken from Rosner. 1990. 201 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Logistic regression involves relating one or more independent variables to the probability of occurrence of an outcome variable that is characterized by the binomial distribution. However, the independent variables used to predict the probability of an outcome may not be characterized by the binomial distribution. To constrain parameter estimates from the model so that predicted values of the event probability will lie between 0 and I, the outcome variable (p) is transformed through the logit function: logit ip) = In [p/(l-p) ] The logit may take on any value from negative infinity to positive infinity: logit (p) = In [p/ (\-p) ] = a + 3 lX l + ... +&xk where: xi, x2, x3 ... x^ are independent variables In multiple logistic regression, the logit (p) is modeled as a linear function of the independent variables and their associated regression coefficients. Solving forp, the model can be expressed as follows: a -p ix l * ... - Pkxk p = --------------------------------- for an individual who has the event; I ~ e a / 1 -p = for an individual who does not have the event. j _ e a -Plxl - ... -Pkxfc 202 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This transforms the probability back to an interpretable value in which the fitted value must always lie between 0 and 1. It represents the individual logistic likelihood (for an individual subject). The total likelihood is obtained by taking the product of the individual likelihood for all study subjects. Because there are many possible values for the estimates o f the parameters Pi-3k, it is useful to assess the amount of support for different probabilities that are obtained based on the study data. The most likely value for 3i...3k, given the study data, is termed the maximum likelihood estimate (MLE), and is obtained by estimating the values of Pi . Pk that achieve the highest values for the total likelihood. In this study, the statistical significance of individual variables was investigated by comparing hierarchical models through the likelihood ratio test. This test compares the total likelihood evaluated at the maximum likelihood estimates o f parameters to the likelihood evaluated at the null value (3 = 0) for the parameter. The test statistic3 ' has an approximate Chi-square distribution with degrees of freedom equal to the number of parameters being tested (if the test hypothesis is true and associated assumptions are met). Reasonable interaction terms were also investigated for statistical significance by comparing hierarchical models (those with and without the interaction term) through the likelihood ratio test. Statistical significance was conducted at alpha=0.05; all statistical testing was two-sided. Univariate associations, referring to the modeling of a single independent variable with respect to grouped Injury Severity Score, are presented along with associations that control for the confounding effects of other variables. - 2 times the natural log of the likelihood ratio 203 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The outcome variable for the polytomous model was a 3-level category for injury severity score (minor: ISS 1-3, moderate: ISS 4-8, serious: ISS ^ 9). Since few serious injuries were observed (33 of 641 earthquake-related injuries), a dichotomous model was also considered. In the dichotomous model, minor injury severity scores (ISS 1-3) were compared to the combined grouping of moderate and severe injury severity scores. Data were grouped by type of variable and analyzed by variable group with respect to injury severity score. Variable groups that were included in analysis are presented in Table 30. Table 30. Variable Groups and Corresponding Variables. Variable G roup Variables Patient Demographics Age, Gender, Ethnicity, Facility Injury Characteristics Body Location of Injury, Mechanism o f Injury Structural Engineering Structure Use, Structure Size, Unit Size, Type of Characteristics Construction, Year of Construction Geologic Characteristics Modified Mercalli Intensity, Peak Ground Acceleration, Evemden 2.5” Soil Code Demographic characteristics (age, gender, ethnicity and facility) were considered potential confounders that might be associated with independent risk factors from other variable groups and the outcome of interest. For example, the elderly and the very young are less mobile; hence injury characteristics may be different for these patients than for others. Additionally, cultural protocols may differ regarding response to earthquakes; some ethnicities may be inclined to stay inside whereas others might 204 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relocate to temporary shelters immediately. Since Level I Trauma Centers are receiving hospitals for very severe injuries and might also be related to independent risk factors, especially location-specific risk factors (e.g., demographics or geologic characteristics), facility type was also considered a potential confounder. Furthermore, since real estate values are determined (in part) by socio-economic status, structural characteristics are likely to be related to the demographic make-up of the community. Therefore demographics were added to each variable group individually, and were included in all models to control for potential confounding bias. Analyses were performed using Statistical Analysis System Software (SAS®, Cary, North Carolina). Polytomous and dichotomous logistic regression used the Proc Logistic module of SAS. This module performs ordered logistic regression using the proportional odds model. Resulting parameter estimates obtained in the trichotomous model are interpreted as estimating the risk of serious injury relative to moderate injury, and the risk of moderate injury relative to minor injury. Power calculations were performed using Power© Version 1.30 (Epicenter Software, Pasadena, California). 205 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C H A P T E R V: R E SU L T S Descriptive results will be presented for variables pertinent to this study in the following section. Although data were obtained regarding details of the hospital visit, this information is useful from a policy and disaster management perspective, and does not provide information directly associated with risk or hazard assessment. Therefore, it is not pertinent to the modeling process and is not presented here. Baseline versus post event descriptive comparisons of each facility will be presented, followed by descriptive characteristics of the study sample. Results of the logistic regression modeling, model selection process and the resulting model parameter estimates (and corresponding confidence intervals) will then be presented by variable group (patient demographics, injury characteristics, structural and geologic components). 1.00. BASELINE VERSUS POST-EVENT DESCRIPTIVE RESULTS. Facility activity and the demographic characteristics of the patients presenting for treatment may be different before the earthquake compared to after the event. In order to investigate this, data are presented by time period (before and after the earthquake) and facility. In addition, the geographic distribution of injured patients may be different before the earthquake compared with after the event. Therefore maps of injuries by time period and facility are also presented. 206 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.01. Facility Activity. Daily frequencies of injury visits to the four facilities (after deleting records with missing mechanisms of injury and invalid addresses) are presented in Figure 13. It appears as though the facilities returned to their respective baseline levels of activity relatively quickly (within 1-3 days) after the earthquake. Hospital ‘A’, a Level II Trauma Center in West Los Angeles, shows no baseline data, but shows a definite spike in activity on the day of the earthquake (January 17, 1994). Hospital ‘B \ a Level II Trauma Center in the Santa Clarita Valley, shows the most impact for the day of the earthquake (January 17). This facility was in a location of more intense ground motion than the other facilities. Hospital ‘C \ a health maintenance facility basic emergency department located in West Los Angeles, was the least impacted on the day of the earthquake, and it also has a higher daily average than the other facilities. Hospital ‘D ’, a Level I Trauma Center in West Los Angeles, shows definite impact the day of the earthquake, but appears to return to normal very quickly. It is easier to view the post-event activity compared to baseline activity after removing the day of the earthquake, since that day had an extreme impact on emergency department activity. Figure 14 shows the data without the totals for the day of the earthquake. The frequency of treated injuries at Hospital ‘B’ on 1/18/94 is nearly the same as the frequency on l/l/94.z Additionally, a slight increase in frequency of treated z January 1 is often associated with increased injuries that are thought to be related to New Year's Eve celebrations. 207 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 13. Daily Frequencies of Injury Visits to 4 Emergency Departments in Los Angeles County. January 1994. 250 200 S' 150 a> 3 100 50 3 t j t t t S ? 2 ? o s e e t e 1 r) % iri (d K c j c ? c * 3 £ Date of ED Arrival K ) O 00 Hospital ’A' □ Hospital 'B' □ Hospital 'C' □ Hospital * D * 1/28/94 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 14. Daily Frequencies of Injury Visits to 4 Em ergency Departments in Los A ngeles County. 1/94. Deleting 1/17/94. I i i § ? ? 1 1 c j c j 5 5 Date of ED Arrival to o VO ■ Hospital 'A' B Hospital 'B' □ Hospital 'C' □ Hospital *D' /31 /9 4 8822 injuries can be seen on 1/14/94 through 1/16/94 compared with the rest of the month at Hospital ~B\ This type of pre-event increase may be due to increased injuries due to the weekend in general or the fact that the weekend was a holiday weekend. Since emergency departments usually show variation in activity by day o f the week, data is presented by facility (for each facility that had data collected for the entire month) and weekday in Figures 15-17. Each of these facilities shows slightly different patterns that will be discussed separately. Facility 'B ’ shows an increase in activity by day of the week for January 17-19 (Monday - Wednesday) (see Figttre 15). Activity seems to be back to normal by the Thursday following the earthquake, and in fact the Friday following the earthquake seems to show lower activity than usual for a Friday. Figure 16 shows activity at Hospital ‘C \ Activity at this facility seems above normal for the day of the earthquake (Monday) and the next day (Tuesday), but activity is only slightly higher on the Wednesday after the earthquake compared to other Wednesdays from the month of January, 1994. Figure 17 shows activity at Hospital ‘D \ This facility shows an obvious increase in activity the day of the earthquake, but only a slight increase for the Tuesday, Wednesday, and Thursday following the event. 210 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. c a > I to Q (b Q o c < b 5 C D S UJ C Q § s C O s £ .3 o C O ,< b 0 c : CD Of 1 10 T — a? = 3 o 11 Oi 0 5 T— £ cc c : CD '" 3 • O -ac C D JD - Q 1 v — 0 0 ! C M C M C O 1 N C c c ■ C e C O C D C D C D C D “ 3 “ 3 - J 1 - > ” > i v n C M o> 0 0 C M C M ■ c c c C e C D C D C O C D C D “ 3 - 3 □ □ □ a □ o m C M o o C M o ID O ID Vt ^ 9 . " a A9uanb9Jj 211 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 16. Frequencies o f Iniurv Visits t o Hospital 'C' Emergency Department bv C D C D Y — £ CD 3 C 3 I < b I T- T- » 9- 9 #■■ 9 • 9 + 9 m * 9 » ■ »■ z 9 * * 9— : j P_ 9- 9- J - .*_ .#• . » ' ' J> '/ V ‘-P_;9 ' • # - ^ »_ .*_ JP_ *' . »_ .#■ ^ • #■_ 9- 9_ 9- 9 - 9- M‘ 9- 9'9— 9- 9- 9' 9- 9- 9— 9- 9'. r_ 9-9 ‘ 9^ 9- 9- 9- 9. 9- 9 ‘ 9~ 9- *_ 9_ 9- 9- 9- . m m m m m m s m i ' w i2 QQ QQ Q QQ SK 2Q QQ Q QQ QQ QQ SQ % g o o oo r^- to o o t o ^ CO CSI Aouanbaij * y ^ 9 212 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. D a y o f Week Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 17, Frequencies of Injury Visits to Hospital 'D' Em ergency Department bv W eekday. January 1994. 120 100 80 □ Jan 1-Jan 7 □Jan 8-Jan 14 □Jan 15- Jan 21 B Jan 22-Jan 28 □ Jan 29-Jan 31 Day of Week 1.02. Demographic Make-up of Patients bv Time Period. Gender (male-female) ratios o f injured patients treated at the four facilities for the month of January, 1994 are presented in Table 31. The managed care facility (Hospital ‘C’) showed similar ratios before and after the earthquake, and for earthquake-related injuries compared to all other injuries. Two facilities showed similar gender ratios for non-earthquake-related injuries before and after the event. However, three facilities treated more females than males for post-event earthquake-related injuries. Age distributions of the injured patients treated at the four facilities for the month of January, 1994 are presented in Table 32. There is only a slight variation in the age distribution between and within facilities by time period (pre and post earthquake). The median age for injured patients that were treated during the baseline time period (January 1-16, 1994) at all facilities was 31 years (range of less than I to 93 years of age). The median age for injured patients that were treated during the post-event time period (January 17-31, 1994) at all facilities was 35 years (range o f less than 1 to 96 years of age). The age distribution appears to shift to a slightly older population (30 years and older) for earthquake-related injuries than for non-earthquake-related injuries during the post-event period. After the earthquake, the median age for patients that were treated at all facilities for injuries that were not associated with the earthquake was 32 years (range of less than 1 to 96 years of age). Patients treated for injuries after the earthquake that were clearly associated with the earthquake were older, with a median age of 42 years (range 1-93 years of age). Patients treated for injuries that were assumed to be 214 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. K ) . M W . W M w w . . w . M U M . f » » a y « y w January 1994 (Pre and Post Earthauake). Hospital & Time Period Not EQ-Relaied Males Females Ratio1 EQ-Related (Direct & Indirect) Males Females Ratio1 Hospital 'A'2 Baseline3 Post-Event4 267 258 1.03 31 53 0.58 Hospital 'B ' Baseline3 Post-Event4 192 112 127 66 1.51 1.70 0 119 0 145 0 0.82 Hospital 'C Baseline3 Post-Event4 336 286 333 279 1.01 1.03 0 96 3 96 0 1.00 Hospital 'D * Baseline3 Post-Event4 195 191 134 127 1.46 1.50 0 54 0 90 0 0.60 Aggregate Baseline3 Post-Event4 723 856 594 730 1.22 1.17 0 300 3 364 0 0.78 Totals 1579 1324 1.19 300 387 0.78 1 Ratio of injured males to females presenting at emergency department during 2-week period. 2 Incomplete Data: No assumed earthquake-related injuries, no baseline data. 3 January 1-16,1994 4 January 17-31,1994 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 32. A ge D istribution o f Inlured P a tien ts T rea ted a t 4 E m e rg e n c y D e p a rtm en ts In L os A n g e le s C ounty. Jan u ary 1994. P ro a n d P o s t N orth ridae E arthauake. 10^19 2039 3039 4049 5039 6039 Unknown ___________________________ < 10 Yrs1 Yrs1 Vra1 Yrs1 Yrs1 Yrs1 Yrs1 70 + Yrs1 Age1 Totals Hospital 'A'2 Baseline,3 Non-EQ-Related 0 0 0 0 0 0 0 0 0 0 Post-Event,4 Non-EQ-Related 47 (8.90) 49 (9 28) 111 (21.02) 128 (24.24) 66 (1250) 35 (663) 30 (5.68) 57 (10 80) 5 (095) 528 Post-Event,4 EQ-Related 0 (0.00) 1 (1.19) 13 (15.48) 27 (32.14) 13 (15.48) 1 1 (13.10) 2 (238) 17 (20.24) 0 (000) 84 Totals 47 (7.60) 50 (0.17) 124 (20.26) 155 (25.33) 79 (12.91) 46 (7 52) 32 (5.23) 74 (12.09) 5 (0.02) 612 >spital 'B' Baseline,3 Non-EQ-Related 62 (19.44) 55 (17.24) 53 (16.61) 74 (23.20) 32 (10.03) 18 (5.64) 1 1 (3.45) 13 (4.08) 1 (0.31) 319 Post-Event,4 Non-EQ-Related 23 (12.92) 31 (17.42) 41 (23.03) 28 (15.73) 29 (16.29) 13 (7.30) 5 (2.81) 8 (4.49) 0 (0.00) 178 Post-Event,4 EQ-Related 17 (6.44) 23 (8.71) 44 (16.67) 69 (26 14) 50 (18.94) 19 (7.20) 19 (7.20) 21 (795) 2 (0.76) 264 Totals 102 (13.40) 109 (14.32) 138 (18.13) 171 (22.47) 111 (14.59) 50 (6.57) 35 (4.60) 42 (5.52) 3 (0.39) 761 to ON Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 32. (Continued) T cT l9 < 10 Yrs1 Yrs1 Hospital 'C ' B aseline,3 N on-E Q -R elated 70 116 (10.42) (17.26) Post-E vent,4 N on-E Q -R elated 60 76 (10 62) (13 45) Post-E vent,4 E Q -R elated 6 16 (3.13) (8.33) Totals 136 208 (9.52) (14.56) Hospital 'D ' B aseline,3 N on-E Q -R elated 24 28 (7.29) (8.51) Post-E vent,4 N on-E Q -R elated 25 35 (7.86) (11.01) P ost-E vent,4 E Q -R elated 4 6 (2.78) (4.17) Totals 53 69 (6 7 0 ) (8.72) 20-29 30-39 40-49 50-59 60-69 Unknown Yrs1 Yrs1 Yrs1 Yrs1 Yrs1 70 + Yrs1 Age1 Totals 121 (18.01) 90 (13 39) 102 (15.18) 59 (8.78) 50 (7.44) 59 (8 78) 5 (0 7 4 ) 672 95 (1 6 8 1 ) 106 (18.76) 79 (1 3 9 8 ) 49 (8 6 7 ) 41 (7 26) 53 (9 3 8 ) 6 (1 0 6 ) 565 32 (1 6 6 7 ) 38 (19.79) 32 (1 6 6 7 ) 30 (15 63) 18 (9 38) 15 (7 81) 5 (2 60) 192 248 (17.35) 234 (16.38) 213 (14 91) 138 (9.66) 109 (7.63) 127 (8.69) 16 (1.12) 1429 100 (30.40) 75 (22.80) 32 (9.73) 28 (8.51) 15 (4.56) 27 (8.21) 0 (0 0 0 ) 329 105 (33.02) 51 (1 6 0 4 ) 41 (12.89) 18 (5 6 6 ) 21 (6.60) 20 (6.29) 2 (0.63) 318 33 (22.92) 32 (22 22) 27 (18.75) 12 (8.33) 8 (5.56) 22 (15.28) 0 (0 0 0 ) 144 238 (30.09) 158 (1 9 9 7 ) 100 (12.64) 58 (7.33) 44 (5 5 6 ) 69 (8 72) 2 (0 2 5 ) 791 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 32. IContinued) < 10 Yrs' 10-19 Yrs1 20-29 Yrs1 30-39 Yrs1 40-49 Yrs1 50-59 Yrs1 60-69 Yrs1 70 + Yrs1 Unknown Age1 Totals Aggregate Data B aseline,3 N on-E Q -R elated 156 (11 02) 199 (15.08) 274 (20 76) 239 (18.11) 166 (12.58) 105 (7.95) 76 (5.76) 99 (7.50) 6 (0.45) 1320 P ost-E vent,4 N on-EQ -R elated 155 (9 7 5 ) 191 (1 2 0 2 ) 352 (22.15) 313 (19.70) 215 (13 53) 115 (7 24) 97 (6.10) 138 (8 68) 13 (0 82) 1589 P ost-E vent,4 E Q -R elated 27 (3.95) 46 (6.73) 122 (17.84) 166 (24.27) 122 (17.84) 72 (10 53) 47 (6.87) 75 (10 96) 7 (1 0 2 ) 684 Totals 338 (9.41) 436 (12.13) 748 (20.82) 718 (19.98) 503 (14.00) 292 (8 1 3 ) 220 (6 1 2 ) 312 (8 6 8 ) 26 (0.72) 3593 1 F requency (P ercent) 2 Incom plete: No a ss u m e d earthquake-related injuries, no b aselin e d a ta 3 Jan u ary 1-16, 1994. 4 Jan u ary 17-3 1 ,1 9 9 4 . K > OP associated with the earthquake were slightly younger than those with clearly earthquake- related injuries, with a median age of 37 years (range 1— 95 years of age). Patients treated for injuries that were assumed to be indirectly related to the earthquake were closer in age to the baseline and non-earthquake-related injured patient age, with a median age of 32 years (range 1 - 47 years of age). The ethnic distribution of injured patients treated at the four facilities for the month of January, 1994 are presented in Table 33. Hospital ‘C’ shows approximately half of the ethnicity data as missing in both time periods. Two facilities (Hospital ‘B ’ and Hospital ‘D’) show an increased proportion o f missing ethnicity data for earthquake- related injuries compared to all others. Hospital ‘A’ has no baseline data, so comparisons by time period are not possible. This facility shows a smaller proportion of earthquake-related injuries in those of Hispanic and Asian/Pacific Island ethnicity compared with non-earthquake-related injuries. All other ethnic groups show similar proportional distributions between earthquake-related and non-earthquake-related injuries. Hospital ‘C’ shows very little ethnic variation of earthquake versus non-earthquake-related injuries and baseline versus post-event injuries. Hospital C B’ shows similar ethnic distributions of non-earthquake-related injuries for both baseline and post-event time periods. This provides some support for the validity of some of the assumptions used to attribute injuries to the earthquake; one would expect demographic distributions to be similar for non-earthquake-related injuries regardless of 219 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 33. Ethnic Distributions of Injured Patients Treated at 4 Emergency Departments in Los Angeles County. January 1994. Pre- and Post Northridoe Earthauake. Facility Non-Hispanic Caucasian1 African American1 Hispanic1 Native American1 Asian / Pacific Isl1 Other1 Missing1 Totals Hospital ‘ A*2 Baseline,3 Non-EQ-Related . . . . . . . . . . . . — . . . . . . . . . Post-Event,4 Non-EQ-Related 353 69 37 0 23 0 26 528 (66.9) (16.9) (70) (0.0) (4.4) (0.0) (4.9) Post-Event,4 EQ-Related 58 15 3 0 1 2 5 84 (69.0) (17.9) (3.6) (0.0) (1.2) (2.4) (6.0) Totals 411 104 40 0 24 2 31 612 (67.2) (17.0) (6.5) (0.0) (3.9) (0.3) (5.1) Hospital 'B * Baseline,3 Non-EQ-Related 238 10 50 0 3 0 18 319 (74.6) (3.1) (15.7) (0.0) (0.9) (0.0) (5.6) Post-Event,4 Non-EQ-Related 131 6 26 0 3 0 12 178 (73.6) (3.4) (14.6) (0.0) (17) (0.0) (6.7) Post-Event,4 EQ-Related 170 1 18 0 5 0 70 264 (64.4) (6.8) (0.4) (0.0) (1.9) (0.0) (26.5) Totals 539 17 94 0 11 0 100 761 (70.8) (2.2) (12.4) (0.0) (1.4) (0.0) (13.1) N > N ) o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 33. (ContinuedI Non-Hispanic Facility__________________________Caucasian1 Hospital 'C' Baseline,3 Non-EQ-Related 89 (13.2) Post-Event,4 Non-EQ-Related 66 (117) Post-Event,4 EQ-Related 27 (14.1) Totals 182 (12.7) Hospital 'D' Baseline,3 Non-EQ-Related 194 (59.0) Post-Event,4 Non-EQ-Related 206 (64.8) Post-Event,4 EQ-Related 81 (56.3) Totals 481 (60.8) African tmerican1 Hispanic1 Native American1 Asian / Pacific Isl1 Other1 Missing1 Totals 176 28 0 8 3 368 672 (26.2) (4.2) (0.0) (1.2) (0.5) (54.8) 142 38 0 2 3 314 565 (25.1) (6.7) (0.0) (0.4) (0.5) (55.6) 44 12 0 2 2 105 192 (22.9) (63) (0.0) (10) (10) (54.7) 362 78 0 12 8 787 1429 (25.3) (5.5) (0.0) m (06) (55.1) 42 49 0 32 2 10 329 (12.8) (14.9) (0.0) (97) (0.6) (3.0) 22 47 2 27 0 14 318 (6.9) (14.8) (0.6) (8.5) (0.0) (4.4) 12 14 2 7 1 27 144 (8.3) (9.7) (14) (4.9) (0.7) (18.8) 76 110 4 66 3 51 791 (9.6) (13.9) (0.5) (83) (0.4) (6.4) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. T able 33. (C o n tin u ed I Facility Non-Hispanic Caucasian1 African American1 Hispanic1 Native American1 Asian / Pacific 1st1 Other1 Missing1 Totals Aggregate Data Baseline,3 Non-EQ-Related 521 (39.5) 228 (17.3) 127 (9.6) 0 (0.0) 43 (33) 5 (0.4) 396 (30.0) 1320 Post-Event,4 Non-EQ-Related 756 (47.6) 259 (16.3) 148 (9.3) 2 (0.1) 54 (3.4) 3 (0.2) 367 (23.1) 1589 Post-Event,4 EQ-Related 336 (49.1) 72 (10.5) 47 (6.9) 2 (0.3) 15 (2.2) 5 (0.7) 207 (30.3) 684 Totals 1613 (44.9) 559 (15.6) 322 (9.0) 4 (0.1) 112 (3.1) 13 (0.4) 970 (27.0) 3593 1 Frequency (Percent) 2 Incomplete: No assumed earthquake-related injuries, no baseline data. 3 January 1-16,1994. 4 January 17-31,1994. the time period. Hospital ‘B ’ shows a smaller proportion of injury visits for those of Hispanic and Caucasian ethnicity treated with earthquake-related injuries compared to non-earthquake-related injuries, and a larger proportion of injury visits for those of African-American ethnicity treated with earthquake-related injuries compared to non earthquake-related injuries. Hospital ‘D’ shows similar ethnic distributions of non-earthquake-related injuries by time period for those o f Hispanic and Asian/Pacific Island ethnicity with reductions in earthquake-related injuries. There were proportionally fewer African-Americans treated after the earthquake at this facility than before the earthquake, regardless of earthquake- relatedness. A larger proportion of Caucasians was treated for non-earthquake-related injuries after the event than during the baseline period and compared with earthquake- related injuries at Hospital ‘D \ Aggregate data for three hospitals shows proportional increases in post-event versus baseline frequencies for treatment of injuries to patients of Caucasian ethnicity. Additionally, the proportion of non-earthquake-related injuries was similar regardless of time period for patients of African-American and Hispanic ethnicity. However, in the aggregate data, a proportional decrease was noted in the treatment of earthquake-related injuries versus non-earthquake-related injuries for those of African-American and Hispanic ethnicity. Other ethnic groups showed no remarkable variation between time period and earthquake-relatedness. 223 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The external cause of injury for patients treated at the four facilities is presented in Table 34. There was a noticeable increase in unreported mechanisms of injury at three facilities for earthquake-related injuries compared to non-earthquake-related injuries. In general, there appeared to be fewer earthquake-related injuries caused by motor-vehicle collisions, firearms, suicide attempts, falls, poisonings, environmental exposures, fires and bums, and being caught in or between objects compared to non-earthquake-related injuries. More earthquake-related injuries appeared to be due to being struck by objects or cutting and piercing mechanisms than injuries not attributed to the earthquake. 1.03. Geoeraphic Comparison of Injured Patients bv Time Period. A map prepared by EQE Center for Advanced Planning and Research showing the distribution of injuries treated at the 4 facilities (from January 17-31, 1994) by earthquake-relatedness is presented in Figure 18. Most of the treated injuries cluster around the facilities, and are more sparsely distributed around the epicenter. It is also apparent that most of the injuries are non-earthquake-related. In order to compare the geographic distribution of injuries after the earthquake with baseline injuries, a map showing injuries treated at the facilities for the baseline time-period (January 1-16, 1994) is presented in Figure 19. The clustering about the facilities is still noted, but does not appear as dense during the baseline time period. There is also a sparse distribution of injuries throughout the San Fernando Valley in the area of what would become the epicenter of the Northridge earthquake. It is notably more sparse than injuries recorded after the earthquake (compare Figure 18 to Figure 19). 224 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 34. External Mechanisms ol Inlurv lor Patients Treated at 4 Emergency Departments In Los Angeles County. January 1994. Pre- and Post-Northrldae Earthquake. f i r e a r m / S e x u a l A s s a u l t / N o n - P o i s o n i n g s I C a u g h t i n o r F a c i l i t y M o t o r V e h i c l e 1 S p e c i f i c A s s a u l t 1,2 F a l l s 1 E n v i r o n m e n t a l E x p o s u r e 1,1 C u t t i n g / P i e r c i n g 1 S t r u c k b y O b j e c t 1 b e t w e e n O b j e c t 1 F ir el B u r n 1 O v e r e x e r t i o n 1 O t h e r 1 4 N o t R e p o r t e d 1 T o t a l s H o s p i t a l 'A'5 B a s e l i n e , ' N o n - E Q - R e l a t e d . . . . .. . . . . .. . .. . .. . . . P o s t - E v e n t / N o n - E Q - R e l a t e d 8 0 1 5 1 4 6 4 6 7 4 6 4 7 1 3 1 4 2 0 4 9 5 2 8 ( 1 5 2 ) ( 2 8 ) ( 2 7 7 ) ( 8 . 7 ) ( 1 4 0 ) ( 1 2 . 1 ) ( 1 . 3 ) ( 2 5 ) ( 2 7 ) ( 3 8 ) ( 9 3 ) P o s t - E v e n t / E Q - R e l a l o d 0 0 2 7 5 1 3 2 6 0 0 6 3 4 8 4 ( 0 0 ) ( 0 . 0 ) ( 3 2 . 1 ) ( 6 . 0 ) ( 1 5 . 5 ) ( 3 1 0 ) ( 0 0 ) ( 0 0 ) ( 7 1 ) ( 3 6 ) ( 4 8 ) Totals SO 15 173 51 8 7 90 7 13 20 23 53 612 (13 t) (2 5) (28 3) (83) (142) (14.7) (1 D (2 1) (3 3) (3 8) (8 7) H o s p i t a l 'B ' B a s e l i n e , 1 N o n - E Q - R e l a t e d 5 2 4 8 4 3 2 4 1 4 5 5 5 6 2 6 1 9 3 1 9 ( 1 6 3 ) ( 1 . 3 ) ( 2 6 . 3 ) ( 1 0 . 0 ) ( 1 2 . 9 ) ( 1 4 1 ) ( 1 . 6 ) ( 1 6 ) ( 1 9 ) ( 8 . 2 ) ( 6 . 0 ) P o s t - E v e n t / N o n - E Q - R e l a t e d 5 0 4 4 1 1 4 1 7 1 7 2 5 7 1 1 1 0 1 7 8 ( 2 8 . 1 ) ( 2 . 3 ) ( 2 3 . 0 ) ( 7 9 ) ( 9 6 ) ( 9 6 ) (1 1> ( 2 8 ) ( 3 9 ) ( 8 . 2 ) ( 5 6 ) P o s t - E v e n t , ' E Q - R e l a t e d 1 0 0 4 7 6 7 3 7 7 1 1 4 7 3 8 2 6 4 ( 3 8 ) ( 0 . 0 ) ( 1 7 . 8 ) ( 2 . 3 ) ( 2 7 . 7 ) ( 2 9 . 2 ) ( 0 . 4 ) ( 0 4 ) ( 1 - 5 ) ( 2 7 ) ( 1 4 . 4 ) Totals 112 8 172 5 2 131 139 8 11 17 44 6 7 7 6 1 (147) (1.1) (22 6) (6 8) (17.2) (18.3) (1.1) (14) (2.2) (5 8) (8.8) H o s p i t a l ' C B a s e l i n e , 1 N o n - E Q - R e l a t e d 7 3 1 7 1 9 6 4 6 9 6 8 8 2 2 1 4 1 4 7 3 3 3 6 7 2 ( 1 0 . 9 ) ( 2 5 ) ( 2 9 . 2 ) ( 6 9 ) ( 1 4 . 3 ) ( 1 3 . 1 ) ( 3 3 ) ( 2 1 ) ( 2 1 ) ( 1 0 . 9 ) ( 4 9 ) P o s t - E v e n t , ' N o n - E Q - R e l a t e d 7 7 2 1 1 5 5 3 7 8 6 8 2 1 0 1 2 1 7 4 7 2 1 5 6 5 ( 1 3 . 6 ) ( 3 7 ) ( 2 7 . 4 ) ( 6 . 6 ) ( 1 5 . 2 ) ( 1 4 5 ) ( 1 8 ) ( 2 1 ) ( 3 . 0 ) ( 8 3 ) ( 3 . 7 ) P o s t - E v e n t , ' E Q - R e l a t o d 0 0 4 0 1 5 6 6 2 0 0 2 6 2 5 1 9 2 ( 0 . 0 ) ( 0 0 ) ( 2 0 . 6 ) ( 0 5 ) ( 2 9 . 2 ) ( 3 2 . 3 ) ( 0 . 0 ) ( 0 0 ) ( 1 . 0 ) ( 3 . 1 ) ( 1 3 0 ) T o t a l s 150 3 8 391 84 238 232 32 2 6 33 126 79 1429 (10.5) (2.7) (27.4) (5.9) (16.7) (16.2) (2.2) (18) (2.3) (8.8) (5.5) K » C A Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Tibia 34. (Continued) F a c i l i t y M o t o r V e h i c l e ’ F i r e a r m / S e x u a l A s s a u l t / N o n - S p e c i f i c A s s a u l t ’ '2 F a l l s ' P o i s o n i n g s 1 E n v i r o n m e n t a l C u t t i n g 1 E x p o s u r e ’ ’2 P i e r c i n g ’ S t r u c k b y O b j e c t ’ C a u g h t i n o r b e t w e e n O b j e c t ’ F i r e / B u r n ’ O v e r e x e r t i o n ' O t h e r ’ * N o t R e p o r t e d ' T o t a l s ' H o s p i t a l 'D ' B a s e l i n e * N o n - E Q - R e l a t e d 4 6 ( 1 4 . 6 ) 1 2 ( 3 . 7 ) 6 2 ( 2 4 . 9 ) 3 0 ( 9 . 1 ) 3 7 ( 1 1 . 3 ) 3 5 ( 1 0 . 6 ) 1 3 ( 4 0 ) 4 ( 1 2 ) 1 3 ( 4 . 0 ) 4 2 ( 1 2 . 8 ) 1 3 ( 4 0 ) 3 2 9 P o s t - E v e n t / N o n - E Q - R e l a t e d 4 4 ( 1 3 . 8 ) 1 3 ( 4 . 1 ) 9 1 ( 2 8 6 ) 2 8 ( 8 8 ) 4 6 ( 1 4 . 5 ) 4 2 ( 1 3 2 ) 5 ( 1 . 6 ) 3 ( 0 . 9 ) 9 ( 2 8 ) 2 4 ( 7 6 ) 1 3 ( 4 1 ) 3 1 8 P o s t - E v e n t , ' E Q - R e l a t e d 0 ( 0 0 ) 0 ( 0 0 ) 4 4 ( 3 0 6 ) 1 ( 1 7 ) 2 6 ( 1 8 . 1 ) 3 0 ( 2 0 . 8 ) 5 ( 3 5 ) 1 ( 0 7 ) 5 ( 3 5 ) 1 8 ( 1 2 5 ) 1 4 ( 9 7 ) 1 4 4 Totals 9 2 {"■6) 2 5 (3.2) 2 1 7 (27.4) 5 9 ( 7 . 5 ) 109 (136) 107 (13.5) 2 3 (2.9) 0 (1.0) 27 (3 4) 64 (10.6) 40 (31) 791 A g g r e g a t e D a t a B a s e l i n e , * N o n - E Q - R e l a t e d 1 7 3 ( 1 3 1 ) 3 3 ( 2 . 5 ) 3 6 2 ( 2 7 . 4 ) 1 0 8 ( 8 . 2 ) 1 7 4 ( 1 3 2 ) 1 6 8 ( 1 2 . 7 ) 4 0 ( 3 . 0 ) 2 3 ( 1 7 ) 3 3 ( 2 5 ) 1 4 1 ( 1 0 . 7 ) 6 5 ( 4 . 9 ) 1 3 2 0 P o s t - E v e n t , ' N o n - E Q - R e l a t e d 2 5 1 ( 1 5 . 8 ) 5 3 ( 3 3 ) 4 3 3 ( 2 7 . 2 ) 1 2 5 ( 7 9 ) 2 2 3 ( 1 4 0 ) 2 0 5 ( 1 2 . 9 ) 2 4 ( 1 . 5 ) 3 3 ( 2 . 1 ) 4 7 ( 3 . 0 ) 1 0 2 ( 6 . 4 ) 9 3 ( 5 . 9 ) 1 5 8 9 P o s t - E v e n t , ' E Q - R e l a t e d 1 0 ( 1 . 5 ) 0 ( 0 . 0 ) 1 5 8 ( 2 3 . 1 ) 1 3 ( 1 . 9 ) 1 6 8 ( 2 4 . 6 ) 1 9 5 ( 2 8 5 ) 6 ( 0 . 9 ) 2 ( 0 . 3 ) 1 7 ( 2 . 5 ) 3 4 ( 5 . 0 ) 8 1 ( 1 1 8 ) 6 8 4 Totals 434 ( 1 2 .1) 0 6 ( 2 . 4 ) 9 5 3 ( 2 6 . 5 ) 246 (6.6) 5 6 5 ( f 3 . 7 ) 5 6 0 ( f 3 . » ) 70 (1.9) 5 0 (1-6) 97 (2.7) 277 (7.7) 2 3 9 (6.7) 3 5 9 3 ’ F r e q u e n c y ( P e r c e n t ) 2 I n c l u d e f i r e a r m I n j u r i e s , s e x u a l a s s a u l t s , a n d r a p e s , a l s o n o n - s p e c i f i c s u i c i d e a n d h o m i c i d e a t t e m p t s , c h i l d a b u s e , e l d e r a b u s e , a n d d o m e s t i c v i o l e n c e . I n c l u d e e x p o s u r e s t o t o x i c c h e m i c a l s , b i t e s & s c r a t c h e s ( a n i m a l , h u m a n , I n s e c t , r e p t i l e , a m p h i b i a n ) , u n i n t e n t i o n a l I n g e s t i o n , i n h a l a t i o n , o r a b s o r p t i o n o f t o x i c s u b s t a n c e s , e n v i r o n m e n t a l - r e l a t e d I n j u r i e s I n c l u d i n g i n j u r i e s f r o m l a n d s c a p e . * I n c l u d e s d r o w n l n g s / n e a r - d r o w n l n g s , s l i p p i n g / t r i p p i n g ( u n r e l a t e d t o f a ll) , e a r t h q u a k e - r e l a t e d ( n o t f u r t h e r s p e c i f i e d ) 9 I n c o m p l e t e D a t a : N o a s s u m e d e a r t h q u a k e - r e l a t e d i n j u r i e s , n o b a s e l i n e d a t a . * J a n u a r y 1 - 1 6 , 1 9 9 4 . ' J a n u a r y 1 7 - 3 1 , 1 9 9 4 . 227 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. _ c <» £ f •o ■o « c 5 g"= SO S l l s f i f 3 o 3 £ ? I 3 3 o | 3^. s.? f s as 51 T O II I1 ?? o « :5 s So 3 If 2 iS ? 0 % cn / 228 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. There is a relatively wide dispersion between the three baseline earthquake- related injuries (one in the West Service Planning Area (SPA), one in the South SPA, and one in the South Bay/Long Beach SPA). There was a small seismic event 5-10 miles off the coast of Santa Monica on January 9, 1994. If these injuries were truly related to that event, it is worthwhile to note the distance between these locations and the event. Two o f them are at least 15 miles from the epicenter of that small event. 2.00. DESCRIPTIVE CHARACTERISTICS OF THE STUDY SAMPLE. Demographic and injury characteristics of the patients treated for earthquake- related injuries at the four facilities will be presented by injury severity group (minor, moderate and serious) in the following sections. Maps of estimated MMI and PGA will also be presented in this section. Building structure characteristics and estimates of geologic factors have already been presented in Chapter IV, Sections 5.01 and 5.02. 2.01. Demographic Characteristics of Injured Patients. Demographic characteristics are displayed by Injury Severity Score grouping (minor, moderate or serious) in Table 35. As expected, the age distribution varies by injury severity. The age group with the greatest proportion of minor (84% of 160) and moderate (13% of 160) injuries (combined) is 30-39 year-olds. However, the age group with the most serious injuries included those patients aged 70 or older. Twenty percent of the 66 injuries in patients aged 70 or older were serious, and 20% were of moderate severity. 229 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 35. Dem ographic Characteristics (Includin g M<««q o Values) o f Patients Injured from th e N orthridae Earthquake th at P resented a t 4 E m ergency Departments in L os A n geles County. January 17-31. 1994. Number (%) of Earthquake-Related Injuries by Level o f Injury Severity Dem ographic C haracteristics Minor1 Moderatei* Serious3 Totals Age < 1 0 Years 25 (93) 1 (3-5) 1 (3.5) 27 10-19 Years 36 (88) 4(10) 1 (2) 41 20-29 Years 100(88) 12(10) 2 (2 ) 114 30-39 Years 135(84) 21 (13) 4 (3 ) 160 40-49 Years 90(79) 20(18) 4 (3 ) 114 50-59 Years 51 (77) 11 (17) 4 (6 ) 66 60-69 Years 33(72) 9(19) 4 (9 ) 46 2 70 Years 40 (60) 13(20) 13 (20) 66 Unknown Age 7 (100) 0 0 7 Ethnicitv (Including Unknown) Caucasian 239 (76) 53 (17) 21 (7) 313 African-American 57 (85) 9(13) 1 (2) 67 Hispanic 32 (80) 4(1 0 ) 4(1 0 ) 40 Other4 18 (90) 2(10) 0 20 Unknown 171 (85) 23 (11) 7 (4 ) 201 Gender Female 274(77) 60 (17) 2 3 (6 ) 357 Male 243 (86) 31 (11) 10(3) 284 Hospital Hospital 'A ' 48 (63) 22(29) 6 (8 ) 76 Hospital 'B' 200 (79) 36 (14) 18(7) 254 Hospital 'C ' 172(92) 14(7) 1 (1) 187 Hospital 'D ' 97 (78) 19 (15) 8 (7 ) 124 Totals 517 (81) 91 (14) 33(5) 641 1 1njury Severity Scores 1-3 2 Injury Severity Scores 4-8 3 Injury Severity Scores > 9 4 Egyptian, Iranian, Persian, Armenian, Ethiopian, and Non-Categorizable 5 No assumed earthquake-related injuries (no baseline data) 230 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Nearly one-third of the ethnicity information was missing, reducing the sample size to 440. Based on available ethnicity information, patients of Caucasian ethnicity represented the majority of the injuries (71%), followed by African-American (15%), and those of Hispanic ethnicity (9%). Five percent of the patients treated for earthquake- related injuries were of Egyptian. Iranian, Persian, Armenian, Ethiopian, and ethnicities that did not fit any category. Within ethnicities, African-Americans had the largest proportion of minor injuries; 85% of all identified earthquake-related injured African- Americans (n=67) were treated for minor injuries. Those of Hispanic ethnicity had the largest proportion of serious injuries; 10% of all identified earthquake-related injured Hispanics (n=40) were treated for serious injuries. Females represented a larger proportion of injuries (56%) than males. Within genders, females showed slightly increased proportions of moderate (17% versus 11%) and serious injuries (6% versus 3%) than males. This is an unexpected finding, since males usually tend to seek care for injuries at emergency departments more than females, a finding confirmed by reviewing activity for the baseline time-period. 2.02 Iniurv Characteristics of Study Patients. Available injury characteristics included body location (n=637) and mechanism of injury (n=512). Other injury characteristics were collected including whether an individual was extricated from an area of entrapment, intentionality of the mechanism o f injury (unintentional, intentional, unknown intent), whether the injury was related to employment, and the use of the property at the injury scene. Elements other than 231 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. mechanism and body location of injury were not included in the analysis since they were not considered risk factors for severity o f injury in this particular earthquake. Most of the injuries did not require extrication (95% of 641 earthquake-related injuries), were not work-related (91%), and were unintentional (92%). Categories and distributions for injury characteristics of patients identified with earthquake-related injuries are displayed in Table 36. Ninety-two percent of the injuries were unintentional. Two injuries were self- inflicted injuries, two were related to assaults, eight were of undetermined intention (the information was ambiguous regarding the intent), and no information regarding intention was available on the remainder (n=40). The two self-inflicted injuries occurred in conjunction with pre-existing psychiatric problems. Only five patient records indicated that the patient had been extricated from the scene of the injury, and 25 records were ambiguous regarding whether the patient was extricated. It was assumed that the remainder of the patients (n=611) did not require extrication. Four percent o f the injuries (n=23) were job-related. Twenty-two percent (n=5) of the job-related injuries occurred in hospitals. This is not surprising considering the source of the data. One work-related injury occurred on a freeway, one at a school, one at a warehouse, one at a construction site, five (22%) occurred at unknown locations. The other seven work-related injuries occurred at locations other than any of the previous mentioned groupings. Five percent (n=32) of the injuries might have been related to the patient’s work, but there was insufficient information in the medical record to be able to 232 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 36. Injury C h a ra cteristics o f P a tien ts T rea ted a t 4 E m e rg e n c y D e p a rtm en ts in L os A n g elo s C ounty. Jan u ary 17-31. 1994. w ith E a rth q u a k e-R ela ted Injuries. Number (% ) of Earthquake-Related Injuries ____________ by Level of Injury Severity________ Injury Characteristics__________________________ Minor1 Moderate2 Serious* Total Body Location Head & Neck Upper Extremities Lower Extremities Back Chest Abdomen T runk Non-Specific Unknown External Cause (Mechanism) of Inlurv Motor-Vehicle Fall Poisonings, Exposures, Bites / Stings I Scratches Cutting or Piercing Mechanism Struck by Object Caught in or between Object(s) Fire / Burn / Electrocution Overexertion Slipping / Tripping, not related to a fall Earthquake-Related, Non-Specific Extrication Extricated from In ju ry Location Not Extricated Unknown / Ambiguous Information 1 5 5 (8 6 ) 1 7 (9 ) 9 (5 ) 181 1 0 6 (7 3 ) 32 (22) 8 (5 ) 146 225 (83) 37 (13) 1 0 (4 ) 272 1 7 (7 1 ) 3 (1 2 ) 4 (1 7 ) 24 9 (9 0 ) 0 1 (1 0 ) 10 1 (3 3 ) 1 (3 3 ) 1 (3 3 ) 3 0 1 (100) 0 1 1 (100) 0 0 1 3 (1 0 0 ) 0 0 3 8 (8 9 ) 0 1 (1 1 ) 9 94 (63) 39 (26) 1 7 (1 1 ) 150 5 (7 1 ) 2 (2 9 ) 0 7 1 5 5 (9 2 ) 1 1 (7 ) 2 (1 ) 168 166 (88) 1 8 (9 ) 5 (3 ) 189 3 (6 0 ) 1 (20) 1 (2 0 ) 5 0 0 2 (1 0 0 ) 2 7 (5 0 ) 5 (3 6 ) 2 (1 4 ) 14 1 5 (6 5 ) 7 (3 1 ) 1 (4 ) 23 64 (86) 8 (1 1 ) 2 (3 ) 74 2 (4 0 ) 1 (2 0 ) 2 (4 0 ) 5 501 (82) 8 2 (1 3 ) 2 8 (5 ) 611 14 (56) 8 (3 2 ) 3 (1 2 ) 25 N > U ) u > Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 36. (Continued) Number (% ) of Earthquake-Related Injuries by Level of Injury Severity Injury Characteristics Minor1 Moderate2 Serious1 Total Intention of In iu rv Unintentional 471 (80) 87(15) 30 (5) 588 Intentional, Self-Inflicted 2(100) 0 0 2 Intentional, Inflicted by Other 1(50) 0 1(50) 2 Undetermined Intent 7(88) 0 1 912) 8 Unknown 36 (88) 4(10) 1(2) 41 Work-Related In iu rv Work-Related 22 (96) 1(4) 0 23 Assumed Not Related to W ork 471 (80) 83(14) 32 (6) 586 Insufficient Information to Attribute to W ork 24 (75) 7(22) 1(3) 32 Use of Property at Scene of Iniurv Home 228 (80) 43(15) 14(5) 285 Residential Institution 0 2(67) 1(33) 3 Public Building 2(100) 0 0 2 Street or Sidewalk 2(40) 1 920) 2(40) 5 Freeway, Highway, Thoroughfare 3(75) 0 1(25) 4 Hospital 5(100) 0 0 5 School 1(50) 1(50) 0 2 Factory or Warehouse 1 (100) 0 0 1 Industrial or Construction Site 1 (100) 0 0 1 Other4 12(80) 0 3(20) 15 Unknown 262 (82) 44(14) 12(4) 318 Totals 517(81) 91 (14) 33(5) 641 11 n ju ry Severity Scores 1-3 2 In ju ry Severity Scores 4-8 3 In ju ry Severity Scores a 9 includes work-place (not specified further); parking lot or structure; grocery / retail / drug store; housing shelter; outside (not specified further), intersection, park, mobile home clearly attribute these to work-related incidents. It was assumed that the remaining 91% of the patients did not have work-related injuries. Work-related injuries were systematically documented in the medical record differently than other injuries. When the injury was clearly identified as work-related, an incident report sheet was generated at the emergency department and was completed by the attending physician. This report detailed the events surrounding the injury and usually provided more detailed incident information than what was usually recorded. The more detailed documentation in work- related injuries could lead to differential validity of the incident-specific information compared to injuries that were not related to work. Therefore, the difference between those two groups (work-related injuries and those not related to work) may be lessened, obscured or increased, or observed when there really is no difference (Abramson, 1990). Fifty percent of the records associated with earthquake-related injuries lacked information on the use of the property at which the injury occurred. Forty-five percent of the injuries occurred in residences, and the remaining 4% fell into a variety of categories (residential institution, public building, street/sidewalk, freeway/highway, hospital, school, factory/warehouse, industrial/construction, parking lot/structure, grocery/retail/drug store, housing shelter, outside, street intersection, park, mobile home, nonspecific work-place). The largest proportion of earthquake-related injuries were reported to have impacted the patients’ lower extremities (42%), followed by head and neck injuries (28%), upper extremity injuries (23%), and back injuries (4%). The body location grouping that showed the largest proportion of serious injuries (16%) was the back or 235 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. spine (4 of 24 back/spine injuries). The body location group that showed the largest proportion of minor injuries (86%) was the head and neck (155 of 181 head/neck injuries). The largest proportion of earthquake-related injuries was reported to have been caused by striking mechanisms (i.e., being struck by an object or objects) (30%), followed by cutting and piercing mechanisms (26%), falls (23%), and non-specific earthquake-related injuries (12%). The largest proportion of serious injuries (11%) was caused by falls (17 o f 150 falls). The largest proportion of minor injuries (92%) was caused by cutting and piercing mechanisms (155 of 168 cutting/piercing ). 2.03. Geographic Distribution of Earthquake-Related Injuries bv Estimated Ground-Shaking and Intensity. Maps created by EQE showing the geographic distribution o f injuries by MMI and PGA are presented in Figures 20-25. Figure 20 shows the distribution of injuries by MMI for the most heavily-impacted region o f Los Angeles County. The total area of Los Angeles County is 4089 square miles. Fifty-three percent of that area was in MMI=V or less, and is not the focus of the map. The area presented on the map concentrates on locations of MMI £ VI. Fifty-three percent o f this area was in MMI=VI, 33% was in MMI=VII, 14% was in MMI=VIII, and less than 1% was in MMI=IX. A total o f 226 (out of 641) earthquake-related injuries were matched to the geologic database. Eight percent o f those that matched were in MMT=VT, 49% were in MMI=VII. 41% were in MMI=VIII, and 2% were in MMI=IX. Figure 21 shows a regional enhancement focusing on the San Fernando Valley. The enhanced maps facilitate identification of 236 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 237 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 238 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 239 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. levels of earthquake-attributability (clearly earthquake-related, assumed directly earthquake-related, and assumed indirectly earthquake-related). Figure 22 shows a regional enhancement focusing on Los Angeles’ West Side and South Bay areas. Figures 23-25 show PGA contours that were provided courtesy of Dave Wald, U.S. Geological Survey, and were incorporated into maps generated by EQE Center for Advanced Planning and Research. Concentric contours usually represent decreasing seismic activity relative to the center. However, one must look closely to ensure that this is the case, since there are central pockets of lowered seismic activity as well. Figure 23 shows the distribution of injuries by PGA for the most heavily- impacted region o f Los Angeles County. It is interesting to note that the epicenter was not associated with a central area of high ground shaking (PGA £ 80% g). However, there were three pockets of high ground shaking North of the epicenter (2 North-East and I North-West), and an additional pocket of high ground shaking in Santa Monica. Figure 24 shows a regional enhancement of the map focusing on the San Fernando Valley. This map shows more clearly the three pockets of high ground shaking in the San Fernando Valley, all North of the epicenter. Additionally, the earthquake- related injuries (denoted as triangles and diamonds in the figure) and the hospitals that treated those injuries can be seen more clearly in the enlarged map. It is worth noting that although there is a clustering of injuries around Hospital ‘ B ’ , earthquake-related injuries that occurred near the epicenter and in areas of higher ground shaking were seen at that hospital. 240 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. S a n F e rn a n d o ^ n te lo p o Valloy 1 0 & S a n G a b r i e l P asadena t « a ^ ie tro Figure 23. Injurlos Recordod al Emorgoncy Departments botwoon 1/17 and 1/31 following tho 1904 NorlhrkJgo Earthquako (Mw 6 7 ) and Poak Ground Acceleration Contours (%g| Regional Extent ^ E piconlor Z ip co d e C onlroid of Hospital hospital A (n*305, Incomplole) c Q j Hospital 0 (n-210) Hospital C (n«4IO) HospllAl D (n-265) Color corresponds lo Hospital S hape Indicates sourco of Injury aEwTR|uafe«(«M M (t;/) •N o t tiu a n o d 10 bo MiTtquofco leW ed (M l) •AMgmod lo bo fOQIW (HI) to bo rxfcatl>r m i (6 ) PGA contour data coudosy ot Dava WaW. U S Gaotogcal Survey CoropWod from COMO, lAOWP, SCC. USC, and USGS acceferogiaph racordngs Center lor Advanced Planning and Research 4590 MacArthur Blvd, Suite 400 Newport Beach, CA 92660-2027 P hone 949-833-3303 1 2 /U M F ig u ie 24 , Injunos Hm>rdod at Hmofijflncy Departments betw een 1 /1 / a n d 1/31 following Iho 1 9 9 4 Northndgo FmthquaKo (Mw 6 /) and PoaH OiiMind Acceleration Contours (%g) San Tornando Gitont "jf f-piccntcf 2rp co d o C entroid nl H ospital HosjHtnl A (n -3 0 5 , incom plete) H ospital (1 (n -2 1 0 ) ^ 1 H ospital C (n -4 1 0 ) H ospital D (n -2 6 5 ) C olor c o rre sp o n d s to HospilAl S h a p e in d icates so u rc e ol injury 4Canhgw«k« iMi«l(irr) • » « tttu m o d ia M M flbqu«k« rtfaiod (M l) • A ttianw ] to t - M H ftyjoM ItM lM |? 6J) •A ttum «4 to bo kxfe-cOy MrtbQuoko loUiod (I) 1*04 contour (Mt4 courlety ol Olv* WakJ, I) S Geolwjcol Survey Compilet) trorn COMO. IADW11 , SCE. IJ5C. «nil IJSG5 iccetorogrtph • ea r n in g s Center lor Advanced Wanning and Rosoatch 4590 MacArthur nivri , Smio 400 Newport Ooach. CA 92660 2027 Phono 949 033 3303 1 * 7 V M Figure 25 shows a regional enhancement that focuses on Los Angeles’ West Side and South Bay areas. This map shows more clearly the pocket of high ground shaking in Santa Monica. Additionally, the earthquake-related injuries (denoted as triangles and diamonds in the figure) and the hospitals that treated those injuries can be seen more clearly. It is worth noting that earthquake-related injuries sustained near the epicenter and in areas of high ground-shaking were seen at Hospitals ‘ A \ C \ and 'D '. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 243 « n O o 55 I 111 244 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I U M I 3.00. LOGISTIC REGRESSION: IDENTIFICA TION OF FACTORS ASSOCIATED WITH MODERATE AND SERIOUS INJURY. Demographic, injury, building structure, and geologic characteristics were each considered separately with respect to the grouped injury severity score. Polytomous logistic regression considered a three-level grouping of injury severity (minor = ISS < 3, moderate = ISS 4-8, and serious = ISS > 9). Age, gender, ethnicity of the injured patients, and facility type were identified a priori as potential confounding variables. Therefore, these variables were included in all other models to control for potential confounding bias. As components of the multivariate model were merged and the effective sample size was reduced (due to missing data), the power to detect an association was also reduced. Table 37 presents the minimum detectable odds ratios corresponding to various sample sizes and different referent exposure prevalence values. The table is based on a two-sided test, a=0.05 (Type I error rate) and (3=0.20 (Type II error rate). The power to detea any given association in the table is 80%. Since the referent exposure prevalence to risk factors is unknown, a range of values is presented. Based on no loss of data, the sample size of 641 shows that regardless of the exposure prevalence, the minimum deteaable odds ratio is less than 2.0. As the sample size decreases, the minimum deteaable odds ratio increases, especially at the extreme ranges of exposure prevalence (less than 0.3 and greater than 0.7). Therefore, it is possible that certain associations 245 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 37. Minimum Detectable1 Odds Ratios for Various Sample Sizes Based on Selected Exposure Prevalence Values. Sample Size Referent Exposure Prevalence 0.10 0.20 0.30 0.40 0.50 0.60 0.07 0.80 0.90 641 1.63 1.46 1.40 1.38 1.38 1.40 1.44 1.54 1.86 512 1.73 1.53 1.46 1.43 1.43 1.45 1.51 1.63 2.04 447 1.79 1.57 1.50 1.47 1.47 1.50 1.56 1.70 2.17 418 1.82 1.60 1.52 1.49 1.49 1.52 1.58 1.73 2.25 330 1.95 1.69 1.60 1.57 1.57 1.61 1.69 1.88 2.59 259 2.12 1.81 1.70 1.67 1.67 1.72 1.82 2.07 3.10 196 2.35 1.97 1.84 1.80 1.81 1.88 2.02 2.38 4.11 88 3.41 2.71 2.50 2.46 2.52 2.72 3.20 4.75 undefined 1 a = 0.05, p = 0.20, 2-sided K > * betw een risk factors and injury severity might not be detected as statistically significant, given the worst case scenario (e.g., smallest sample size). The remaining sections in this chapter first describe the results of the likelihood ratio tests (LRT) associated with the model selection process for each variable group (patient demographics, injury characteristics, structural and geologic components). The tables reflecting the LRT results are included in Appendix 7 . Each section will also present the resulting model fit with associated parameter estimates (odds ratios) and confidence intervals. The final section of this chapter describes the full model that includes variables identified from each variable group with associated parameter estimates and confidence intervals. 3.01. Demographic Characteristics with Respect to Iniurv Severity Score. Polytomous Logistic Regression. Only records with values for all demographic variables (n=418) were included in the polytomous demographic model since patients with missing information may be misrepresented if grouped together, and this study does not involve imputing values for missing data. Reasonable interaction terms were tested for statistical significance by comparing hierarchical models through the likelihood ratio test. These interaction terms included gender-ethnicity interactions, age-gender interactions and age-ethnicity interactions. Frequencies o f demographic characteristics for patients included in the polytomous logistic regression are presented by level o f injury severity in Table 38. 247 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 38. Pofytom ous Logistic R egression: A ssociations betw een D emographic Characteristics of Inlured P a tien ts1 and Severity o f Iniurv from the Northrldae Earthquake (n°418). Number (%) of Earthquake-R elated Injuries C haracteristic Minor4 M oderate9 Serious* Totals (95% Confidence Interval) (95% C onfidence In Male 137 (82) 21 (13) 9(5) 167 (Referent) (Referent) Female 189(75) 45(18) 17(7) 251 1 48 (0 91-2 41) 1 21 (0 73-2 01) Caucasian 238 (76) 53(17) 21 (7) 312 (Referent) (Referent) African-American 56(85) 9(14) 1(2) 66 0 56 (0 2 7 - 1.15) 0 5 6 (0 2 6 -1 19) Hispanic 32 (80) 4(10) 4(10) 40 0 8 5 (0.38-1 90) 1 31 (0 57 - 3.04) Less than 20 Years 36(88) 3(7 ) 2(5) 41 0 6 7 (0.23-1.92) 0 8 4 (0.28 - 2.49) 20-29 Years 69(86) 9(11) 2(3) 80 0.75 (0.33-1 68) 0 76 (0.33-1.73) 30-39 Years 88(82) 16(15) 3(3) 107 (Referent) (Referent) 40-49 Years 50(71) 16 (23) 4(6) 70 1.83 (0.89 - 3.75) 1.93 (0.92 - 4.03) 50-59 Years 31 (76) 8(20) 2(4) 41 1.49 (0.63 - 3.55) 1.70 (0.69-4.16) l - 60 Years 52(66) 14(18) 13(16) 79 2 6 9 (1 3 8 - 5 27) 2 6 9 (1.34-5.40) Age Group Trend7 1.34 (1.16-1.56) 1.33 (1.14-1.55) Gender Ethnicity Aoe to •u 00 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 38. fContinued). Number (%) of Earthquake-Related Injuries by Level of Injury Severity Univariate Odds Ratio3 Adjusted Odds Ratio3 ,9 Characteristic Minor4 Moderate9 Serious8 Totals (95% Confidence Interval) (95% Confidence Interval) Hospital Hospital 'A '8 43 (63) 19(28) 6(9) 68 272 2 05 Hospital 'B' 141 (77) 28 (15) 13(7) 182 (1 51 - 4 87) (Referent) (1 55-5.24) (Referent) Hospital 'C' 75 (96) 3(4) 0 78 (Referent) (Referent) Hospital 'D' 67 (74) 16(18) 7(8) 90 1 69 1 58 Totals 326 (78) 66 (16) 26(6) 418 (0 95 - 2 98) (0 88 - 2 83) ' P a t i e n t s t r e a t e d vrk h e a r t h q u a k e - r e l a t e d I n ju r ie s a t 4 em ergency d e p a r t m e n t s a lt e r t h e N o r t h r ld g e E a r t h q u a k e , L o t A n g e l e s C o u n t y , J a n u a r y 1 7 - 3 1 , 1 9 9 4 . 2 S e v e r e : M o d e r a t e In ju ry ; M o d a r a te :M ln o < In ju ry . * A d j u t l a d for aH o t h e r d e m o g r a p h i c v a r i a b l e s 4 I n ju r y S e v e r i t y S c o r e s 1 - 3 A In ju r y S e v e r i t y S c o r e s 4 8 In ju r y S e v e r i t y S c o r e s * 9 ’ O d d s r a t io s ( p e r a g e group) r e p r e s e n t e s t i m a t e s o f r e l a t h e r is k fo r m o r e s e r i o u s In ju ry a s a g a g r o u p I n c r e a s e s . 8 N o a s s u m e d e a t l h q u a k e - t e l e l e d I n ju r ie s ( n o b a s e l i n e d a t e ) . __________________________________________________________________ -U < o Ethnicity was grouped into 3 categories for analysis: Caucasian, African-American, and Hispanic. The ethnicity category o f ‘Other’ (n=20, no serious injuries) was deleted, and patients of Caucasian ethnicity served as the reference category. The two lowest (ages 0-9 and 10-19) and highest age categories (ages 60-69 and those aged 70 or older) were combined due to sparse observations. Resulting categories were those less than age 20, 20-29, 30-39, 40-49, 50-59, and those aged 60 or older. The reference category for age was 30-39 year-olds, since this group had the largest number o f injuries. Hospitals C B’ and ‘C’ served as the reference facilities for Hospitals ‘A’ and ‘D’ (Level I Trauma Centers) since Hospitals ‘A’ and ‘D’ were expected to treat the most serious injuries. Furthermore, Hospital ‘C’ treated no serious injuries, therefore this hospital was not considered an appropriate reference group, and it was combined with Hospital ‘B’ for modeling purposes. A summary table of the development of the polytomous demographic model with pertinent likelihood ratio tests is presented in Appendix 7, Table A 7-1. Although only records with values for all demographic variables (n=418) were included in the polytomous demographic model, prior to data reduction, gender (n=641) and age group (n=634) were univariately (unadjusted) associated with injury severity. No statistically significant interactions were noted between gender and ethnicity, age and gender, and age and ethnicity. The assumption for proportional odds between serious, moderate, and minor injuries was satisfied for this model. All available demographic variables (age, gender, ethnicity, and facility level o f service) were retained for further model development to control for possible confounding bias. 250 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Estimates of relative risk for the demographic polytomous logistic regression model are also presented in Table 38. .After controlling for gender, ethnicity, and facility type, the risk of serious injury relative to moderate injury and of moderate injury relative to minor injury was 2.7 times higher among patients aged 60 or older, relative to patients aged 30-39 (n=418). Additionally, risk of serious injury relative to moderate and of moderate injury relative to minor injury increased with age (adjusted OR=1.33 per category of age, p < 0 .0 0 1). Increased risk for serious injury relative to moderate and of moderate injury relative to minor was also noted in Hospital ‘A ’ relative to Hospitals ‘B’ and ‘C ’. This finding is expected because Hospital ‘A’ is a Level I Trauma Center, and Hospitals ‘B ’ and ‘C’ are not. No statistically significant associations were noted between ethnicity and injury severity. The estimate of relative risk for serious and moderate injury for those of Hispanic ethnicity changed directions from reduced risk to increased risk (relative to Caucasians after adjusting for age, gender, and facility type), suggesting confounding of ethnicity by age, gender, or facility type. Closer investigation suggested that the relationship between ethnicity and injury severity was confounded by age. D ichotom ous Logistic Regression. A summary o f the dichotomous demographic model development with pertinent likelihood ratio statistics is presented in Appendix 7, Table A 7-2. As in the polytomous model, both the age o f the patient and the facility type showed a statistically significant association with injury severity after controlling for all other demographic variables. A trend for increased relative risk for more severe injuries was noted by age group prior to data reduction (n=634) and after data reduction (n=438). 251 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. As in the polytomous model, no statistically significant interactions were noted between gender and ethnicity or age and gender, confirming the findings from the polytomous model. Insufficient observations were available in all categories to investigate interactions between ethnicity and age. All demographic variables were retained for further model development. Frequencies of demographic characteristics for patients (n=438) included in the dichotomous logistic regression are presented by level o f injury severity in Table 39. The same demographic reference categories were used in the dichotomous model and the polytomous model. Although no observations in the highest group of injury severity (ISS > 9) were noted for the ethnic category of ‘other’ (Egyptian, Iranian, Persian, Armenian, Ethiopian, and mixed ethnicities that did not fit any one category), this group was included in the dichotomous model since the outcome was dichotomized at the cut- point of ISS=4. This resulted in a slightly larger sample size (n=438) than in the polytomous model (n=418). The results for the dichotomous model were similar to those generated from the polytomous model, and are also presented in Table 39. There were no striking differences between unadjusted estimates of risk for moderate and serious injuries and those obtained after adjusting for other available demographics. After controlling for gender, ethnicity, and facility type, the risk of severe or moderate injury relative to minor injury was 2 .2 times higher among patients aged 60 or older, compared to patients aged 30-39. Increased risk for more severe injury was also confirmed in Hospital ‘A’ relative 252 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 39, Dichotomous Logistic Regression: Associations between Demographic Characteristics of Patients1 and Severity of Injury from the Northridae Earthquake ln=438l. N u m b e r ( % ) o f E a r t h q u a k e - R e l a t e d I n j u r i e s b y L e v e l o f I n j u r y S e v e r i t y U n a d j u s t e d O d d s R a t i o 1 A d j u s t e d O d d s R a t i o 1 , 1 M o d e r a t e o r C h a r a c t e r i s t i c M i n o r 4 S e r i o u s * T o t a l s ( 9 5 % C o n f i d e n c e I n t e r v a l ) ( 9 5 % C o n f i d e n c e I n t e r v a l ) G e n d e r M a l e 1 4 8 ( 6 3 ) 3 1 ( 1 7 ) 1 7 9 ( R e f e r e n t ) ( R e f e r e n t ) F e m a l e 1 9 6 ( 7 6 ) 6 3 ( 2 4 ) 2 5 9 1 5 3 1 2 6 ( 0 9 5 - 2 4 8 ) ( 0 7 6 - 2 0 8 ) E l h n i c i t v C a u c a s i a n 2 3 8 ( 7 6 ) 7 4 ( 2 4 ) 3 1 2 ( R e f o r o n t ) ( R o f o r o n t ) A f r i c a n - A m e r i c a n 5 6 ( 8 5 ) 1 0 ( 1 5 ) 6 6 0 5 6 0 5 6 ( 0 2 7 - 1 1 5 ) ( 0 2 6 - 1 2 0 ) H i s p a n i c 3 2 ( 8 0 ) 8 ( 2 0 ) 4 0 0 8 5 1 . 1 6 ( 0 3 8 - 1 . 9 0 ) ( 0 4 9 - 2 . 7 6 ) O t h e r ‘ 1 8 ( 9 0 ) 2 ( 1 0 ) 2 0 0 3 6 0 3 7 ( 0 0 8 - 1 5 8 ) ( 0 . 0 8 - 1 7 1 ) A a e < 2 0 Y e a r s 3 6 ( 8 8 ) 5 ( 1 2 ) 4 1 0 6 6 0 7 9 ( 0 2 3 - 1 . 8 9 ) ( 0 2 7 - 2 3 5 ) 2 0 - 2 9 Y e a r s 7 4 ( 8 7 ) 1 1 ( 1 3 ) 8 5 0 7 1 0 7 1 ( 0 . 3 2 - 1 . 5 7 ) ( 0 3 1 - 1 6 1 ) 3 0 - 3 9 Y e a r s 9 5 ( 8 3 ) 2 0 ( 1 7 ) 1 1 5 ( R e f e r e n t ) ( R e f e r e n t ) 4 0 - 4 9 Y e a r s 5 3 ( 7 3 ) 2 0 ( 2 7 ) 7 3 1 . 7 9 1 . 8 3 ( 0 8 9 - 3 6 3 ) ( 0 8 8 - 3 7 9 ) 5 0 - 5 9 Y e a r s 3 2 ( 7 4 ) 1 1 ( 2 6 ) 4 3 1 . 6 3 1 . 7 7 ( 0 7 1 - 3 . 7 7 ) ( 0 . 7 2 - 4 2 1 ) 6 0 Y e a r s 5 4 ( 6 7 ) 2 7 ( 3 3 ) 8 1 2 3 8 2 . 2 1 (1 2 2 - 4 6 3 ) ( 1 . 1 0 - 4 4 2 ) A g e G r o u p T r e n d ' 1 3 2 1 2 9 ( 1 . 1 4 - 1 . 5 3 ) ( 1 1 1 - 1 . 5 1 ) N > V * u > Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 39. (Continued). N u m b e r ( ' / • ) o f E a r t h q u a k e - R e l a t e d I n j u r i e s b y L e v e l o f I n j u r y S e v e r i t y U n a d j u s t e d O d d s R a t i o 2 A d j u s t e d O d d s R a t i o 2' 2 C h a r a c t e r i s t i c M i n o r 4 M o d e r a t e o r S e r i o u s 9 T o t a l s ( 9 5 % C o n f i d e n c e I n t e r v a l ) ( 9 5 V . C o n f i d e n c e I n t e r v a l ) H o s p i t a l H o s p i t a l ' A ' ' 4 5 ( 6 3 ) 2 6 ( 3 7 ) 7 1 2 8 8 (1 6 1 - 5 1 3 ) 2 9 1 (1 5 8 - 5 3 5 ) H o s p i t a l 'B ' 1 4 6 ( 7 8 ) 4 1 ( 2 2 ) 1 8 7 ( R e f e r e n t ) ( R o f e r e n t ) H o s p i t a l 'C ‘ 7 8 ( 9 5 ) 4 ( 5 ) 8 2 ( R e f e r e n t ) ( R e f e r e n t ) H o s p i t a l 'D ' 7 5 ( 7 7 ) 2 3 ( 2 3 ) 9 8 1 5 3 ( 0 8 7 - 2 6 9 ) 1 5 0 ( 0 8 3 ■ 2 6 9 ) T o t a l * 3 4 4 ( 7 9 ) 9 4 ( 2 1 ) 4 3 8 P a ten ts I m M with aarthquaka-ralatM injures at 4 emergency departm ents aRat the Northndge Earthquake. l o t Angalat County, January 17 3 1 1834 Serious Of Moderate Injury Mlnof Injury ’ Adjusted lor all othar demographic variables Injury Seventy Scotea 1-3 Injury Sevarity Scoraa <4 Egyptian, Iranian. P anian. Armenian. Ethiopian, and unable lo categorue Odda ratios (per age group) repreaent attim ataa or ralattva risk lor injury a t aga gioup m cteatei No a ttu m ed earthquake-related injures (no batelm a data) K > to Hospitals ‘B’ and ‘C’. Similarly, the previously noted reversal of association between estimated risk for more severe injury in patients of Hispanic ethnicity compared to Caucasians after controlling for other demographics (gender, age, and facility type) was confirmed in the dichotomous model. Again, the association was not statistically significant, but the reversal indicates confounding of ethnicity by one or more of the other demographic variables. 3.02. Injury Characteristics with Respect to fniurv Severity Score. Polytomous Logistic Regression. Mechanical circumstances surrounding the injury may be closely correlated with the body location of impact. For example, knife cuts are less likely to occur on the feet than on the hands, and motorcycle injuries have the most severe impact on the head when no helmet is worn. Additionally, their combined effect determines the injury severity score. Since all categories of mechanism and body location may be associated with a range of values for injury severity, these two variables were retained in the model. However, only categories that contained sufficient observations in all three levels of injury severity were included (n=330). Mechanisms of injury that were retained in the model included falls (n=108), cutting or piercing injuries (n=94), and a collapsed group of injuries resulting from being struck by or caught in or between objects (n=128). The remaining mechanisms were diverse with sparse numbers of observations, insufficient diversity with respect to injury severity score (e.g., no 255 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. observed serious injuries), and they were not included in the analysis. “ The category with the largest proportion of injuries (struck by or caught in/between objects) was used as the reference group. Non-specific earthquake-related injuries (n=74) were not included in this analysis since the external cause of the injury was not specified further than an indication that the earthquake was associated with the cause. This information was insufficient for grouping with other mechanisms o f injury. These injuries were also not included as a unique category because of possible misclassification bias. Body location categories were collapsed into 4 groups for analysis due to sparse numbers. The collapsed groups included head or neck (n=181), upper extremity (n=146), lower extremity (n=272), and trunk (back, thorax, chest, abdomen, side or flank) (n=38). Since the category with the largest proportion of injuries was the lower extremity, this was used as the reference category. A summary table showing the development of the injury model with pertinent likelihood ratio statistics is presented in Appendix 7 , Table A7-3. Both body location and mechanism of injury were associated with injury severity prior to adjusting for other variables. This is to be expected, however, since both are used to score the injury severity. Mechanism of injury was associated with severity of injury after controlling for body location, gender, ethnicity, age, and facility type. No other statistically significant associations were detected. No statistically significant interactions between specific “ Deleted observations included motor-vehicle-related injuries (n=9). poisonings and environmental exposures (n=7). bums and electrocutions (n=2). overexertion (n= 14). and slipping or tripping not resulting in a fall (n=23). 256 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. mechanisms and body locations were detected (fall-arm, fall-head, fall-trunk, cut-foot, cut-head, cut-arm). The assumption of proportional odds between serious, moderate, and mild injuries was satisfied for this model. Frequencies of injury characteristics for patients included in the polytomous logistic regression are presented by level o f injury severity in Table 40, along with estimates of relative risk. No remarkable changes in unadjusted and adjusted estimates of risk were detected except for patients injured in body locations grouped as ‘trunk’ and injuries due to falls. Overall, body location was not associated with increased relative risk o f more severe injury. However, injuries to the trunk showed a reversal in direction of association after controlling for mechanism of injury, suggesting confounding of body location by mechanism of injury. Certain mechanisms o f injuries were associated with an elevated relative risk for more severe injury. After adjusting for body location o f injury, gender, ethnicity, age group and facility type, risk for severe injury relative to moderate and o f moderate injury relative to minor increased 2.7 times among patients who fell relative to patients who were struck by or caught in objects. This odds ratio was slightly lower compared to the unadjusted odds ratio of 3.7. Although not statistically significant, risk for serious injury relative to moderate and of moderate injury relative to minor was reduced 0.4 times among patients who were cut or pierced by an object relative to patients who were struck by or caught in objects, after adjusting for body location of injury, gender, ethnicity, age group and facility type. 257 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 40. Polytomous Logistic Regression: Associations between Inlurv Characteristics of Patients1 and Severity of Injury from the Narthrldge Earthquake (n*330l. Number (%) of Earthquake-Related Injuries by Level of Injury Severity Unadjusted Odds Ratio2 Adjusted Odds Ratio2 ' 3 Characteristic Minor4 Moderate3 Serious' Totals (95% Confidence Interval) (95% Confidence Interval) Bcdy Location of injury Lower Extremity 109(60) 17(13) 10(7) 136 (Reference) (Reference) Upper Extremity 55(70) 2 1 (26) 3(4) 79 1 60 (0 8 5 -3 0 2 ) 1 42 (0.70-2.87) Mead & Neck 6 6 (69) 7(7) 4(4) 97 051 (0 24- 1 08) 051 (0 22-1 17) Trunk 14(78) 2(11) 2(11) 18 1 20 (0 38 -384) 0 78 (0 23 - 2 69) Mechanism (External Cause) of Injury Struck by / caught in or between object 110 (86) 13(10) 5(4) 128 (Reference) (Relerence) Cut or pierced by an object 87(93) 6(6) 1(1) 94 049 ( 0 t 9 - 1 22) 039 (0.15-1.04) Falls 67 (62) 28(26) 13(12) 108 373 (1 9 9 -6 9 8 ) 2.72 (1.38 - 536) Totals 264(80) 47(14) 19(6) 330 Patients treated at 4 emergency departments for earthquake-related injuries, Los Angeles County, January 17-31,1994 2 SeriousModerate Injury; Moderate:Mlnor Injury. 3 Adjusted for other Injury characteristics and patient demographics (age, gender, ethnicity, and facility level of care) 4 Injury Severity Scores 1-3 3 Injury Severity Scores 4 8 6 Injury Severity Scores ^9 N ) L /> 00 Dichotomous Logistic Regression. A summary of the dichotomous model development for injury and demographic characteristics with pertinent likelihood ratio statistics is presented in Appendix 7 , Table A 7-4. Both body location and mechanism o f injury were associated with more serious injuries prior to adjusting for the effect of other variables. Mechanism was also associated with more serious injuries after controlling for body locations. Unlike the polytomous model, body location was also associated with more serious injury after controlling for mechanism of injury. No interaction was detected between body location and mechanism of injury. Frequencies of demographic and injury characteristics for patients included in the dichotomous logistic regression are presented by level of injury severity in Table 41. The same reference categories were used in the dichotomous model and the polytomous model. The ethnicity category of ‘other’ (Egyptian, Iranian, Persian, Armenian, Ethiopian, and mixed ethnicities that did not fit any one category) was included in the dichotomous model.b b Additional mechanisms of injury were also included (overexertion, slipping or tripping that did not result in a fall, motor-vehicle collisions, poisonings or being injured by an animal, insect, amphibian, reptile, or plant) since observations were noted in the injury severity category o f greater than or equal to ISS=4. The resulting sample size was increased from 330 in the polytomous model to 375 in the dichotomous model. b b Justification for inclusion of ‘Other' category previously presented in Section 3.01 of Chapter V . 259 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 41. D ichotom ous Logistic R egression: A ssociations betw een Injury Characteristics of P a tien ts1 and Severity oflnlury from the Northrldae Earthquake {n=375l. Number (*/•) of Earthquake-Related Injuries ________ by Level of Injury Severity________ Characteristic Minor4 Moderate or Serious9 Totals Unadjusted Odds Ratio (95% Confidence Interval! Adjusted Odds Ratio2 3 (95% Confidence Interval) Body Location of Injury Lower Extremity 118(79) Upper Extremity 59 (69) Head & Neck 99 (88) Trunk 19 (73) Mechanism (External Cause) o f Struck by / caught in or between object 119 (87) Cut or pierced by an object 90 (92) Falls 68 (62) Slip / Trip (Not Resulting in Fall) 5 (63) 32(21) 27 (31) 14(12) 7(27) 18(13) 8 (0) 42(38) 3(37) 150 86 113 26 137 98 110 (Reference) 1 69 (0 93 -3 07) 0 52 (0 26 -1 03) 1 36 (0 53 - 351) (Reference) 0 59 (0.24-1 41) 4.08 (2.18 - 7.65) 3 9 7 (0 87 -18.04) (Reference) 1 83 (0 91 - 3 67) 0 58 (0 2 6 - 1 28) 0 6 8 (0.22 - 2 07) (Reference) 0 49 (0 1 9 -1 .2 5 ) 3 0 6 (1 5 5 -6 .1 3 ) 4 9 8 (0 97 - 25 57) K » <3\ © 32 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 41. (Continued). Number (%) of Earthquake-Related Injuries by Level of Injury Severity Unadjusted Odds Ratio2 Adjusted Odds Ratio2,3 Characteristic Minor4 Moderate or Serious* rotals (95% Confidence Interval) (95% Confidence Interval) Mechanism (Continued) Motor-Vehicle Collision 5(83) 1(17) 6 1 32 (0 15 -11.98) 2 0 3 (0 1 9 -2 1 73) Poisoning 2(50) 2(50) 4 661 (0 88 - 49 93) 5 8 5 (0 65 - 52 75) Overexertion 6(50) 6(50) 12 661 (1 92 - 22 74) 6 1 2 (1 43 - 26 12) Totals 295(79) 80(21) 375 1 Patients Healed with earthquake-related Injuries at 4 emergency departments alter the Northrldge Earthquake, Los Angeles County, Januaiy 17-31,1994. 2 Serious or Moderate ln|uty:Mk)or Injuiy. 1 Ad|ustad (or other Injury characteristics and patient demographies (age, gender, ethnicity, and facility level ol care). 4 Injury Severity Scores 1 -3 5 Injury Severity Scores 1 4 Estimates of relative risk for the dichotomous injury model are also presented in Table 41. The results were similar for the dichotomous model as for the polytomous model. In the multivariate model, body location (upper extremity) was associated with more severe injury after controlling for mechanism of injury, age, gender, ethnicity and facility. Mechanism of injury was also associated with more severe injury after controlling for body location, age, gender, ethnicity and facility. There were no striking differences between unadjusted estimates of risk for moderate and serious injuries and those obtained after adjusting for other available variables except in patients sustaining injuries to the trunk. These injuries showed a reversal in the direction of association after controlling for mechanism of injury. This finding confirms the reversed association detected in the polytomous model, providing additional support for confounding of body location by mechanism of injury. As in the polytomous model, after adjusting for body location of injury, gender, ethnicity, patient age, and facility type, risk for serious or moderate injury relative to minor injury increased by 3.1 among patients who fell relative to patients who were struck by or caught in objects. 3.03. Building Structure Characteristics with Respect to Iniurv Severity Score. Polytomous Logistic Regression. Nearly all (97%) of the available structure types were wood-framed houses. Since insufficient observations precluded use of any other category as an appropriate reference, all other categories (n=13) were combined with 262 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. structures reflecting missing data for this field. The comparison between wood-framed structures to all others showed no statistical increase in risk of more serious injury. Due to the homogeneity of this variable, structure type was not included in further analysis. Square footage of the building and the unit were categorized by quartile. No associations or trends between square footage of the building or square footage of the unit were detected in relation to injury severity prior to adjusting for the effect of other variables. Similarly, no associations or trends between the categorized year of construction in relation to injury severity were noted prior to adjusting for the effect of other variables. However, a non-significant increase in risk of serious injury (prior to adjusting for the effect of other variables) was noted in structures built prior to 1976 relative to those built during or after 1976. No association was noted between red- and yellow-tagged structures and injury severity prior to adjusting for the effect of other variables. Since an inspection report may have been a surrogate for individual motivation for inspection and perceived damage, tagged structures were modeled relative to non-tagged structures. No association between injury severity and tagged structures was noted prior to adjusting for the effect of other variables. However, only one-sixth of the earthquake-related injury addresses (n=100) were associated with a city building inspection report subsequent to the earthquake. After merging the tagged structures with other structural characteristics, the sample size was reduced to 88 observations. Consequently, the minimum detectable odds ratio would have been 2.52 if equal proportions of the sample were exposed to structural damage as were unexposed (see Table 37) at this reduced sample size, 80% 263 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. power and a=0.05. Note that if the referent exposure prevalence was very low (i.e., 20%) or higher than 50%, the minimum detectable odds ratio would have been greater than 2.52. The odds ratio that was generated based on these data was 1.4. Therefore, the power to detect a relationship between structural damage for inspected structures and injury severity was less than 80%. Therefore, tagging information was not included in further model development and is considered only as a descriptive variable. Most of the injury scene addresses that were successfully linked to structures were residential housing. Types of residences that were represented included single and duplex housing, and multi-family structures. As with structure type, the observations in the remaining categories were few. For modeling purposes, those records (n=22) were deleted along with missing values. Multi-family structures were then compared to single/duplex housing. No association was noted between structure use and injury severity prior to adjusting for the effect of other variables. Structures with use coded were also compared to non-coded structures. No association between injury severity in coded versus non-coded structures was noted prior to adjusting for the effect of other variables. A summary o f the polytomous structural model development with pertinent likelihood ratio statistics is presented in Appendix 7, Table A 7-5. All previously identified variables except structure type and tagging information were included in a model to investigate the overall association between structural characteristics and injury severity. Records with missing values for component structural characteristics were deleted, reducing the sample size to 393. Results of this model are presented in Table 42. 264 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 42. Polytom ous Logistic R egression: A ssociations betw een Structural Factors Linked to Scene of Inlurv * and Severity of Inlurv from the Northrldae Earthquake fn=393). s> o\ cn Characteristic Number (%) of Earthquake-Related Injuries by Level of Injury Severity Minor4 Moderate5 Serious* Totals Univariate Odds Ratio2 (95% Confidence Interval) Adjusted Odds Ratio2 1 (95% Confidence Interval) Use of Structure Single or Duplex Housing 174 (82) 29(14) 8(4) 211 (Roferonce) (Soo Table 46) Multi-family Housing 139(76) 32(18) 11 (6) 182 1 40 (Soo Table 40) (0 90 - 2 39) Year of Construction Pre-1943 83 (79) 17(16) 5(5) 105 1 50 (See lablo 40) (0 74 - 3 04) 1943-1960 60(77) 14(18) 4(5) 78 1 69 (Soo Table 40) (0 80 • 3 56) 1961 -1975 80(77) 17(16) 7(7) 104 1 72 (See Table 40) (0 86 - 3 46) *1976 90(85) 13(12) 3(3) 106 (Reference) (See Table 40) Trend - Age Group of Structure 7 0 9 0 n a (0 7 3 - 1 11) Size of Buildina Less than 1468 Square Feet 72 (81) 14(16) 3(3) 89 (Reference) (Reference) 1468 - 2217 Square Feet 81 (81) 13(13) 6 (6 ) 100 0 9 5 0 98* (0 49 • 1 85) (0 38 - 2 54) 2218 - 8479 Square Feet 81 (79) 18(17) 4(4) 103 0 85 0 46* (0 43 • 1 68) (0 1 5 - 1 43) Greater than 8479 Square Feet 79 (78) 16(16) 6 (6 ) 101 0 83 0 39* (0 4 1 -1 68) (0 13- 1 18) Trend - Square Footage of Building4 1 07 1 37* (0 8 6 -1 34) (0 9 5 - 1 97) Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 42. (Continued). Number (%) of Earthquake-R elated Injuries by Level of Injury Severity Univariate O dds Ratio2 A djusted O dds Ratio2 ' 5 C haracteristic Minor4 M oderate 5 Serious* Totals (95% C onfidence Interval) (95% C onfidence Interval) Size of Unit Less than 1351 Square Feet 66 (75) 18 (20) 4(5 ) 88 (Reference) (Reference) 1351 -1836 Square Feet 84 (83) 1 2 (12) 5(5) 101 1 01 (0.51 -1.99) 0 8 5 (0 34 - 2 15) 1837 - 4991 Square Feet 82 (79) 17(17) 4(4) 103 0.82 (0 4 0 - 1 66) 0 95 (0 3 2 - 2 84) Greater than 4991 Square Feet 81 (80) 14(14) 6 (6) 101 1 30 (0 66 - 2 58) 1 19 (0 40 - 3 58) Trend - Square Footage of Unit9 Totals 313(80) 61 (15) 19(5) 393 0 94 (0 7 5 - 1 18) 0 94 (0 6 6 -1 36) Patients treated at 4 emergency departments for earthquake-related injuries, Los Angeles County, January 17-31,1994. 2 Serious: Moderate Injury; Moderate: Minor Injury. 3 Adjusted for all other structural characteristics. 4 Injury Severity Scores 1-3 5 Injury Severity Scores 4-8 s Injury Severity Scores ; ■ 9 7 Odds ratios (per year of construction group) represent estim ates of relative risk for more serious injury in older housing. a Odds ratios represent estim ates of relative risk for more serious injury per square footage increase in building size. 9 Odds ratios represent estim ates of relative risk for more serious injury per square footage increase in unit size._________ N > & A non-significant increase in the unadjusted relative risk for more serious injury was noted for those patients injured in multi-family housing relative to single or duplex housing. Additionally, a non-significant increase in the unadjusted relative risk for more serious injury was noted for patients injured in structures built prior to 1976 relative to those injured in structures built during or after 1976. No remarkable increases in relative risk for more serious injury were associated with the size of the building or the structure, before or after adjusting for other structural characteristics. However, an interaction was detected between year of construction and structure use (p < 0.025). The resulting estimates of relative risk associated with the interaction variables are presented in Table 43 Although not statistically significant, multi-family structures built after 1975 and those built prior to 1961 were associated with increased risk for more serious injury (adjusting for square footage of the building and unit). Single or duplex housing built between 1961 and 1975 were also associated with a non significant increase in relative risk for more serious injury (adjusting for square footage of the building and unit). Since the size of the building and the unit were not associated with increased relative risk for more severe injury after adjusting for other structural characteristics, these variables were dropped from subsequent analyses. Structural characteristics (structure use and year of construction) were merged with patient demographics, retaining only observations with valid data for all the variables. This resulted in a reduced sample size of 259 observations. Frequencies of characteristics of structures 267 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 43. Polytomous Logistic Regression: Odds Ratios1 for Interaction between Structure Use and Year Built with Respect to Severity of Inlurv fn=393).2 Year Built Structure Use • 1976 1961-75 1943-'60 < 1943 Single or Duplex Housing (Reference) 3.14 1.37 1.36 (0.80 - 12.24) (0.32 - 5.90) (0.35 - 5.23) Multi-family Housing 2.29 0,38 1.95 3.39 (0.59 - 8.96) (0.07 - 2.02) (0.31 -12.38) (0.56 - 20.29) Severe:Moderate Injury; ModerateMlnor Injury (96% Confidence Interval), adjusting for square footage of the building and the unit 2 Patients identified with earthquake-related' j I ss and treated at 4 emergency departments In Los Angeles County, January 17-31,1994 to 0\ 00 2 linked to injury scenes and patient demographics included in the polytomous logistic regression are presented by level of injury severity in Table 44. Two multivariate models were generated: one incorporating the interaction term between year of construction and structure use and one without. The corresponding univariate models (with unadjusted estimates of relative risk) and multivariate models (with adjusted estimates of relative risk) are also presented in Table 44. No interaction was detected between the ethnicity of the patient and the structure use or year of construction. Although the interaction term between year of construction and structure use was no longer detectable after sample reduction, relative risks are still presented for the interaction between structure use and year of construction in Table 45. The assumption for proportional odds between severe, moderate, and minor injuries was satisfied for all models. There were no striking differences between unadjusted estimates of risk for moderate and serious injuries and those obtained after adjusting for other variables. Patients injured in housing built between 1943 and 1960 had 3.2 times the risk for serious injury relative to moderate and of moderate injury relative to minor compared to those injured in housing built after 1975 (controlling for structure use, gender, ethnicity, age, and type of facility). Similarly, patients injured in housing built between 1961-1975 had 2.6 times the risk for serious injury relative to moderate and of moderate injury relative to minor compared to those injured in housing built after 1975 (controlling for structure use, gender, ethnicity, age, and type of facility). However, the combined effect of year of construction and use of the structure showed some interesting relationships. Although 269 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 44. Polytomous Logistic Regression: Associations between Selected Structural Factors Linked (o hUurv Scones with R espect (o Iniun Severity' fn=269). Characteristic Number (%) of Earthquake-Related Injuries by Level of Injury Severity Minor4 Moderate* Serious* Totals Unadjusted Odds Ratio7 (95% Confidence Interval) Adjusted Odds Ratio2 1 (95% Confidence Interval) Use of Structure Single or Duplex Housing 102 (79) 22(17) 5(4) 129 (Reference) (Reference) Multi-family Housing 96(75) 22(17) 10(8) 130 1 28 157 (0.72 - 2 28) (0.79-311) Year of Construction Pre-1943 40(72) 12(21) 4(7) 56 317 294 (1 25 - 8 06) (096 - 885) 1943- I960 35(72) 10(20) 4(8) 49 322 320 (1 2 4 -8 3 7 ) (1 13 - 906) 1961 -1975 61 (74) 16(20) 5(6) 82 273 262 (1 13 - 660) (1 0 5 -6 5 7 ) Post -1975 64(89) 6 (8) 2(3) 72 (Reference) (Reference) Trend - Age Group of Structure7 1 36 1 34 (1.06-1 78) (0 9 8 -1 9 0 ) Total? 200(77) 44(17) 15(6) 259 Patients treated at 4 emergency departments for earthquake-related injuries, Los Angeles County, January 17-31,1994. SeriousiModerate Injury; Moderate:Minor Injury. Adjusted for the other structural characteristic and patient demographics (age, gender, ethnicity, and facility level of care) Injury Severity Scores 1-3 Injury Severity Scores 4-8 Injury Severity Scores * 9 Odds ratios (per year of construction group) represent estimates of relative risk for more serious Injury in older housing. ts» -J O Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 45. Polytomous Logistic Regression: Odds Ratios1 for Interaction between Structure Use and Year Built with Respect to Severity o f Injury.2 Adjusting for Patient Demographics (n=2S9). Year Built Structure Use >1976 1961-75 1943-'60 s 1942 Single or Duplex Housing (Reference) 6.45 5.56 4.78 (0.72 - 57.67) (0.56 - 55.55) (0.53-43.10) Multi-family Housing 3.33 0.28 0.50 0.70 (0.37 - 30.42) (0.02 - 3.23) (0.04 - 6.89) (0.05 - 9.45) * Severe Moderate Injury, Moderate.Mlnor Injury (96% Confidence Interval), adjusting for patient demographics (aige, gender, ethnicity, and facility level of care) 2 Patients Identified with earthquake-related injuries and treated at 4 emergency departments In Los Angeles County. January 17-31,1994_________________ to not statistically significant, patients injured in single or duplex housing built between 1961-1975 were at increased risk for more serious injury relative to patients injured in single or duplex housing built after 1975 ( Table 45). Conversely, patients injured in multi-family housing built between 1961-1975 were at lower risk for more serious injury relative to patients injured in single or duplex housing built after 1975. Dichotomous Logistic Regression. A summary of the dichotomous structural model development with pertinent likelihood ratio statistics is presented in Appendix 7 , Table A 7-6. Results were similar to those from the polytomous model and are presented in Table 46. Structure use was associated with injury severity after controlling for year of construction, and size of the building and unit. The square footage o f the building and the unit were not associated with increased relative risk for more serious injury. However, the interaction between structure use and year of construction was detected prior to data reduction and is presented in Table 47. As in the polytomous model, a non significant increase in risk for more serious injury was noted for patients injured in single or duplex housing built between 1961-1975 relative to those injured in single or duplex housing built after 1975. Similarly, patients injured in multi-family housing built after 1975 were at an increased risk for more serious injury relative to those injured in single or duplex housing built after 1975. Additionally, patients injured in multi-family structures built prior to 1943 were at a non-significant increased risk for more serious injury relative to those injured in single or duplex structures built after 1975. Since the size of the building and the unit were not associated with increased relative risk for more serious injury after adjusting for other structural characteristics, 272 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 46. Dichotomous Logistic Regression: Associations between Structural Factors Linked to Scene ot Inlurv and Severity of Inlurv1 from the Northridoe Earthquake (n*393). Number (%) of Earthquake-Related Injuries by Level of Injury Severity Univariate Odds Ratio2 Adjusted Odds Ratio2 1 Moderate or Characteristic Minor4 Serious5 Totals (95% Confidence Interval) (95% Confidence Interval) Use of Structure Single or Duplex Housing 174 (82) 37(18) 211 (Reference) (Reference) Multi-family Housing 139(76) 43 (24) 182 1 45 (See Table 50) Year of Construction (0 89 - 2.38) Pre-1943 83 (79) 2 2 (21) 105 1 49 (0.73 - 3 03) (See Table 5 0 ) 1943-1960 60(77) 18(23) 78 1.69 (0.80 - 3 57) (See Table 5 0 ) 1961 -1975 80(77) 24 (23) 104 1.69 (0.84 - 3.40) (See Table 50) P o s t-1975 90(85) 16(15) 106 (Reference) (See Table 50) Trend - Age Group of Structure6 1 11 n.a. Size of Buildina (0.90- 1.37) Less than 1468 Square Feet 79 (78) 2 2 (22) 101 (Reference) (Reference) 1468 - 2217 Square Feet 81 (79) 2 2 (21) 103 0 9 8 (0.50-1.90) 1.01 (0.39 - 2 63) 2218 - 8479 Square Feet 81 (81) 19(19) 100 0.84 (0.42-1 68) 0 4 7 (0.15 -1.49) Greater than 8479 Square Feet 72 (81) 17(19) 89 0.85 (0.42-1.72) 041 (0.14-1.26) Trend - Square Footage of Building7 0.93 (0.75-1.18) 0.74 (0.51-1.08) Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 46. (Continued). Number (%) of Earthquake-Related Injuries by Level of Injury Severity Univariate Odds Ratio2 Adjusted Odds Ratio12 Characteristic Minor4 Moderate or Serious6 Totals (95% Confidence Interval) (95% Confidence Interval) Size of Unit Less than 1351 Square Feet 81 (80) 20 (20) 101 (Reference) (Reference) 1351 -1836 Square Feet 82 (80) 21 (20) 103 1.04 (0 52 - 2 06) 0 8 9 (0 35 - 2 27) 1837 - 4991 Square Feet 84 (83) 17(17) 101 0 82 (0 4 0 -1 68) 0 97 (0 3 2 - 2 91) Greater than 4991 Square Feet 66(75) 22 (25) 88 1 35 (0 68 - 2 68) 1 30 (0 43 - 3 95) Trend - Square Footage of Unit6 0 9 4 (0 7 5 - 1.17) 0 92 (0 6 4 -1 .3 3 ) Patients identified with aarthquake-iatetsd tn|uites and heated at 4 emergency departments in Los Angeles County, January 17 31,1994 Moderate and Serious ln|ury:Mlnor Injury Adjusted (or all other structural characteristics Injury Severity Scores 1 -3 Injury Severity Scores 1 4 Odds ratios (per year ot construction group) represent estimates o) lelathie ilsti for more serious Injury In older housing. Odds ratios represent estimates ot rekrtkre risk tot mote serious ln|ury pet square footage increase In building size. Odds ratios represent estimates ot telathre risk for mote serious Injury pet square footage Increase In unit size._______________________ K J 4 ? Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 47. Dichotomous Logistic Regression: Odds Ratios1 for Interaction between Structure Use and Year Built with Respect to Severity o f Injury2 from the Northridae Earthquake fn=393i. Year Built Structure Use . 1976 1961-75 1943-'60 < 1943 Single or Duplex Housing (Reference) 1.69 (0.84 - 3.40) 1.69 (0.80- 3.57) 1.49 (0.73 - 3.03) Multi-family Housing 1.45 (0.89 - 2.38) 0.34 (0.06- 1.85) 1.87 (0.29- 11.97) 2.80 (0.45- 17.37) Serious or Moderate ln)ury:Minor Injury (96% Confidence Interval), adjusting for square footage of the building and the unit 2 Patients identified with earthquaKe-related injuries and treated at 4 emergency departments in Los Angeles County, January 17-31,1994 these variables were dropped from subsequent analyses. Structural characteristics (structure use and year of construction) were merged with patient demographics, retaining only observations with valid data for all the variables. The same reference categories were used in the dichotomous model and the polytomous model. The ethnicity category o f ‘other’ (Egyptian, Iranian, Persian, Armenian, Ethiopian, and mixed ethnicities that did not fit any one category) was included in the dichotomous model. The resulting sample size was increased from 259 to 275. Frequencies o f characteristics of structures included in the dichotomous logistic regression are presented by level of injury severity in Table 48. Two multivariate models were generated: one incorporating the interaction term between year o f construction and structure use and one without. The corresponding univariate models (with unadjusted estimates of relative risk) and multivariate models (with adjusted estimates of relative risk) are also presented in Table 48. No interaction was detected between ethnicity of the patient and structure use or year o f construction. Although the interaction term between year of construction and structure use was no longer detectable after sample reduction, relative risks are presented for the interaction between structure use and year of construction in Table 49. The assumption for proportional odds between serious, moderate, and minor injuries was satisfied for all models. There were no striking differences between unadjusted estimates of risk for moderate and serious injuries and those obtained after adjusting for other variables. 276 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 48. Dichotomous Logistic Regression: Associations between Structural Factors Linked to Scene of Inlurv with Respect to Severity of Inlurv1 from the Northrtdae Earthquake. Adlusttna for Patient Demographics fn»275). Number (*/.) of Earthquake-Related Injuries __________ by Level of Injury Severity____________ Unadjusted Odds Ratio2 Adjusted Odds Ratio2 '3 Moderate or Characteristic_____________________________________ Minor4_________ Serious3 _________Totals________(95% Confidence Interval) (95% Confidence Interval) Use of Structure Single or Duplex Housing 1 1 1 (80) 28(20) 139 (Reference) (Reference) Multi-family Housing 103 (76) 33 924) 136 1.27 1 47 (0 7 2 -2 25) (0.75 -2 89) Year of Construction Pre-1943 45 (73) 17(27) 62 316 313 (1 26 - 7 95) (1 0 7 -9 1 8 ) 1943-1960 40 (74) 14(26) 54 293 3 03 (1 13-7.60) (1.07-8 57) 1961 -1975 62 (74) 22(26) 84 297 285 (1.23-7.16) (1 14-7.12) Post-1975 67 (89) 8(11) 75 (Reference) (Reference) Trend -Age Group of Structure6 075 07 3 (0 5 8 -0 97) (0.53-101) Totals 214(78) 61 (22) 275 ^ Patients identified with earthquake-related injuries and treated at 4 em erjjency departm ents in Los A ngeles County. January 17-31, 1994 2 M oderate and Serious Injury Minor Injury 3 Adjusted for all other structural characteristics Injury Severity S cores 1-3 3 Injury Seventy S co ie s j4 £ O dds ratios (per year of construction group) repiesent estim ates of relative risk for more serious iniuty in older housina K > ^4 --4 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 49. Dichotomous Logistic Regression: Odds Ratios1 for Interaction between Structure Use and Year Built with Respect to Severity of Inlurv from the Northridge Earthquake.2 Adjusting for Patient Demographics In-275). Structure Use >1975 Year Built 1961-75 1943-60 < 1943 Single or Duplex Housing (Reference) 8.08 (0.92 - 70.78) 5.54 (0.56 - 54.94) 5.20 (0.58 - 46.47) Multi-family Housing 3.42 0.22 0.48 0.74 (0.38 - 30.99) (0.02-2.51) (0.04 - 6.43) (0.06 - 9.82) 1 Serious or Moderate lnjury:Minor Injury (96% Confidence Interval), adjusting for patient demographics (age. gender, ethnicity, and facility level of care) 2 Patients identified with earthquake-related Injuries and treated at 4 emergency departments in Los Angeles County, January 17-31,1994 tsl '-j o o Similar to estimates in the polytomous (main effects) model, patients injured in housing built between 1943 and 1960 had 3 .0 times the risk for serious or moderate injury relative to minor injury, compared to those injured in housing built after 1975 (controlling for structure use. gender, ethnicity, age, and type of facility). Similarly, patients injured in housing built between 1961-1975 had 2.9 times the risk for serious or moderate injury relative to minor, compared to those injured in housing built after 1975 (controlling for structure use. gender, ethnicity, age, and type of facility). However, the combined effect of year of construction and use of the structure showed some interesting relationships. Although marginally significant (OR=8.1, 95% CI=0.9 - 70.8), patients injured in single or duplex housing built between 1961-1975 were at increased risk for more serious injury relative to patients injured in single or duplex housing built after 1975 (controlling for demographic characteristics). It appears in this model that all patients injured in single or duplex housing built during or prior to 1975 were at increased risk for more serious injury relative to those injured in single or duplex housing built after 1976. Conversely, patients injured in multi-family housing built before 1975 were at lower risk for more serious injury' relative to patients injured in single or duplex housing built after 1975. 3.04 Geologic Characteristics with Respect to Iniurv Severity Score. Polytom ous Logistic Regression. A summary o f the model development process with pertinent likelihood ratio statistics for geologic and demographic characteristics associated with injury severity is presented in Appendix 7, Table A 7-7. Once all geologic characteristics were merged, the sample size was reduced to 414. Cross-tabulation of geologic data relative to injury severity score showed insufficient observations for any 279 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. interaction terms to be considered. No trend was detected between injury severity and categories of PGA or MMI prior to merging with the demographic variables. No noticeable improvement was noted in the fitted model estimating relative risk for more serious injury due to: a) MMI, controlling for PGA and soil conditions; b) PGA, controlling for MMI and soil conditions; and c) soil conditions, controlling for PGA and MMI. The two types of sedimentary soil (liquefiable and normal) were compared to the non-sedimentary soil (rock). Liquefiable soil was also compared to all other types of soil. No statistical improvement in models was noted between the two models. Since liquefiable soil is thought to be a risk factor for more severe injury, the model incorporating liquefiable soil in reference to all others was selected for subsequent analyses, despite the fact that it was not significantly associated with injury severity in these data. Geologic characteristics were merged with demographics, reducing the sample size to 279. Records that did not contain a valid entry for each geologic and demographic variable were deleted as part of the data reduction. Cross-tabulation of available data in relation to injury severity score showed insufficient observations for any interaction terms to be considered. Characteristics of the geology o f the injury scene included in the polytomous logistic regression models are presented by level of injury severity in Table 50. Available geologic characteristics included estimates of soil type (Evemden 2.5” soil classification), MMI, and PGA. 280 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission Table 60. Polytomous Logistic Regression: Associations between Geology of the Inlurv Scene and Severity of Inlurv1 from the Northrldae Earthquake (n=279). Number (%j of Earlhquake-Related Injuries by Level of Injury Severity Unadjusted Odds Ratio2 Adjusted Odds Ratio2 ' 1 Characteristic Minor4 Moderate9 Serious4 Totals (96% Confidence Interval! (96% Confidence Interval) Modified Meseili. Intensity VI & VII 128(76) 31 (19) 9(5) 168 (Reference) (Reference) V III & IX 86(80) 16(14) 7(6) 111 0 8 6 0 5 6 (0.48-152) (0 2 6 -1 21) Peak Ground Acceleration Less than 0.62 g 42(71) 14(24) 395) 59 (Reference) (Reference) 0.62 - 0 67 g 56(82) 10(15) 2(3) 68 05 4 07 9 (0 2 3 -1 25) (0 3 0 -2 08) 0.68 - 0.79 g 54(83) 8 (12) 3(5) 65 052 1 66 (0 22 - 1 23) (0 5 3 -5 1 8 ) Greater than 0.79 g 64(74) 15(17) 8(9) 87 0 96 52 0 (0.46-1.97) (1.50-1801) Trend - Ground Acceleration Group1 1 01 1.72 (0 79-1.29) (1.14-260) 6 fsra&tL§p!! Cede Rock and Sedimentary Soil 193(76) 44(18) 14(6) 251 (Reference) (Reference) Liquefiable Soil 23(82) 3(11) 2(7) 28 0 75 1 29 (0 28 - 2.04) (0.42 - 4 02) Tofolt 216(77) 47(17) 16(6) 279 Patients treated at 4 emergency departments (or earthquake-related injuries, Los Angeles County, January 17-31,1994 Sertous.Moderate Injury; Moderate.Minor Injury. Adjusted (or all other geologic characteristics and patient demographics (age, gender, ethnicity, and (acility level ot care). Injury Severity Scores 1-3 Injury Severity Scores 4-0 Injury Severity Scores 2 9 O d d s r a t i o s ( p e r g r o u p o ( p e a k g r o u n d a c c e l e r a t i o n ) r e p r e s e n t e s t i m a t e s o f r e l a t i v e r i s k ( o r m o r e s e r i o u s i n j u r y a s g r o u n d - s h a k i n g I n c r e a s e d . 0 0 Summary statistics including estimates of relative risk for moderate and serious injuries are also presented in Table 50. PGA was statistically significantly associated with injury severity after controlling for MMI, liquefiable soil, age, gender, ethnicity, and facility type. Neither MMI or soil conditions were associated with injury severity in the reduced model. However, the estimate of relative risk is unexpectedly lower (although non-significant) in MMI areas of VIII and IX in comparison to MMI areas of VI and VTI. This non-significant finding was detected before and after adjusting for age, gender, ethnicity, facility type, PGA and soil type. A change in the direction of the association between injury severity and peak ground acceleration was noted between unadjusted estimates of relative risk and those obtained after controlling for age, gender, ethnicity, facility type, MMI and soil type. After adjusting for all other variables, an increase in relative risk was apparent as PGA increased, particularly for PGA greater than 0.79 g (OR=5.20, 95% CI=1.50 - 18.01). This type of change indicates confounding of peak ground acceleration by one or more of the demographic variables. Dichotomous Logistic Regression. A summary of the model development process with pertinent likelihood ratio statistics for geologic and demographic characteristics associated with injury severity is presented in Appendix 7, Table A7-8. The process was nearly identical to that o f the polytomous model development. Frequencies of characteristics of the geology included in the dichotomous logistic regression are presented by level of injury severity in Table 51. The same reference categories were used in the dichotomous model and the polytomous model. The ethnicity 282 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 51, Dichotomous Logistic Regression: Associations between Geology of the Inlurv Scene and Severity of Inlurv' from the Northridae Earthquake (n=29SI. Number (%) of Earthquake-Related Injuries ________ by Level of Injury Severity________ Unadjusted Odds Ratio2 Adjusted Odds Ratio' 2.3 Characteristic M inor4 Moderate or Serious* Totals (95% Confidence Interval) (95% Confidence ‘ ' “ Modified Mercalli Intensity VI 8 , VII VIII & IX Peak Ground Acceleration Less than 0 62 g 134 (76) 96(81) 45 (71) 60(82) 0 62 - 0.67 g 0.68 - 0.79 g 58 (84) y . 0 80 g 67(74) T rend6 42 (24) 23(19) 18(29) 13(18) 11(16) 23 (26) 176 119 63 73 69 90 (Reference) (Reference) 0 76 0 57 (0 43- 1 35) (0 27 - 1 25) (Reference) (Reference) 0 54 080 (0 24- 1.22) (0 31 - 2 04) 0.47 1 30 (0 20-1.10) (0 43 • 3 98) 086 363 (0.42- 1.77) (1.08- 12.29) 0.97 1.53 (0.76- 1.24) (1 02 - 2 30) 00 < * » 53 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 51. (Continued.) Number (% ) of Earthquake-Related Injuries by Level o f Injury Severity Unadjusted Odds Ratio2 Adjusted Odds Ratio2 3 Characteristic Minor4 Moderate or Serious9 Totals (95% Confidence Interval) (95% Confidence Interval) Evemden Soil Code Rock and Sedimentary Soil 205 (77) 60 (23) 265 (Reference) (Reference) Liquefiable Soil 25 (83) 5(17) 30 0 68 (0 25 - 1 86) 1 14 (0 37 - 3 55) Totals 230 (78) 65 (22) 296 1 Patients Identified wllh earthquake-related Injuries and treated at 4 emergency departments in Los Angeles County, January 17 ■ 31, 1994 2 Moderate and Serious ln|uryMlnot ln|uty 3 Adjusted for all other geologic characteristics and patient demographics (age, gendoi, ethnicity, and facility levol ot care). 4 Injury Severity Scores 1 -3 5 Injury Severity Scores i4 6 Odds ratios (pet quaitHe of ground acceleration) represent estimates of retatkre risk for more serious Injury as ground shaking Increased. to o o -u category o f ‘other’ (Egyptian, Iranian, Persian, Armenian, Ethiopian, and mixed ethnicities that did not fit any one category) was included in the dichotomous model. The resulting sample size was increased from 279 in the polytomous model to 295 in the dichotomous model. The corresponding univariate models (with unadjusted estimates of relative risk) and multivariate models (with adjusted estimates of relative risk) are also presented in Table 51. These are similar to the results from the polytomous model. There were a few differences between unadjusted estimates of risk for moderate and serious injuries and those obtained after adjusting for other available variables. The previously noted reversal in the direction of the unadjusted relationship between PGA and injury severity (after controlling for age, gender, ethnicity, facility type, MMI and soil type) was confirmed in this model. These findings suggest confounding of PGA by one or more demographic characteristics. As in the polytomous model, patients injured in areas of ground movement greater than 0.79 g had 3.6 times the risk for severe or moderate injury relative to minor compared to those injured in areas of ground movement less than 0.62 g (controlling for MMI, soil type, gender, ethnicity, age, and type of facility). 3.0S. Demographic. Injury. Structural and Geologic Characteristics with Respect to Iniurv Severity Score. Polytomous Logistic Regression. Observations that contained valid data for each previously identified variable (gender, ethnicity, age, body location of injury, external cause of injury, structure use, year of construction, MMI, PGA, Evemden soil code, and 285 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. type of facility) were used in the final analysis. A variable to denote earthquake- attributability (clearly earthquake-related, assumed directly earthquake-related, assumed indirectly earthquake-related) was included to determine whether estimates of risk based on previously identified factors were modified due to assumptions for earthquake- relatedness. The sample size was grossly reduced from 641 to 196 after merging all variables. Due to sparse observations of serious injuries by external mechanism, injuries caused by cutting or piercing objects were collapsed with the category of injuries resulting from being struck by or caught in or between objects. These mechanisms were considered similar enough to combine. The combined group served as the reference category for patients injured by falling. Similarly, PGA was dichotomized (at the 500 1 percentile) due to the reduced sample size. Previous stages of this analysis identified possible confounding variables (gender, ethnicity, age, and facility type) and an interaction between year of construction and structure use. The final model development is presented in Appendix 7 , Table A7-9. Two multivariate models were considered: one including the interaction, and one without. The distribution of the variables by injury severity for the fitted models are displayed in Table 52. Unadjusted estimates o f risk as well as adjusted estimates for the two multivariate models (one with the interaction term and one without) are also presented in Table 52. The assumption for proportional hazards between serious, moderate, and minor injuries was satisfied for all models. The interaction term representing effect 286 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 52. Polytomous Logistic Regression: Assocatlons between Demographic. Inlurv. Structural and Geologic Characteristics of the Inlurv Scene with R espect to Severity o f Inlurv1 from the Norlhrldae Earthquake fn*196l. Number (% } ot Earthquake-Related Injuries ________ by Level ot Injury Severity________ Unadjusted O dds Ratio* Adjusted Odds Ratio* ’3 Characteristic Minor4 Moderate* Serious4 Totals (96% Confidence Interval) (95% Contldence Interval) Demographics Gender Ethnicity Age Group to oo -o Male 68(87) 9(12) 1 (1) 78 (Reference) (Reference) Female 90(76) 19(16) 9(8) 118 220 (1 00 - 483) 1 36 (0 52 - 3 60) Caucasian 123(82) 19(13) 8(5) 150 (Reference) (Reference) African-American 19(73) 6(23) 1(4) 26 1 59 (0 61 -41 5 ) 1 84 (0 4 8 -7 10) Hispanic 16(80) 3915) 1 (5) 20 1 13 (036 - 3.62) 204 (0.44-9.47) Less than 20 Years 17(85) 2 (10) 1(5) 20 1.27 (0 29 - 5 62) 423 (0 6 9 -2 6 0 6 ) 2 0 -2 9 Years 29(88) 4(12) 0 33 0 9 4 (024 - 3.68) 0 8 7 (0.18-428) 30 -39 Years 42(88) 5(10) 1 (2) 48 (Reference) (Reference) 40 -49 Years 29(78) 8 (22) 0 37 1 82 (0 5 7 -5 8 6 ) 1 93 (050 - 7 46) 50 - 59 Years 15 (75) 3(15) 2 (10) 20 2 46 (0 6 6 -9 1 5 ) 1.91 (0.40-9.23) Greater than 56 Years 26(68) 6(16) 6(16) 38 361 (1 22- 10 67) 6 06 (1.64-22 40) Age Group Trend7 1.35 (1.08-1.70) 1.38 (1.03-1.85) Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 62. (Continued). Number (%) of Earthquake-Related Injuries by Level of Injury Severity Unadjusted Odds Ratio2 Adjusted Odds Ratio2,1 Characteristic Minor4 Moderate* Serious* Totals (95% Confidence Interval) (95% Confidence Interval) Hospital Hospital 'A' 26(68) 8 (21) 4(11) 38 2 60 (1 12-602) 557 (1 62-19.13) Hospitals 'B' 61 (79) 12(16) 496) 77 (Reference) (Reference) Hospitals 'C' 38(96) 2(5) 0 40 (Reference) (Reference) Inlurv Characteristics Body Location Hospital D' Lower Extremity 33(80) 73(82) 6(15) 11(12) 2(5) 5(6) 41 89 1 34 (0 5 3 -3 3 5 ) (Reference) 1 56 (0 4 7 -5 1 2 ) (Reference) Upper Extremity 36(73) 11 (22) 2(4) 49 1 56 (0.68-3.58) 263 (0 96-7.22) Head & Neck 42(86) 5(10) 2(4) 49 0.76 (0.29 -1.98) 1.23 (0.39-3.91) Trunk External Cause (Mechanism of Injury) Cut or Pierced by Object 7(78) 54(96) 1 (11) 3(5) 1 (11) 0 9 57 1.38 (0 27 - 699) (Reference) 1.41 (0 22 - 9.24) (Reference) Struck by or Caught In or between Object(s) 64(85) 9(12) 2(3) 75 (Relerence) (Reference) Fall 40(63) 16(25) 8 (12) 64 5.24 (2.49-11.06) 531 (2 1 3 -1 3 2 5 ) N > o o o o Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table 62. (Contlnuedl. Number (%) of Earthquake-Related Injuries by Level of Injury Severity Unadjusted Odds Ratio2 Adjusted Odds Ratio2,9 Characteristic Minor4 Moderate9 Serious1 1 Totals (95% Confidence Interval) (95% Confidence Interval) Structural Characteristics Use of Structure Single or Duplex Housing 63 (86) 1 1 (11) 3(3) 97 (Reference) (Reference) Multi-family Housing 75(76) 17(17) 7(7) 99 192 (0 93 - 3 97) 476 (1.61-14.00) Year of Construction Pre-1943 36(78) 6(13) 4(9) 46 2 19 (0 73-6 5 6 ) 459 (098 - 21 50) 1943-I960 29(74) 7(18) 3(8) 39 2 63 (087 - 801) 460 (103-20.63) 1961 -1975 48(80) 10(17) 2(3) 60 1 85 (0 64 - 5.36) 1.90 (0 52 - 6 97) Post-1975 Year of Construction Trend* 45(88) 5(10) 1 (2) 51 (Reference) 1.28 (093-1.75) (Reference) 167 (1 03 - 2.70) Geokwic Characteristics Modified Mercab Intensity M M I V I & M MI V II 92 (82) 14(12) 7(6) 113 (Reference) (Reference) M MI VIII & M M I IX 66(80) 14(16) 3(4) 83 109 (053 - 2.21) 0.42 (015-1.17) K > 00 VO Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 52. (ContinuedI. Number (%) of Earthquake-Related Injuries by Level of Injury Severity Unadjusted O dds Ratio2 Adjusted Odds Ratio2,1 Characteristic Minor4 Moderate1 Serious Totals (95% Confidence Interval) (95% Confidence Interval) Geologic Characteristics (Continued) Peak Ground Acceleration Less than or equal to 0.67 g 47 (76) 10(16) 5(8) 39 (Reference) (Reference) 0.62 - 0.67 g 36(84) 7(16) 0 52 (Reference) (Reference) Greater than 0 68 g 75 (82) 11(12) 5(6) 91 1 22 (0 6 0 -2 48) 13 47 (35 6 -5 1 0 6 ) Trend - Ground Acceleration (per 0 25 g )9 Evemden Soil Code Rock and Sedimentary Soil (Low Liquefaction Risk) 143(81) 25914) 9(5) 177 1 19 (060 - 236) (Reference) 9 1 0 (2 41 - 34 32) (Reference) Liquefiable Soil T efels 15(79) 158(81) 3(16) 28(14) 1(5) 10(5) 19 196 1 12 (0.35 - 3 56) 1 37 (0 3 4 -5 4 8 ) Patients treated at 4 emergency departments for earthquake-related injuries, Los Angeles County, January 17-31,1984. Serlous:Moderate Injury; Moderate:Minor Injury. Adjusted for all other model variables. Injury Severity Scores 1-3 Injury Severity Scores 4-8 Injury Severity Scores ^9 Odds ratios (per age group) represent estimates of relative risk for more serious Injury as age group increases. Odds ratios (per year of construction group) represent estimates of relative risk for more serious Injury in older housing. Odds ratios (per 0.25 g Increased ground acceleration) represent estimates of relative risk for more serious Injury as ground-shaking Increased. to 8 modification by earthquake-attributability was not significant, and was not included in the fitted model. The age group of the patient, mechanism o f injury, structure use, year of construction, PGA and facility type were each individually associated with injury severity after controlling for all other variables. Since the interaction is not frankly statistically significant, and odds ratios were very similar regardless of inclusion of the interaction, results are presented in the following paragraph without controlling for its effect. The interaction will be presented separately following statistically significant findings. Individuals aged 60 or older had 6 .1 times the risk (relative to 30-39 year olds) of a serious earthquake-related injury relative to moderate and a moderate earthquake- related injury relative to minor (controlling for all other variables). A significant trend for increasing risk of more serious injury as age group increased was also noted (p < 0.05). Patients who sustained upper extremity injuries had marginally increased risk of serious injury relative to moderate and of moderate injury relative to minor (OR=2.6, 95° o CI=T .0-7.2) compared to those sustaining injuries to the lower extremities (controlling for all other variables). Patients who fell had 5 .3 times the risk of serious injury relative to moderate and of moderate injury relative to mild compared to those struck by or cut by objects (controlling for all other variables). Those injured in multi family housing had 4.8 times the risk of serious injury relative to moderate and moderate injury relative to minor compared to those injured in single or duplex housing (controlling for all other variables). Those injured in housing built between 1943-1960 were at 4.6 times the risk of serious injury relative to moderate and moderate injury 291 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relative to minor compared to those injured in structures built after 1975. Patients injured in structures built prior to 1943 had a marginal increased risk of more serious injury compared to those injured in structures built after 1975 (OR=4.6, 95% Cl: 1.0 - 21.5) A significant trend (p < 0.05) was noted with increased risk for more serious injury in older housing Patients in areas of ground motion greater than 0.67 g had 13 .5 times the risk for serious injury relative to moderate and moderate injury relative to minor compared to those in areas of ground shaking less than or equal to 0.67 g. However, the odds ratio had a wide 95% confidence interval (3.55 - 51.06). A significant trend (p < 0.001) was also noted with increased risk for more serious injury as peak ground acceleration increased. Those presenting at one Level I Trauma Center (Hospital ‘A’) had 5.6 times the risk for serious injury relative to moderate and of moderate injury relative to minor compared to those treated in non-Level I Trauma Centers. Differences between unadjusted and adjusted parameter estimates were noted in patient gender, body location of injury, structure use, year of construction, PGA and MMI. The relative risk for more serious injury was reduced for female patients from OR=2.20 (95% CI=1.0 - 4.8) to OR=1.4 (95% CI=0.5 - 3.6), controlling for all other variables. Injuries to the upper extremity became marginally significant after controlling for all other variables. Although non-significant, the relationship between injuries to the head and neck and injury severity reversed direction after controlling for all other variables, indicating confounding of body location of the injury by one or more of the other variables. Closer investigation suggested that the relationship between body location of injury and severity o f injury was confounded by the mechanism of injury. 292 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This is not surprising, since both variables are used to help assign injury severity. Patients injured in multi-family housing had a statistically significant increased relative risk for more serious injury after controlling for all other variables, compared with a non significant association prior to controlling for all other variables. Similarly, those injured in structures built prior to 1960 had a marginally significant increased relative risk for more serious injury after controlling for all other variables, compared with a non significant association prior to controlling for the effect of other variables. Although not significant, patients that were injured in MMI areas of VIII and IX had lower risk for more serious injury relative to those injured in areas of MMI=VI and MMI=VTI after adjusting for all other variables compared with the unadjusted association. Patients injured in areas of peak ground acceleration greater than 0.68 g had a significant increased risk for more serious injury after controlling for all other variables, compared with a non-significant unadjusted association. Dichotomous Logistic Regression. Model development is presented in Appendix Table A 7-10. The process was similar for the dichotomous model as for the polytomous model. Frequencies of demographic, injury, structural and geologic characteristics for patients included in the dichotomous logistic regression are presented by level of injury severity in Table 53. Additional mechanisms of injury were included in this model (overexertion, slipping or tripping that did not result in a fall, motor-vehicle collisions, poisonings or being injured by an animal, insect, amphibian, reptile, or plant) since there were sufficient observations in the category of moderate or serious injuries 293 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. r«Wt M. Ql(ho<moui Loalftlf R w l w : AMOCtoflon* 0«mogfP/i(ci. Inlurv. S(mc(unl tn d Qtrtoais Ct}»nt(»ritOc$ o f tf>» In lu rv S cent ind Stvtrltv o f Inlurv1 fr o m m N orthridm GmVnuHi* (n*W ), I f _ • • 2 H r o m _ u j « w > 2 .3 £ £ i E s I I * x 0 1 o I ST C - s o o A o « O a » Ct n n r* 2 s - m C4 I i 1 u .2 c J n rt ° a s e 3 > o w e • £ £ c £ Uj O $ 2 s I s I > S': 2 " 8 s ” • ~S " 5 2- d 6 2 5*? - a ? a s 3 v A T ? s a n w o S c. ct 8 2 o o N » e 3 > S ! 5 294 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (Nl'MO) <Kl-S0t> K'l « l dnojg •fly • pu»jj. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. T a M e S 3 . fCooUnuadl. N u m b e r ( % ) o f E a r t h q u a k e - R e l a t e d I n j u r i e s b y L e v e l o f I n j u r y S e v e r i t y U n a d j u s t e d O d d s R a t i o 1 A d j u s t e d O d d s R a t i o 1' * M o d e r a t e o r C h a r a c t e r i s t i c M i n o r 4 S e r i o u s * T o t a l s ( 9 6 % C o n f i d e n c e I n t e r v a l ) ( 9 6 % C o n f i d e n c e I n t e r v a l ) Hospital H o s p i t a l 'A ' 2 9 ( 8 2 ) 1 8 ( 3 8 ) 4 7 3 2 5 ( 1 5 3 - 6 8 9 ) 5 2 1 < 1 . 9 4 • 1 7 . 9 9 ) H o s p i t a l s 'O' 6 8 ( 7 9 ) 1 8 ( 2 1 ) 8 8 ( R e f e r e n c e ) ( R e f e r e n c e ) H o s p i t a l s ' C ' 4 2 ( 9 3 ) 3 ( 7 ) 4 5 ( R e f e r e n c e ) ( R e f e r e n c e ) H o s p i t a l ' D ' 4 1 ( 7 9 ) 1 1 ( 2 1 ) 5 2 1 . 4 1 ( 0 . 9 2 - 3 1 7 ) 1 8 6 ( 0 . 5 8 - 9 . 1 2 ) Inlurv C h a r a c t e r t a b c a B o d y l o c a t o r ) L o w e r E x t r e m it y 7 9 ( 8 0 ) 2 0 ( 2 0 ) 9 9 ( R e f e r e n c e ) ( R e f e r e n c e ) U p p e r E x t r e m it y 3 7 ( 7 0 ) 1 6 ( 3 0 ) S 3 1 . 7 1 ( 0 9 0 - 3 8 7 ) 2 . 9 8 ( 1 . 0 1 - 8 9 7 ) H e a d A N e c k 8 3 ( 6 8 ) 9 ( 1 5 ) 8 2 0 9 7 ( 0 2 6 - 1 5 9 ) 0 9 9 ( 0 2 3 - 2 . 0 9 ) T r u n k E x t e r n a l C a u s a (Machanlam of Injuqr) 1 1 ( 8 9 ) 5 ( 3 1 ) 1 8 1 9 0 ( 0 . 5 8 - 5 7 9 ) 0 4 7 ( 0 . 0 9 - 2 . 8 3 ) S t r u c k b y , C a u g h t in o r b e t w e e n O f a ) e c t ( a ) 7 1 ( 8 7 ) 1 1 ( 1 3 ) 6 2 ( R e f e r e n c e ) ( R e f e r e n c e ) P a l l 4 1 ( 6 2 ) 2 5 ( 3 8 ) 8 8 3 9 4 ( 1 . 7 9 - 8 . 9 2 ) 3 . 2 6 ( 1 . 2 5 - 9 . 5 9 ) C u t o r P i e r c e d b y O b j e c t 5 8 ( 9 3 ) 4 ( 7 ) 9 0 0 4 8 ( 0 . 1 4 - 1 . 5 3 ) 0 4 1 ( 0 1 0 - 1 9 2 ) S l i p o r T r ip ( N o t R e s u l t i n g In P a l l ) 4 ( 8 7 ) 2 ( 3 3 ) 9 3 2 3 ( 0 . 5 3 - 1 9 . 7 7 ) 9 . 7 4 ( 0 . 9 1 - 5 8 . 4 5 ) M o t o r - V e h l c l e C o l l i s i o n 3 ( 7 5 ) 1 ( 2 5 ) 4 2 . 1 5 ( 0 2 1 • 2 2 . 9 7 ) 7 . 3 0 ( 0 . 4 4 • 1 2 2 . 1 4 ) P o i s o n i n g , ( M / S t u n g / S c r a t c h e d b y A n i m a l , I n s e c t , R e p t i l e , A m p h i b i a n , P l a n t 2 ( 6 7 ) 1 ( 3 3 ) 3 3 2 3 ( 0 2 7 - 3 8 . 9 9 ) 2 9 9 ( 0 1 7 - 4 1 3 1 ) O v e r e x e r t i o n 3 ( 3 3 ) 8 ( 8 7 ) 9 1 2 . 9 1 ( 2 . 9 1 • 5 9 . 2 9 ) 2 0 . 5 2 ( 3 . 0 1 - 2 3 3 . 7 9 ) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Tab/e 53. fContlnutdl, N u m b e r ( % ) o f E a r t h q u a k e - R e l a t e d I n j u r i e s b y L e v e l o f I n j u r y S e v e r i t y U n a d j u s t e d O d d s R a t i o 1 A d j u s t e d O d d s R a t i o 1- 1 C h a r a c t e r i s t i c M i n o r 4 M o d e r a t e o r S e r i o u s 1 T o t a l s ( 9 6 % C o n f i d e n c e I n t e r v a l ) ( 9 6 % C o n f i d e n c e I n t e r v a l l S t r u c t u r a l C h a r s c t e r l s t l c s Use of Structure S i n g l e o r D u p la n H o u s i n g 9 5 ( 8 1 ) 2 2 ( 1 9 ) 1 1 7 ( R e f e r e n c e ) ( R e f e r e n c e ) Year o f Construction M u lt i- f a m ily H o u s i n g P r e - I M S 8 5 ( 7 5 ) 4 1 ( 7 2 ) 2 8 ( 2 5 ) 1 6 ( 2 6 ) 1 1 3 5 7 1 . 4 2 ( 0 . 7 8 - 2 6 7 ) 3 . 2 5 (1 1 7 - 9 0 6 ) 2 5 3 ( 1 0 1 - 6 3 8 ) 3 9 6 ( 1 0 3 - 1 5 1 3 ) 1 M 3 • 1 9 6 0 3 5 ( 7 3 ) 1 3 ( 2 7 ) 4 8 3 . 1 0 ( 1 0 7 - 6 . 9 3 ) 2 . 7 3 ( 0 6 6 - 1 1 . 0 6 ) 1 M 1 • 1 9 7 5 5 4 ( 7 8 ) 1 5 ( 2 2 ) • 9 2 3 1 ( 0 . 9 3 • 8 4 3 ) 2 . 1 4 ( 0 6 3 - 7 . 2 6 ) P o a t - 1 9 7 5 5 0 ( 8 9 ) 6 ( 1 1 ) 5 8 ( R e f e r e n c e ) ( R e f e r e n c e ) T r e n d • A g e G r o u p o f S t r u c t u r e ' 0 7 2 ( 0 5 4 - 0 9 5 ) 0 6 6 ( 0 . 4 3 - 1 0 0 ) Modified Mercalll Intensity M M I V I 1 4 ( 1 0 0 ) 0 1 4 ( R e f e r e n c e ) ( R e f e r e n c e ) M W VII 8 9 ( 7 3 ) 3 2 ( 2 7 ) 1 2 1 ( R e f e r e n c e ) ( R e f e r e n c e ) M M VIII & M W IX 7 7 ( 8 1 ) 1 6 ( 1 9 ) 9 5 0 . 7 5 ( 0 . 3 9 - 1 . 4 4 ) 0 3 9 ( 0 . 1 4 - 1 . 1 1 ) N u m b e r (% ) of E a r t h q u a k e - R e l a t e d I n j u r l e a by L e v e l of Injury S e v e r i t y U n a d |u « t e d O d d a R a t i o 1 A d j m t e d O d d a R a t i o 1 , 1 Characterlatlc Minor4 Sellout* Total* ( 9 6 % Confidence Interval) ( 9 6 % Confidence Interval) 297 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (ISS ^ 4). The ethnicity category o f ‘other’ (Egyptian, Iranian, Persian, Armenian, Ethiopian, and patients of mixed ethnicities that did not fit any category) was also included in the dichotomous model since there were sufficient observations in the category of moderate or serious injuries (ISS ^ 4). Additionally, a variable to denote earthquake-attributability (clearly earthquake-related, assumed directly earthquake- related, assumed indirectly earthquake-related) was also included to determine whether estimates of risk based on previously identified factors were modified due to assumptions for earthquake-relatedness. The resulting sample size was increased from 196 in the polytomous model to 230 in the dichotomous model. The same reference categories were used in the dichotomous model and the polytomous model except regarding the mechanism of injury and PGA. The reference category for mechanism of injury was the group of injuries caused by striking and being caught in or between objects, since there were sufficient observations in both categories of injury severity to use this group as the reference. Similarly, observations were sufficient in the quartiles of PGA to estimate risk relative to the lowest quartile ( < 0 62 g). Unadjusted estimates of risk as well as adjusted estimates for the two multivariate models (one with the interaction term and one without) are also presented in Table 53. Dichotomous results were similar to those generated from the polytomous model. The term representing levels of assumptions for earthquake-attributability was not significant, and was not included in the fitted model. Statistically significant associations were detected between body location of injury, mechanism of injury, structure use, year of 298 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. building construction, ground motion and type of facility (Level I Trauma Center versus non-Level I Trauma Centers) and injury severity. Since the interaction is not frankly significant (p > 0.05), and the odds ratios were similar regardless of inclusion of the interaction, results are presented in the following paragraph without controlling for its effect. Although marginally significant, those aged 60 or older and those less than 20 years of age were at increased relative risk for more serious injury (controlling for all other variables). No trend was noted between relative risk for more serious injury as age group increased after controlling for all other variables. Those with upper extremity injuries had 2.7 times the risk of moderate or serious injury relative to minor injury compared to those with lower extremity injuries (controlling for all other variables). Patients injured from falling had 3 .3 times the risk for moderate or serious injury relative to minor injury, and those who sustained injuries from overexertion had 26.5 times the risk for moderate or serious injury relative to minor injury compared to patients injured from being struck by, caught in, or caught between objects (controlling for all other variables). Those injured due to slipping or tripping that did not result in a fall were at 6.7 times the risk (non-significant, 95% CI=0.81 - 46.45) for moderate and serious injury relative to minor compared to those struck by, caught in, or caught between objects. Those injured in multi-family housing were at 2.5 times the risk of moderate or serious injury relative to minor injury compared to those injured in single or duplex housing (controlling for all other variables). A significant trend was detected for increased relative risk for more serious injury as housing age increased (p < 0.05). Patients injured 299 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in structures built before 1943 were at 3 .9 times the risk of moderate or serious injury relative to minor injury compared to those injured in structures built after 1975 (controlling for all other variables). Those injured in structures built between 1943 and 1960 had a non-significant increased risk of serious or moderate injury relative to minor injury compared to those injured in structures built after 1975 (controlling for all other variables, OR=2.7, 95% CI=0.68 - 11.08). A significant trend was detected for increased relative risk for more serious injury as peak ground acceleration increased. Patients injured in areas of ground motion greater than or equal to 0.80 g had 7.8 times the risk for moderate or serious injury relative to minor injury compared to those injured in areas of ground motion less than 0.62 g. Patients treated at one Level I Trauma Center (Hospital ‘A’) had 5.2 times the risk of serious or moderate injury relative to minor compared to patients treated at non-Level I Trauma Centers. Differences between unadjusted estimates of risk for moderate and serious injuries and those obtained after adjusting for other available variables were noted for those of'O ther’ ethnicity, body locations of injury that included the trunk, those injured in areas of ground shaking greater than or equal to 0 .6 8 g, and those injured in areas with liquefiable soil. There was a non-significant increase in risk for moderate and serious injury relative to minor in those o f ‘Other’ ethnicities compared to Caucasians (controlling for all other variables). This was a reversal of the unadjusted association, indicating confounding of this ethnic group by one or more of the other model variables. Closer investigation suggested that the relationship between ethnicity and injury severity was confounded by mechanism of injury. There was a non-significant decrease in risk 300 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. for moderate and serious injury relative to minor in those patients sustaining trunk (or torso) injuries compared to those with lower extremity injuries. This was a reversal of the unadjusted association, indicating confounding of body location of injury by one or more of the other model variables. Closer investigation suggested that the relationship between body location of injury and severity of injury was confounded by mechanism of injury, confirming this finding from the polytomous logistic regression. Those injured in areas of peak ground acceleration greater than or equal to 0.68 g were at increased risk for moderate and serious injury relative to minor compared to those injured in areas of ground shaking less than 0.68 g. This was also a reversal of the unadjusted association, indicating confounding of PGA by one or more of the other model variables. Closer investigation suggested that the relationship between PGA and injury severity was confounded by facility type (licensed level of care). Those injured in geologic areas of liquefaction were at increased risk for moderate and serious injury relative to minor compared to those injured in geologic areas of rock or sedimentary soil of low risk for liquefaction (after adjusting for all other variables). This was a change from the non significant unadjusted association between soil type and injury severity. 301 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C H A P T E R VI: D I S C U S S I O N This is the first epidemiologic study of risk for more serious (relative to less serious) earthquake-related injuries for a relatively large sample of patients presenting at urban emergency departments that incorporates a multi-disciplinary approach. Highlights of the study findings will be discussed in the following paragraphs including implications from the estimates of relative risk, a comparison between observed injuries and those predicted by the Early Post-Earthquake Damage Assessment Tool (EPEDAT), a discussion of the modeling and methodology, and the study limitations. 1.00. ESTIMATES OF RELATIVE RISK. Caution must be issued with respect to interpretation of existing study findings. If a relationship is not detected, it does not necessarily imply that no relationship exists. As stated earlier, since there are no registries or surveillance systems to monitor injuries in the population in Los Angeles County, underlying (baseline) injury patterns are unknown. Therefore identifying increases beyond what is normally observed, or decreases in usual activ ity is difficult. Also the referent exposure prevalence (the proportion of the population that is exposed to risk factors such as high ground shaking, fragile housing, non-structural hazards, or other environmental hazards that might result in injury) is unknown, highly variable, and probably modified by demographic characteristics of the population that may be unmeasured or misrepresentative of the population at risk for injury. Therefore, even estimating the power to detect a relationship based on various sample sizes for available 302 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. risk factors is difficult. As the component variable groups were merged, the power to detect a relationship was reduced (due to associated reduction in sample size), and any detectable association would necessarily have been relatively large. 1.01. Demographic Characteristics of Patients with More Serious Injuries. Patients over age 60 were at increased relative risk for more serious injury compared with younger patients. However, more patients were treated between the ages of 30-40 than any other age group. Similarly, although more females presented for treatment than males, risk for more serious injury was not elevated for either gender after controlling for other factors such as age, ethnicity, and mechanism of injury. Confounding of injury severity for patients of ‘Other’ ethnicities was noted due to external cause of injury in the final dichotomous model. This type of confounding association w as not detected in the polytomous model, but fewer mechanisms of injury were included in that model (due to sparse or no serious injuries from specific mechanisms), and no injuries to those of ‘Other’ ethnicities. This association is difficult to explain and may be an artifact of the data due to sparse observations. A possible explanation for this finding considers inclusion of ethnically-specific behaviors that might have precipitated injury incidents. This is, however, speculation, and more in-depth examination of the injury circumstances is necessary to better explain this relationship. Since the relationship between ethnicity and injury severity may be confounded by the mechanism of injury, ethnicity is an important factor to include in all modeling processes. 303 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.02. Characteristics of M ore Serious Injuries. Although more patients sustained injuries to their lower extremities compared to any other body location, patients sustaining injuries to the upper extremities were at risk for more serious injury relative to moderate and minor injury throughout the analysis. This association became more pronounced and statistically significant after adjusting for all other variables (demographics, structural, and geologic characteristics) in the final model. A possible explanation for this finding might be that people are more likely to reach for or try to catch objects with their arms, and this type of action might leave the upper extremities particularly vulnerable to more serious injuries. Alternately, people may be more likely to brace themselves with their arms, and this act may also expose the arms to more hazards, depending on characteristics of objects that are used for bracing. Traditional response recommendations have included instructions to ‘duck, cover, and hold’. How one holds might be better described, and maintaining a compact, tucked position (as recommended for vehicular crashes) might also be considered as an appropriate response. Much of the information gleaned regarding risk for more serious injuries from specific mechanisms is not surprising from an epidemiologic viewpoint. Falls were associated with risk for more serious injury after adjusting for gender, age, ethnicity, and facility. This increase in risk even after adjusting for age suggests that specific education and development of interventions that focus on reducing falls versus any other mechanism of injury (in the event of an earthquake) would be particularly useful. Falls are the second 304 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. leading cause of death due to unintentional injury, and are the leading cause of hospital admission for trauma, so the finding is not surprising (Baker et al.„ 1992). This finding may be important especially if health education can be employed to reduce injuries from falls during and after an earthquake. Further investigation of these falls will provide more detail regarding the exact circumstances surrounding events. If individuals fell down flights of stairs or in showers, for example, these issues are easily addressed by health education. Some of these injuries may have been avoided or less sev ere if patients had prepared differently or had been educated about behaviors that might reduce the risk for more serious injury (i.e., remaining in bed, placing slip-resistant shower mats on shower floors). Although there were too few observations in all categories of severity in the polytomous model to include more external causes o f injury than those caused by falls, the dichotomous model did include other mechanisms. None of these mechanisms were statistically significantly associated with risk for more severe injury except injuries due to overexertion. .Although the confidence interval surrounding this estimate was wide, overexertion can also be prevented through education and preparedness. Proper lifting techniques are taught regularly in some work environments, and could easily be expanded to apply to the home. Patients that were injured due to slipping and tripping that did not result in a fall were at a non-significant increased relative risk for more serious injury. Behavior patterns of these patients are important to investigate more thoroughly since comparing their behavior with the behavior of patients that fell might identify points of intervention. 305 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.03. Characteristics of Structures Associated with More Serious Injuries. One of the most important findings in this study shows that structural damage (reflected in city building inspection reports) was not associated with injury severity. This may be due to the paucity of data available based on building inspection reports. However, these reports are the only existing data that reflect damage, and the data-set is assumed to be accurate. The absence of an association between structural damage and injury severity is important because hazard estimation is often based on structural damage or the likelihood of structural damage based on structural fragility curves. Incorporation of real data into the modeling process shows that other factors (i.e., mechanism of injury, patient behaviors, age of the patient and year of building construction) may be better predictors of injury severity. Those involved with hazard estimation processes may benefit from refining modeling techniques to incorporate details of other risk factors besides structural damage. The relationship between the structural characteristics and risk for more severe injury after adjusting for other variables has been interpreted with caution. Since the interaction between structure use and year of construction was not significant after data reduction and adjustment for patient demographics, injury characteristics, and geologic components of the injury scene, the value of the associated estimates is questionable. How ever, it is possible that this interaction is real, and not detectable due to reduced power subsequent to data reduction. This interaction may be an important relationship that is worth high-lighting. Urban environments are likely to be quite diverse with respect to 306 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. housing characteristics. Models that do not consider the interactive effect of structural components may be misleading when predicting numbers and severities of injuries, and overly simplistic with respect to recommending strategies for reinforcing structures. Even without consideration of the interaction, an increased relative risk for more serious injuries in structures built prior to 1960 was noted after adjusting for demographic characteristics. This is important because patient demographics (age, ethnicity and gender) may impact their choice of housing including newer versus older structures. As noted in Chapter 2, age and gender have also been associated with increased relative risk for more serious injuries for specific mechanisms. Therefore, by excluding these characteristics one may introduce biased estimates of risk due to a possible confounding effect. However, the results presented in this study show that even when all demographics are equal, there is still some increased risk for more serious injury associated with these older structures relative to structures built after 1960. After adjusting for geologic conditions, injury characteristics and demographics, the increased estimated risk in serious injuries relative to moderate and moderate injuries relative to minor was still detected in structures built prior to 1960 relative to those built during and after 1976. The category of housing built between 1943 and 1960 showed a non-significant increased relative risk in the dichotomous model, but a significant increased risk in an older category of housing (pre-1943) was confirmed. The finding may lead one to believe that more rigorous investigation of structural hazards in housing constructed prior to 1960 might be warranted. 307 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The final polytomous and dichotomous models both estimated increased risk for serious and moderate injuries in multi-family housing relative to single or duplex housing after adjusting for geologic conditions, injury characteristics, and patient demographics. This relationship was not detected prior to data reduction after controlling only for patient demographics. However, it was detected prior to merging with the demographic file, along with the interaction between structure use and year of construction. The reappearance of this association at the final step in the modeling process raises some concern about the previously detected interaction between structure use and year of construction. Although inconsistent, the increased risk in multifamily housing and older housing that was detected at early stages in the modeling process may represent a true modification of effect that is not detectable due to lost data. 1.04. Geologic Characteristics of Scenes Associated with More Serious Iniurv. The only geologic characteristic that showed a significant relationship with increased relative risk for more serious injury was peak ground acceleration. M M I was not significant, and the apparent relative reduction in risk for more serious injuries in the polytomous model as M M I increased is counterintuitive. Closer investigation of the distribution of housing characteristics by M M I showed no major differences except for the oldest and newest categories o f housing. A slightly larger proportion o f structures in the polytomous model (n=196) that were located in areas of MMT=VI and V T I were built prior to 1942 (27%) compared to the proportion of structures of that age built in areas of M M I=V m and EX (19%). Conversely, a slightly larger proportion of structures that were 308 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. located in areas of MMI=VHI and IX were built after 1976 (31%) compared to the proportion of structures of that age built in areas of MMI=VI and V T I (22%). Since MMI is a subjective measure, the appropriateness of this measure as a factor on which risk is estimated should be questioned in light of these findings. PGA is more consistently associated with increased relative risk for more serious injury than MMI. The final models estimated significantly increased relative risk for more serious injury in the highest category of ground shaking (50* percentile in the polytomous model, 75th percentile in the dichotomous model). PGA also consistently showed a trend for increased relative risk for serious or moderate injury as ground shaking increased in both dichotomous and polytomous models (after adjusting for patient demographics, injury characteristics and structural characteristics). Additionally, the relationship between PGA and injury severity was clearly confounded by facility type in the dichotomous model. Additionally, although one might expect more serious injuries to be treated at Level I Trauma Centers, and rapid ground movement is thought to be associated with more serious injury, these injuries were treated at both Level I and Level II Trauma Centers, and the injuries were sustained in all quartiles of ground shaking. This is important for researchers from other disciplines to realize and to take into consideration when estimating hazard. Demographic characteristics influence housing choices, but also choices of property location, and access to medical care. These demographic trends are also not stable over time, but in places like Los Angeles (highly mobile, urban environments), are likely to shift relatively quickly. Confounding by demographic characteristics may result in biased (high or low) estimates of relative risk due to ground-shaking. 309 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.00. COMPARISON OF OBSERVED AND PREDICTED INJURIES USING EPEDAT (EARLY POST-EARTHQUAKE DAMAGE ASSESSMENT TOOL) SOFTWARE. As reported in the Final Report submitted by EQE (see Appendix 3), for MMI VI and VH, the observed injury rate was higher than the predicted injury rate (generated from EPEDAT, the Early Post-Earthquake Damage Assessment Tool software) for those injured in wood-framed housing built prior to 1950. However for MMI Vm, the observed injury rate was lower than the predicted rate in the same type of structures. Taking into account the variance in the observed injury rate, and following Chebychev’s Inequality, at least 75% of the observed injuries would fall within the upper and lower boundaries of the injury rate predicted using EPEDAT-ATC methods. The observed injury rates do not coincide as well with the rate predicted using EPEDAT-Whitman methods. The report notes limitations with the sample size and the location that was the focus of this comparison, and these might be associated with reduced precision in calculating observed injury rates. In addition to those limitations, however, one might hypothesize that the EPEDAT-Whitman injury rates may represent all injuries, including those that did not receive treatment. .Alternatively, the higher EPEDAT - WTiitman injury rates may represent a broader assessment of injuries and other indirect health problems resulting from an earthquake (including myocardial infarction, psychological problems, exacerbation of pre-existing conditions, and infectious diseases). Further investigation of agreement between predicted and observed injuries is necessary to improve precision and knowledge regarding the expected point estimates from EPEDAT. 310 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.00. DISCUSSION OF THE METHODOLOGY. One of the most important findings in this study relates to methodology, especially sample selection and case identification, when dealing with injury-specific information that is associated with geologic and engineering parameters. This, in itself^ is a difficult issue to address. However, it is further complicated by shifts from normal operating functions due to a disaster. Two aspects o f the methodology, facility selection and sample representation, will be discussed in detail, since the study results hinge on this approach. 3.01. Sample Selection. With respect to sample or facility selection, the task needs to take into consideration the geographic area affected, the ability to obtain data that is reasonably reliable, and the time involved in accessing the corresponding records. There are enormous obstacles and time involved in the review of highly confidential records such as medical records. In studies such as this that are funded for a finite time-period, one cannot waste time attempting to obtain data from facilities that have a history of being slow, unresponsive, or hostile to data requests. The task of medical record review is labor-intensive and time-consuming. Experienced personnel may review a maximum of 45-50 records per day, assuming the individual has prior training and knowledge regarding interpretation of medical abbreviations, terminology and navigating patient medical records. Nurses, physicians, physician assistants or medical students perform this type of task well, but the nature of the work is somewhat mundane and contrary to 311 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the type of personality that is required. Furthermore, the pay associated with this level of expertise is beyond most research study budgets. Epidemiologists perform this task well, but appropriate training is necessary to ensure that the data is gathered and grouped in a manner that is meaningful from an injury epidemiology point of view. Exaggerated numbers of deaths due to earthquakes have been documented (Durkin, 1995 versus Peek- Asa, 1998) and undercounts of injuries have also been documented (Durkin, 1995; Mahue & Weiss, 1996b; Peek-Asa et al., 1998). Critics of this methodology might hypothesize that only facilities within a certain radius of the epicenter should be included, or that the data collection should be based on some standard distance from the epicenter. This is problematic for several reasons. Location near the epicenter did not necessarily relate well to ground motion. Ground- shaking also did not follow expected patterns o f dissemination (Holmes & Somers, 1996; International & Services, 1995; Noji, 1997). Additionally, documentation of transportation for injured patients that subsequently died showed that some patients were transported unexpectedly large distances rather than being delivered to the nearest facility (EQE, International & California Governor’ s Office of Emergency Services, 1995). Discussions with Emergency Medical Services administrators for the County of Los Angeles explained that flexibility in usual protocols was permitted after the earthquake. Therefore, if paramedics experienced problems en route to an intended destination or if they could not access a destination, they were allowed flexibility regarding the choice of the receiving hospital. Another point of caution regarding inclusion of facilities in 312 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. heavily-impacted areas pertains to availability and quality of data. Facilities in some areas lost data both before and after the earthquake. The ability to obtain reasonably reliable data from facilities that have been damaged is also not a straightforward process. This damage may result in loss or disorganization of existing data in such a manner that baseline or event data is not obtainable. This much was known from the preliminary study. What was not known, but is now documented, is the loss or inability to retrieve medical records subsequent to the disaster. It was a bit of a surprise to find that one facility in a higher-impact area had lost a relatively large proportion (25%) of medical records after the earthquake. One would expect that once record-keeping was initiated, barring further disasters, the medical records would eventually be filed. Although this problem was apparent only at one of the four facilities, that facility was in an area of high MMI (V LL1) and PGA (0.48). The other three facilities were in areas of lower PGA, and may have treated fewer patients that were injured in higher PGAs. Since concern exists regarding which facilities are more representative of the population at risk for injuries and are also worth the effort necessary to review medical records for research; this is an important issue. Validation of the problem of lost post-event data needs to be documented at other facilities before concluding that these facilities may not be worth the effort. A similar issue that is not well-documented is the ability to retrieve and review medical records in general after a period of time (i.e., 4 years). Although only one facility showed a dramatic change in the number of missing records after the earthquake than before, two other facilities were not able to retrieve a fairly large proportion of 313 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. records (15-20%) regardless of the time period. Still another facility housed these older emergency department records by microfiche. This manner of storage added another dimension of encryption to records that are traditionally notorious for being difficult to read. Still another difficulty in determining facility selection pertains to the assessment of ‘high impact’ or ‘heavily impacted’ areas. The wide range of PGA values associated with various levels o f MMI was unexpected (see Table 21). This may be due to the qualitative and highly variable nature of the MMI measure. However, an important take away message from this study is that clearly high areas of ground-shaking may occur in areas of low intensity, and vice-versa. 3.02. Specificity of Case Identification. Sixty-three percent of the cases in this study were assumed to be earthquake- related. A larger fraction of the assumed earthquake-related injuries were minor and a smaller fraction were more severe (compared with clearly earthquake-related injuries). If estimates of risk are based only on records that clearly attribute injuries to the earthquake, not only is the sample size reduced, but also the distribution of injuries by severity is altered. Additionally, the distributions o f external causes and other risk factors may also change by ability to attribute the injury to the earthquake, and the methods through which this is accomplished. These components effect the magnitude of risk estimates, variance around estimates and the power to detect associations. 314 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Although non-earthquake-related injuries may have been included in the group identified as cases, the assumptions for case identification and the manual review of each potential case were conservative. What is not known is how sensitive and specific this methodology is. A qualitative comparison of baseline non-earthquake-related injuries and post-event non-earthquake-related injuries may give some indication regarding specificity (categorization of true non-earthquake-related injuries as such). Although the post-event earthquake-related distributions might be different from the other two categories (injuries recorded as caused by the earthquake and those that were assumed to be caused by the earthquake), one would expect the baseline and post-event zion-earthquake-related distributions to be similar. This was observed for aggregate data regarding the gender of the patient, the age, and the ethnicity of the patient. Therefore it is possible that the specificity is high in this study. 4.00. BASELINE EARTHQUAKE-RELATED INJURIES A notable finding in this research is the baseline activity for earthquake-related injuries. Most research to date has assumed that baseline levels of earthquake-related injuries would be zero, and that only moderate or large seismic events would lead to injury in communities with high seismic preparation (i.e., Southern California and Japan). If the findings in this study based on medical record review are true, the assumption of zero baseline earthquake-related injuries is not valid. Although there were very few injuries 315 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. detected during the study baseline time period, this finding has implications for future research. Future research on earthquake-related injuries would benefit from investigation of baseline time periods in order to determine whether minor earthquakes preceding a moderate or severe event might have also generated injuries. Additionally, future research that models early warning, real-time estimates and hypothetical estimated risk for injury due to earthquake-related environmental hazards might include ongoing efforts to monitor injuries at a sample of emergency departments. Seismic events that are moderate and large are clearly more hazardous than minor events, but some minor events might be characterized by unique ground-shaking in areas that have no seismometers. If this unique ground-shaking generates hazards that cause injury, the injured patients might present at an emergency department for treatment. Surveillance for these types of injuries might benefit policy-makers and practitioners in order to prepare for possible larger events, and researchers in order to identify areas of seismic activity that might be monitored. 5.00. LIMITATIONS. This research is limited by the sample selection. It is not known whether those people included in this study represent the general population at risk for injuries during and immediately after an earthquake. Additionally, this sample focuses only on those patients that appeared at emergency departments for treatment. Notably excluded are those that were treated in the field, those that sought care at clinics or physicians’ offices, those that were injured and did not seek care, those that were cared for by dentists, veterinarians, 316 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. podiatrists, chiropractors, non-traditional care-givers, and those that sought care outside of Los Angeles County. This research is also limited by the data on which estimates of risk were based. Although quality control procedures were in place regarding the data abstraction process for the medical information, the accuracy of the information recorded in the medical records is not known. Additionally, necessary demographics, particularly the ethnic identity of the patient, were missing in nearly one-third of the data used for modeling. Similarly, structural data based on the Los Angeles County Assessor’s Tax Roles did not match with 30% of the injury scene addresses that were obtained from the medical records. This could be due to errors in either addresses from the Assessor’s database or addresses in the injury database. In order to assess this problem, additional funding has been granted through the National Science Foundation to sample addresses from both databases (injury and structural) and verify that they actually exist in the County of Los Angeles and are valid addresses. A correlated problem stems from the intermediary matching of injury data to the Assessor’s database to estimate geologic activity. Notably dropped from analysis during this step were any injuries related to motor vehicle crashes unless those occurred in a structure or in a driveway associated with a structure. This is an important loss of information to consider in hazard estimation. If the Northridge earthquake would have occurred two or three hours later (at 6:30 or 7:30 A.M.), motor-vehicle crashes might have 317 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. been associated with more serious injury. Matching locations by geographic coordinates (latitude and longitude rather than address) is recommended to address this shortcoming. This epidemiologic approach to estimating relative risk of severe or moderate injury due to environmental factors is new to the field of disaster epidemiology. At the onset of this study, no knowledge regarding loss of data due to merging disparate data sets was known. The loss of nearly two-thirds of the injury data due to missing demographic information or building structure details emphasizes the need for better documentation and maintenance of existing medical records and real estate inventories. Without this information, one necessarily must impute values, which creates an additional step in hazard estimation and adds an unknown amount of error to the point estimates. A bit of information that would help to provide a more complete comparison between earthquake-related injuries and non-earthquake-related injuries and perhaps reduce variability of risk estimates is to score injury severity for baseline injuries. This would enable a different approach to the modeling process. Controls could be identified from the baseline time period and matched to the cases on geographic location (census tract) and age. Conditional logistic regression could then be used to investigate risk factors for severe injury for patients exposed to the earthquake relative to those that were not exposed. 318 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C H A P T E R VII: C O N C L U S t O N S There are several important conclusions that are based on this study of injuries that were documented at emergency departments in Los Angeles County after the Northridge Earthquake. These are listed in the next few pages, and explained in more detail in the rest of the chapter. 1 Although neglected from existing hazard estimation methodology, earthquake- related injuries due to falls were consistently identified throughout the modeling process of this study as significantly associated with more severe injury. 2. Most injuries were caused by non-structural elements and may have involved inappropriate response behaviors on behalf o f the patient. 3 .Although not associated with serious injuries, being cut by objects (particularly glass) caused many injuries. 4. Since the modal mechanism of injury was the striking mechanism caused by objects falling on individuals, funding might be directed towards providing education to the general public and providing low-income residents with material and training to secure non-structural objects. 5. Although patients sustained more injuries to their lower extremities compared to any other body location, patients sustaining injuries to the upper extremities were at risk for more serious injury relative to moderate and minor injury. 319 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 . There may be modification of the effect of the association between injury severity, structure use (single versus multi-family housing) and year of construction (pre-1943, 1943-1960, 1961-1975, post-1975). 7. Due to inconsistencies between measuring MMI and PGA, and since they may be mathematically correlated, it may be optimal to include both variables in hazard estimation procedures. 8 . The relationship between geologic parameters and severity of injury may be confounded by demographic characteristics of the injured patient. 9 Since MMI is a measure that is subjective and inconsistently associated with injury severity and building damage, it is a concern from a policy perspective to use MMI to determine disbursement o f emergency funds. 10 It is advisable for hospitals to take extra precautions to ensure that documentation is stored in the usual manner or is retrospectively transcribed into the usual format and housed in a manner so that information can be retrieved at a future date. 11. File cabinets in hospitals need to be secured to walls or ceilings (to prevent damage and injury) and locked during off-peak hours to prevent spillage. 12. Facilities that did not maintain documentation could not be reimbursed for excess services rendered due to the disaster. 13. Without maintaining ongoing surveillance of injuries, it is impossible to know what the baseline or normal rates of injury are, and subsequently impossible to be able to attribute (in the absence of specific documentation) injuries to an event such as an earthquake. 320 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The most consistent significant finding in this study identified a mechanism of injury associated with more serious injury that is not currently included in earthquake hazard estimation methodologies. Falls were associated with more severe injury throughout the modeling process: unadjusted and after controlling for the effect o f other v ariables (patient demographics, structural characteristics and estimates of geologic activity). This echoes findings from Peek-Asa and colleagues, who found that although falls were not associated with fatalities, 56% of hospitalized earthquake-related injuries were due to falls (Peek-Asa, et al., 1998). Investigation in more detail of these injuries will help focus prevention efforts. If the injuries were due to falls from flights o f stairs or falls in showers, community-oriented health education may target at-risk populations. If people were injured while trying to go somewhere else (get out of the shower, run out of the house, etc.), education addressing the appropriate response (stay in the shower and hold onto something, make sure shoes are tied before running down the stairs, etc.) can be developed. Simple solutions such as placing shower mats in the shower can help provide safety measures that will reduce injuries. Furthermore, education is less costly than treating the injuries. An important finding from this research shows that very few structurally-related injuries might be expected in an urban earthquake of moderate intensity. Only 7 (1%) of the 641 earthquake-related injuries identified in this study were caused by structural mechanisms. Of these, only 2 were caused by structural collapse. This echoes findings of previous studies of the Northridge earthquake. Peek-Asa and colleagues found only 8% of hospitalized injuries caused by structural mechanisms (Peek-Asa, et al., 1998). 321 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Similarly Shoaf and colleagues found less than 1% of earthquake-related injuries detected through their population-based survey were caused by structural mechanisms (Shoaf, et al., 1998). Current hazard estimation models have refined methodologies to estimate casualties and injuries due to earthquakes based on structural fragility. Re-evaluation of this foundation might be advisable in light of the current findings, since most injuries in a developed, seismically active urban environment might be caused by non-structural elements. Additionally, behavioral elements and existing information regarding differential distribution of injuries based on demographic characteristics of the population might be considered for inclusion in future models. Although not associated with risk for severe or moderate injury, objects, particularly glass, cut many patients. This also confirms findings from the population- based survey conducted after the Northridge earthquake (Shoaf, et al., 1998). To reduce these types of injuries, it is recommended that people always have a pair of slippers or shoes close to their beds and that they wear these if they leave their beds during a disaster. Additionally, glass that is engineered to shatter with rounded edges (similar to that used in car windows) might be installed for large picture windows or thick windows. Safety latches may also be installed on cupboards and on refrigerators. These types of storage units usually contain breakable bottles and housewares. Installing safety latches may not keep the glass within the units from breaking, but they will help to reduce the dissemination of broken glass onto floors. Objects that fell striking people caused the most frequent type of earthquake- related injury, sometimes while they were lying in bed. Recommendations stemming 322 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. from this finding include planning a room in such a way that heavy furniture and appliances (wardrobes, bureaus, file cabinets, bookshelves, refrigerators and microwaves) are anchored or bolted to a wall. Other artifacts that are heavy or breakable should be anchored or secured to the furniture on which they are positioned. Trophies, televisions, pictures and mirrors that are positioned near or above beds may be dislodged during an earthquake and fall and injure a person asleep in bed. It is recommended that these types of items be placed far from beds, preferably in low-traffic areas. Policy makers might consider funding projects that would provide low-income residents with material, training, or assistance in order to secure heavy furniture. Additionally, policy-makers might consider funding educational efforts to help people respond appropriately during and after a disaster. Alternatively, safety education courses could include basic information regarding appropriate behaviors and risk reducing strategies. Insurance agencies might also consider providing rebates to residents that show proof of safety measures taken within their homes. The interaction between structure use and year of construction that was detected prior to data reduction might be of interest to other researchers as they investigate and model earthquake-related injuries in the United States. If structures built during specific time periods that are used for multi-family or single-family housing can be targeted for reinforcement, injuries may be less severe or prevented altogether. Some inconsistencies were noted for estimates of risk for more severe injury from MMI & PGA. Specifically, high ground-shaking may occur in areas of low perceived intensity, and vice versa, with relative risk for more serious injury reduced in areas of 323 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. high perceived intensity and increased in areas of low perceived intensity. This challenges traditional perceptions and theories that have driven hazard estimation techniques. MMI is useful since it is comparable to historical measures o f earthquake magnitude. However, it is limited by its subjective nature and has shown differential distribution by housing characteristics in a counterintuitive manner. Regardless, it may be optimal to include both variables in risk estimation models since they can be mathematically correlated (although in this study they were independent). Additionally, the effect may be confounded by patient demographics, and it is recommended to always include demographics with geologic measures when modeling earthquake hazards. It is a concern from a policy perspective to think that an inconsistent measure such as MMI has been used in the past to determine financial assistance (Eguchi et al., 1997). City building inspection reports for the City of Los Angeles showed that only 12% of the buildings that were inspected in MMI areas greater than or equal to VIII were red- or yellow-tagged. This was nearly the same proportion as was found in buildings inspected in MMI areas less than V m (13%) (EQE, International & California Governor’ s Office of Emergency Services, 1995). Inconsistencies between building damage from the Northridge earthquake and MMI have been noted by other researchers (Shoaf & Bourque, 1999). Although monies need to be distributed rapidly to those who need it most after an earthquake, policy-makers should be informed of the inconsistencies associated with estimates of ground-shaking and perceptions o f intensity and damage. More conservative or incremental distribution of monies might be more effective than issuing the maximum allowable coverage for loss of habitat to individuals that have not 324 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. requested aid and may not need any. Alternatively, distribution of monies based on PGA or determined through a population-based survey may be more appropriate. Other conclusions stem not so much from what was found in this research, but rather what was not found. Incomplete documentation is of limited use. It is advisable that hospitals maintain at least minimal information on patients treated and released during a disaster. If documentation is maintained, reimbursement for care and treatment of injured patients that exceed a normal day’s activity may be obtained. However, fairly specific information is required for reimbursement including the name o f the patient, the date and time of treatment, and an explanation of the services or procedures provided. Without documentation, a facility will probably absorb this expense. Additionally, if documentation is maintained, investigators can identify risk factors and help provide information that will reduce the numbers and severities of injuries in future events. Hospitals are also advised to take extra precautions to ensure that documentation is stored in the usual manner, or is retrospectively transcribed into the usual format for storage. Documentation of patient activity during a disaster is of no use if it cannot be retrieved. Therefore, filing systems need to be maintained after a disaster. Regarding earthquake preparedness, file cabinets need to be secured to walls or ceilings. Additionally, files should be locked during off-peak hours to prevent spillage if the cabinets open and topple. The importance o f maintaining good patient records and being able to retrieve those records should not be circumvented or neglected because of a disaster. Facilities that do not take precautions surrounding patient medical records may be held accountable for treatment or inaction. 325 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Finally, one of the most important findings of this study addresses the importance of injury surveillance at emergency departments. Without this information, baseline levels o f injuries can not be taken into account, necessitating precise documentation of what is and what is not attributable to the earthquake. If ongoing surveillance of emergency department activity includes proper monitoring of injury mechanisms, researchers, policy makers, and health care practitioners may all benefit from this information. Researchers will have baseline levels of activity from which deviations may be easily monitored after a disaster. Policy makers will have information that may be used to focus prevention efforts and prepare better for future events. Health care practitioners will have information that will help guide resources and staff treatment areas appropriately. Furthermore, if baseline injuries include earthquake-related injuries and they can be identified as such, researchers, policy makers, and health care practitioners will benefit from data that may enhance existing efforts in early-warning and real-time assessment models. All of the preceding conclusions and recommendations are based on findings from the Northridge earthquake and pertain predominantly to residential housing in Southern California. However, many of the recommendations can be applied equally to businesses and public buildings. They may also be applicable to areas other than Southern California. Furthermore, they may help reduce injury from other seismic events such as explosions and bombings. Although the likelihood of these activities is low, terrorist campaigns occur with little or no warning, and associated morbidity and mortality may be reduced by proper preparation. 3 2 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. B I B L I O G R A P H Y Abramson, J. A. (1990). Survey M ethods in Community M edicine. (Fourth ed.). Edinburgh, England: Churchill Livingstone. Alberts, K. A., Goldie, I., Edler, D., & Svanstrom, L. (1991). Injury surveillance at emergency departments in an urban area: Methodology and quality control. Scandinavian Journal o f Prim ary H ealth Care, 9 ,239-243. American Association for Automotive Medicine. (1976). The A bbreviated Injury Scale (AIS), 1976 R evisio n. Des Plaines, IL: American Association for Automotive Medicine. American Association for Automotive Medicine. (1980). The A bbreviated Injury Scale, 1980 Revision . Des Plaines, IL: American Association for Automotive Medicine. American Association for Automotive Medicine. (1985). The A bbreviated Injury Scale, 1985 R evision. Des Plaines, IL: American Association for Automotive Medicine. Applied Technology Council. (1985). ATC-13 Earthquake Damage Evaluation D atafor C alifornia (Technical Report ). Redwood City, CA: Applied Technology Council. Armenian, H., Melkonian, A., & Hovanesian, A. (1998). Long term mortality and morbidity related to degree of damage following the 1988 earthquake in Armenia. Am erican Journal o f Epidemiology, I-f8 (l 1), 1077-1084. Armenian, H., Melkonian, A., Noji, E., & Hovanesian, A. (1997). Deaths and injuries due to the earthquake in Armenia: A cohort approach. International Journal o f Epidemiology, 26(4), 806-813. Armenian, H., Noji, E., & Oganesian, A. (1992). A case-control study of injuries arising from the earthquake in Armenia, 1988. B ulletin o f the W orld H ealth Organization, 70(2), 251-257. Armenian, H., & Shapiro, S. (1998). Epidem iology and H ealth Services. New York: Oxford University Press. Armitage, P., & Berry, G. (1994). Statistical M ethods in M edical Research. (Third ed.). Oxford, England: Blackwell Science Ltd. 327 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Association for the Advancement of Automotive Medicine. (1990). The A bbreviated Injury Scale . Des Plaines, IL: Association for the Advancement of Automotive Medicine. Baker, S. P., O’Neill, B., Ginsburg, M. J., & Li, G. (1992). The Injury Fact Book. (Second ed.). New York: Oxford University Press. Baker, S. P., O’Neill, B., Haddon, W., & Long, W. B. (1974). The Injury Severity Score: A method for describing patients with multiple injuries and evaluating emergency care. The Journal o f Trauma, 14(3), 187-196. Bonnie, R _, Fulco, C., & Liverman, C. (Eds.). (1999). R educing the Burden o f Injury: A dvancing Prevention and Treatment. Washington, D.C.: National Academy Press. Bourque, L. (1997). Book Review: Eric K. Noji, Editor, The Public Health Consequences of Disasters. Earthquake Spectra, 13(4), 851-855. Bourque, L., Peek-Asa, C., Mahue, M., Nguyen, L., Shoaf, K., Kraus, J., Weiss, B., Davenport, D., & Saruwatari, M. (1997, January, 1997). H ealth im plications o f earthquakes: Physical and em otional injuries during and after the N orthridge Earthquake. Paper presented at the International Symposium on Earthquakes and People’s Health: Vulnerability Reduction, Preparedness and Rehabilitation, World Health Organization Centre for Health Development, Kobe, Japan California Institute of Technology & United States Geological Survey Broadcast of Earthquakes (CUBE). (1992). CUBE Handbook: Third D raft. Pasadena, CA: Southern California Seismographic Network, A Cooperative Project of California Institute of Technology and the United States Geological Survey. Centers for Disease Control and Prevention. (1989). Earthquake-associated deaths— California. C entersfor D isease C ontrol and Prevention M orbidity and M ortality W eekly Reviews, 35(45), 767-770. Centers for Disease Control and Prevention. (1992). Rapid health needs assessment following Hurricane Andrew—Florida and Louisiana, 1992. C entersfor D isease C ontrol and Prevention M orbidity and M ortality W eekly Reviews, 41, 685-688. Cheu, D. H. (1989, ). Earthquake related casualties. Paper presented at the Conference XLIX: A Meeting of the United States Ad Hoc Working Group on Earthquake Related Casualties, Baltimore, Maryland. Cobum, A. W., Spence, R. J., & Pomonis, A. (1992, ). F actors determ ining human casualty levels in earthquakes: M ortality prediction in building collapse. Paper presented at the Tenth World Conference on Earthquake Engineering, Madrid, Spain. 328 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Committee on Medical Aspects of Automotive Safety. (1971). Rating the severity of tissue damage: I. The Abbreviated Scale. Journal o f the Am erican M edical A ssociation, 215, 277-280. Coulson. A. H. (1989,). E pidem iologic concepts and earthquake injury research. Paper presented at the International Workshop on Earthquake Injury Epidemiology for Mitigation and Response, Johns Hopkins University, Baltimor, Maryland. de Ville de Goyet, C., del Cid, E., Romero, A., Jeannee, E., & Lechat, M. (1976). Earthquake in Guatemala: epidemiologic evaluation of the relief effort. B ulletin o f the Pan Am erican H ealth O rganization, 10(2), 95-109. DeHaven, H. (1942). Mechanical analysis o f survival in falls from heights of fifty to one hundred and fifty feet. War M edicine, 586-596. Dewey, J. W., Reagor, B. G., Dengler, L., & Moley, K. (1995). Intensity distrubtion and isoseism al maps fo r the Northridge C alifornia earthquake o f January 17, 1994 (Open File Report). Reston, Virginia: United States Geologic Survey. Durkin, M. E. (Ed.). (1995). F atalities, nonfatal injuries, and m edical aspects o f the Northridge Earthquake. (Vol. 116): California Department of Conservation, Division of Mines and Geology Special Publication. Durkin, M. E., & Thiel, C. C. (1991, ). E stim ating casualties in earthquakes: an assessm ent. Paper presented at the Fourth International Conference on Seismic Zonation, Stanford University, Palo Alto, California. Durkin, M. E., & Thiel, C. C. (1992). Improving measures to reduce earthquake casualties. Earthquake Spectra. 5(1), 95-113. Eaton. G. (1994). Testimony o f Dr. Gordon Eaton, D irector, U.S. G eological Survey, NEHRP R eauthorization H earing. Reston: United States Geological Survey. Eguchi, R. T., Goltz, J. T., Seligson, H. A., Flores, P. J., Blais, N. C., Heaton, T. H., & Bortugno, E. (1997). Real-time loss estimation as an emergency response decision support system. The Early Post-Earthquake Damage Assessent Tool (EPEDAT). Earthquake Spectra 13(4), 815-832. Emmi, P. C., & Horton, C. A. (1993). A GIS-based assessment of earthquake property damage and casualty risk: Salt Lake County, Utah. Earthquake Spectra 9(1), 11- 33. 329 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. EQE, International, & California Governor’s Office of Emergency Services. (1995). The Northridge Earthquake o f January 17, 1994: Report o f D ata C ollection and Analysis: P art A : D am age atuiInventory D a ta . Irvine, California. Fingerhut, L., & Warner, M. (1997). Injury Chartbook. Health, U nited States 1996-97. Hyattsville, MD: National Center for Health Statistics. Ghosh, S. (1993). Structural G eology: Fundam entals and M odem D evelopm ents. Tarrytown, NY: Pergamon Press. Gibson, J. J. (1961). The contribution o f experim ental psychology to the form ulation o f the problem o f safety: A b rie f fo r basic research. New York: Association for the Aid of Crippled Children. Glass, R., Urrutia, J., Sibony, S., Smith, H., Garcia, B., & Rizzo, L. (1977). Earthquake injuries related to housing in a Guatemalan village: Aseismic construction techniques may diminish the toll of deaths and serious injuries. Science, 1 9 7 ,638- 643. Goltz, J., Russell, L., & Bourque, L. (1992). Initial behavioral response to a rapid onset disaster: A case study o f the October 1, 1987 Whittier Narrows Earthquake. International Journal o f M ass Em ergencies and Disasters, 70(1), 43-69. Greenland, S. (1977). Response and follow-up bias in cohort studies. Am erican Journal o f Epidemiology, 106, 184-187. Haddon, W. (1980). Advances in the epidemiology of injuries as a basis for public policy. Public H ealth Reports, 95, 411-421. Hayden, G., Gerberich, S., & Maldonado, G. (1995). Fatal farm injuries: A five-year study utilizing a unique surveillance approach to investigate the concordance of reporting between two data sources. Journal o f O ccupational and Environm ental M edicine, 57(5), 571-577. Hlady, W., Quenemoen, L., & Armenia-Cope, R. (1994). Use of a modified cluster sampling method to perform rapid needs assessment after Hurricane Andrew. Annals o f Em ergency M edicine, 23, 719-725. Holmes, W. T., & Somers, P. (1996). Northridge Earthquake of January 17, 1994 Reconnaisance Report, Volume 2. Earthquake Spectra, 1 /(Supplement C), 1-278. International Classification of Diseases, 9th Revision, Clinical Modification, Fourth Edition, 1992, Medicode, Inc., Med-Index Publications, Salt Lake City, Utah. 330 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Irving, L. M., Norton, R. N., & Langley, J. D. (1994). Injury surveillance in public hospital emergency departments. New Zealand M edical Journal, 1 0 7 ,222-223. Karaffa, M.C. (Ed). (1993) International C lassification o f D iseases. (Ninth ed.). Los Angeles, CA: Practice Management Information Corporation. Kelsey, J., Whittemore, A., Evans, A., & Thompson, W. (1996). M ethods in O bservational Epidemiology. (Second ed.). New York; Oxford University Press. Kennedy, R _, Cornell, C., Campbell, R., Kaplan, S., & Perla, H. (1980). Probabilistic Seismic Safety of an Existing Nuclear Power Plant. N uclear Engineering and Design, 59(2), 315-318. Kimball, A. M., Thacker, S. B., & Levy, M. E. (1980). Shigella surveillance in a large metropolitan area; assessment of a passive reporting system. American Journal o f Public Health, 70, 164-166. Kircher, C., Reitherman, R., Whitman, R., & Arnold, C. (1997a). Estimation of earthquake losses to buildings. Earthquake Spectra 13(4), 703-720. Kircher, C. A., Nassar, A. A., Kustu, O., & Holmes, W. T. (1997b). Development of building damage functions for earthquake loss estimation. Earthquake Spectra 13(4), 663-682. Kleinbaum, D., Kupper, L., & Morgenstem, H. (1982). Epidem iologic Research: Principles and Q uantitative M ethods. New York; Van Nostrand Reinhold. Kloner, R., Loer, J., Poole, W., & Perrin, R. (1997). Population-based analysis of the effect of the Northridge Earthquake on cardiac death in Los Angeles County, California. Journal o f the Am erican College o f Cardiology, 30(5), 1174-80. Landen, D., & Hendricks, S. (1995). Effect of recall on reporting of at-work injuries. Public H ealth Reports, 110(3), 350-354. Langley, J D. (1995). Experiences using New Zealand’s hospital based surveillance system for injury prevention research. M ethods o f Inform ation in M edicine, 34(4), 340-344. Lay, T., & Wallace, T. (1995). M odem G lobal Seismology. (Vol. 58). San Diego; Academic Press. Lees-Haley, P., Williams, C., & English, L. (1996). Response bias in self-reported history of plaintiffs compared with non-litigating patients. P sychological Reports, 79(3 Pt 1 ), 811-818. 331 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Leor. J., & Kloner, R. A. (1996). The Northridge Earthquake as a trigger for acute myocardial infarction. Am erican Journal o f Cardiology, 77(14), 1230-1232. Leor, J., Poole, W. K., & Kloner, R. A. (1996). Sudden cardiac death triggered by an earthquake. The New E ngland Journal o f M edicine, 334(1), 413-419. Lilienfeld, A. M., & Lilienfeld, D. E. (1980). Foundations o f Epidem iology. (Second ed.). New York: Oxford. Litan, I. R. (1993, ). P hysical damage a n d human loss: the economic im pact o f earthquake m itigation m easures in M em phis, Tennessee. Paper presented at the Central United States Earthquake Consortium: 1993 National Earthquake Conference, Memphis, Tennessee. Logue, J. N., Melick, M. E., & Hansen, H. (1981). Research issues and directions in the epidemiology of health effects of disasters. E pidem iologic Reviews, 3, 140-162. Los Angeles County Department of Health Services Disease Control Program. (1994). N orthridge Earthquake R apid H ealth N eeds A ssessm ent (Internal Report). Los Angeles: Los Angeles County Department o f Health Services. MacKenzie, E., Steinwachs, D., & Shankar, B. (1989). Classifying severity of trauma based on hospital discharge diagnoses: Validation o f an ICD-9CM to AIS-85 conversion table. M edical Care, 27, 412-422. Mahue, M., & Weiss, B. (1996a, ). M ethodologies fo r Com paring Injury D ata— II: Im pact o f N orthridge Injuries on Em ergency D epartm ents in Los Angeles County. Paper presented at the Second National Workshop on Modeling Earthquake Casualties for Planning and Response, Los Altos, CA. Mahue, M., & Weiss, B. (1996b, April, 1996). P relim inary Findings o f a Study to A ssess Injuries R esulting fro m the 1994 Northridge Earthquake in Los Angeles, C alifornia Paper presented at the National Disaster Medicine Systems Conference, San Diego, California. Marquis, J. (1998, January 22, 1998). New study lowers toll of quake injuries. Los A ngeles Times, pp. Bl, B5. Mittleman, M., Maldonado, G., Gerberich, S., Smith, G., & Sorock, G. (1997). Alternative approaches to analytical designs in occupational injury epidemiology. Am erican Journal o f Industrial M edicine, 32(2), 129-141. 332 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Moore, D., Okamoto, T., Russo, J., Wilson, R., & Rojahn, C. (1985,). The FEMA Earthquake Dam age and Loss Estim ation System (FEDLOSS). Paper presented at the 1985 Multiconference of the Society for Computer Simulation, San Diego, California. Murakami, H. O. (1992, ). Sim ulation m odel to estim ate human loss fo r occupants o f collapsed buildings in an earthquake. Paper presented at the Tenth World Conference on Earthquake Engineering, Madrid, Spain. National Oceanic and Atmospheric Administration. (1972). A study o f earthquake losses in the San F rancisco B ay Region (Prepared for the Office of Emergency Preparedness by the United States Department of Congress, National oceanic and Atmospheric Administration. Washington: Office of Emergency Preparedness ). National Oceanic and Atmospheric Administration. (1973). A study o f earthquake losses in the Los A ngeles C alifornia area (Prepared for the Federal Disaster Assistance Administration by the Department of Housing and Urban Development / United States Department o f Commerce, National Oceanic and Atmospheric Administration.). Washington: Federal Disaster Assistance Administration. Noji, E. (1994a). Progress in disaster management. Lancet, 343, 1239-1240. Noji, E. (1994b). The role o f epidemiology in seismic vulnerability reduction: An interdisciplinary approach. The Cusac Journal, 8-9. Noji, E. K. (Ed.). (1997). The Public H ealth Consequences o f D isasters. New York: Oxford University Press. Noji, E. K., Jones, N. P., Smith, G. S., & Krimgold, F. R. (1989, ). Use o f quantitative measures o f injury severity in earthquake research. Paper presented at the International Workshop on Earthquake Injury Epidemiology for Mitigation and Response, Johnson Hopkins University, Baltimore, Maryland. Okada, S. (1993). Study on the Evaluation of seismic casualty risk potential in dwellings: Part 1 — Indoor zoning map on dwelling space safety. Journal o f Structural and Construction Engineering(454), 39-49. Olson, R. A, & Eidinger, J. M. (1995). M ethodologiesfor evaluating the socio-econom ic consequences o f large earthquakes: Identification o f critical fa c ilitie s (R24.01 Rev. C). Berkeley, California: University of California at Berkeley. Parker, S. (Ed.). (1994). M cG raw -H ill D ictionary o f Geology an d M ineralogy (Fifth ed.). New York: McGraw-Hill. 333 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Payne, S. R _, & Waller, J. A. (1989). Trauma registry and trauma center biases in injury research. The Journal o f Trauma, 29(4), 424-429. Peek-Asa, C., Kraus, J. F., Bourque, L. B., Vimalachandra, D., Yu, J., & Abrams, J. (1998). Fatal and hospitalized injuries resulting from the 1994 Northridge earthquake. International Journal o f Epidemiology, 27, 459-465. Ribbeck, B. M., Runge, J. W., Thomason, M. H., & Baker, J. W. (1992). Injury surveillance. A method for recording E Codes for injured emergency department patients. A nnals o f Emergency M edicine, 2/(1), 37-40. Risk Management Software. (1993). Assessm ent o f the State-of-the-Art Earthquake Loss Estim ation M ethodologies (Report to the National Institute of Building Sciences ). Washington, DC: National Institute of Building Sciences. Rivara, F. P., Morgan, P., Bergman, A. B., & Maier, R. V. (1990). Cost estimates for statewide reporting of injuries by E-Coding hospital discharge abstract data base systems. P ublic H ealth Reports, 105(6), 635-638. Robinson, A. (1993). Earthshock: Hurricanes, Volcanoes, Earthquakes, Tornadoes, and Other Forces o f Nature. New York: Thames and Hudson. Roces, M., White, M., Dayrit, M., & Durkin, M. (1992). Risk factors for injuries due to the 1990 earthquake in Luzon, Philippines. B ulletin o f the W orld Health Organization, 70(4), 509-514. Rosner, B. (1990). Fundam entals o f B iostatistics. Boston, MA: PWS-Kent Publishing Company. Rothman, K. (1986). M odem Epidem iology. Boston: Little, Brown and Company. Rothman, K., & Greenland, S. (1998). M odem Epidem iology. (Second ed ). Philadelphia: L ippi ncott-Raven. Runyan, C. W., Bowling, J. M., & Bangdiwala, S. I. (1992). Emergency department record keeping and the potential for injury surveillance. The Journal o f Trauma, 32(2), 187-189. Samardjieva, E., & Oike, K. (1992). Modeling the number of casualties from earthquakes. Journal o f N atural D isaster Science, 14( 1), 17-28. Scientists of the United States Geological Survey and the Southern California Earthquake Center. (1994). The magnitude 6.7 Northridge, California Earthquake of 17 January 1994. Science, 266, 389-397. 334 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Seare, S. P., Speirs, L., & Driscoll, P. (Eds.). (1993). International C lassification o f D iseases, f h Revision, C linical M odification. (Vol. 1 & 2). Salt Lake City: Med- Index Publications. Shinozuka, M., Change, S., Eguchi, R., Abrams, D., Hwang, H., & Rose, A. (1997). Advances in earthquake loss estimation and application to Memphis, Tennessee. Earthquake Spectra, 13(4), 739-758. Shiono, K. (1992, ). Assessm ent m odelfor earthquake fa ta lities. Paper presented at the Second U.S.-Asia Conference on Engineering for Mitigating Natural Hazards Damage, Yogyakarta, Indonesia. Shiono, K., Krimgold, F., & Ohta, Y. (1991, ). M ethod fo r the estim ation o f earthquake fa ta litie s and its applicability to the global m acro-zonation o f human casualty risk Paper presented at the Fourth International Conference on Seismic Zonation, Stanford University, Palo Alto, California. Shlesselman, J. (1982). Case-Control Studies: D esign, Conduct, Analysis. New York: Oxford University Press. Shoaf, K. (1998). Personal Communication . Shoaf, K., Sareen, H., Nguyen, L., & Bourque, L. (1999). Injuries as a Result of California Earthquakes in the Past Decade. The Journal o f D isaster Studies and M anagement, (In Press). Shoaf, K. I., & Bourque, L.B. (1999). Correlates of damage to residences following the Northridge Earthquake, as reported in a population-based survey of Los Angeles County residents. Earthquake Spectra, 15( 1), 145-172. Smith, G. S. (1990, ). The application o f epidem iology to the reduction o f earthquake related casualties: Research needs. Paper presented at the Conference XLIX, a Meeting of the United States ad hoc Working Group on Earthquake Related Casualties, Reston, Virginia. Sommer, A., & Mosley, W. (1970). East Bengal cyclone of November, 1970. Epidemiological approach to disaster assessment. Lancet, 1, 1029-1036. Steinbrugge, K. V. (1990, ). Comments on the history o f earthquake casualty estim ation. Paper presented at the Workshop on modeling earthquake casualties for planning and response, Asilomar Conference Center, Pacific, California. 335 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Stojanovski, P., & Dong, W. (1992). Sim ulation m odel fo r earthquake casualty estim ation (Technical Report). Menlo Park, California: Risk Management Software. Teeter, D. (1996). Illnesses and injuries reported at Disaster Application Centers following the 1994 Northridge Earthquake. M ilita ry m edicine, 161(9), 526-30. The National Committee for Injury Prevention and Control. (1989). Injury Prevention: Meeting the Challenge. A m erican Journal o f Preventive M edicine, Supplem ent to Volume 5(3), 4-20. Thomas Brothers. (1993). The Thomas Guide, 1993 Los Angeles County (1993 Census Tract Edition ed.,). Irvine: Thomas Bros. Vogt, R L., Clark, S. W., & Kappei, S. (1986). Evaluation of the state surveillance system using hospital discharge diagnoses, 1982-1983. American Journal o f Epidem iology, 123, 197-198. Wagner, R (1996). A Case-Control Study o f R isk F actorsfor Physical Injury during the M ainshock o f the 1989 Lom a P rieta Earthquake in the County o f Santa Cruz, C alifornia. Unpublished Dissertation, The Johns Hopkins University, Baltimore. Wagner, R (1999). Personal Communication . Wagner, R M., Jones, N. P., & Smith, G. S. (1994). Study methods and progress report: a case-control study o f physical injuries associated with the earthquake in the county o f Santa Cruz (U.S. Geological Survey Professional Paper 1553-A, EERC Library Call #: 705.326/1989L/U75/1553- A). Washington, D C.: United States Government Printing Office. Whitman, R , Anagnos, T., Kircher, C., Lagoria, H., Lawson, R , & Schneider, P. (1997). Development of a national earthquake loss estimation methodology. Earthquake Spectra, 13(A), 643-661. Whitman, R , Cornell, C., & Taleb-Agha, G. (1975,). Analysis o f earthquake risk fo r lifeline systems. Paper presented at the United States National Earthquake Engineering Conference, Ann Arbor, Michigan. Williams, J. M., Furbee, P. M., Prescott, J. E., & Paulson, D. J. (1995). The emergency department log as a simple injury-surveillance tool. Annals o f Em ergency M edicine, 25(5), 686-691. Yin, Z. (1993). A method to predict and classify earthquake disasters. E arthquake R esearch in China, 7(1), 55-68. 336 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX Is Grant Proposal Submitted to the United States Geological Survey (Department of the Interior) and Associated Award Letter 337 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. RISK FACTORS ASSOCIATED WITH MODERATE AND SERIOUS INJURIES ATTRIBUTABLE TO THE 1994 NORTHRIDGE EARTHQUAKE, LOS ANGELES COUNTY Volume II by Maya Louise Mahue-Giangreco A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements of the Degree DOCTOR OF PHILOSOPHY (Epidemiology) December 1999 Copyright 1999 Maya Louise Mahue-Giangreco Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. NEHRP Information Proposal Summary 1. Project Title: Risk and Loss Associated with Earthquake Hazards 2. Principal Investigator: Billie P. Weiss. M.P.H.. Director Injury and Violence Prevention Program Los Angeles County Department o f Health Services 213-240-7785 Fax:213-250-2594 313 N. Figueroa. Room 127 Los Angeles. CA 90012 bweissfa)dhs.co.la.ca.us 3. Regional Panel Designation: SC 4. Element Designation: Element II 5. Key Words: Building response, ground motion, soil-structure interaction. site effects, urban hazards, seismic zonation, earthquake effects, loss estimation, earthquake forecasting, earthquake probabilities, emergency preparedness, policy, real-time earthquake information, database Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 6. Amount Requested: 7. Proposed Start Date: 8. Proposed Duration: 9. Applicants Grants/Contracts Official: 10. New or Renewal Proposal: 11. Active Earthquake-related Research Grants, and Level o f Support: 12. Has proposal been submitted to other agency for funding? U > V O $90,000 11/1/96 12 months Bob Frangenberg 213-351-5264 Fax:213-388-3370 550 S. Vermont Room 412 Los Angeles, CA 90020 rftangenberg@dhs.co.la.ca.us New None No 2 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. NEHR.P Information Proposal Summary Proposal Summary: Risk and Loss Associated with Earthquake Hazards Principal Investigator: Billie P . Weiss. M.P.H.. Director Injury and Violence Prevention Program Los Angeles County Department of Health Services 213-240-7785 Fax:213-250-2594 313 N. Figueroa. Room 127 Los Angeles. CA 90012 bweiss@dhs.co.la.ca.us Regional Panel Designation: SC Element Designation: Element II Proposed Duration: 12 months Application of improved methodology employing epidemiologic investigational techniques to estimate injuries and death due to earthquakes will be conducted to test the following hypotheses: I) The number of injuries reported in emergency departments after the Northridge earthquake that occurred in residences that were damaged is statistically significantly different than those injuries reported in emergency departments that occurred in residences that were not assessed as damaged or inspected by Building and Safety Inspectors, and 2) reported variation in type and severity of injuries detected in emergency departments after the Northridge earthquake can be explained in part by increased risk associated with specific structure types, seismic activity at the injury location, and demographic characteristics. A b stra c t Estim ates o f casualties associated with earthquakes do not typically em ploy real data, but depend heavily on simulated data. Conclusions based on m odels from sim ulated data may be m isleading. The specific problem that will be addressed by this research involves improving m ethodology regarding estimation o f injuries and death due to earthquakes by employing ep id em iologic investigational techniques to obtain geographic locations o f patients who sought care at emergency departments for injuries resulting from the 1994 N orthridge earthquake. The Los Angeies County Injury and Violence Prevention Program (IV PP) will test the following hypotheses by employing standard epidemiologic techniques: 1) The num ber o f injuries reported in em ergency departments after the Northridge earthquake that occurred in residences that were damaged (as assessed by either a red or yellow tag) is statistically significantly different than those injuries reported in emergency departments that occurred in residences that were not assessed as damaged or inspected by Building and Safety Inspectors, and 2) reported variation in type and severity o f injuries detected in em ergency departm ents after the Northridge earthquake can be explained in part by increased risk associated with specific structure types, seismic activity at the injury location, and demographic characteristics (i.e.. age. gender, ethnicity). The ap p ro ach will be to abstract actual data for patients presenting at em ergency departments in the San Fernando Valley, including location o f the patient at the tim e o f injury. The injury data will be grouped by type and severity o f injury and m atched w ith structure and geologic information including dam age sustained at the structure. M M I. PG A. and 2.5 minute soil code for the location. The research will meet the follow ing objectives: 1) abstract medical records to describe injured patients seen at emergency departm ents betw een January 17. 1994 and January 3 1 .1 9 9 4 . 2) link these data with structures. 3) identify those structures that sustained damage during the earthquake. 3) identify geologic com ponents that describe the status o f the ground during the earthquake. 4) determine if earthquake-related injuries occurred in structurally compromised structures. 5) estimate risk associated with structure type, degree o f damage. Modified Mercaiii Index (MMI), Peak Ground Acceleration (PG A ), and 2.5 minute soil code information. It is anticipated that the results will show a positive correlation between injury severity and seismic activity. The implications of the project results will help hazard assessment abilities, and improve mitigation strategies. By defining the m echanism s and severities o f earthquake- related injuries, and describing how and where people were during the quake, recommendations can be made regarding safer building designs. Additionally, by com paring M onte Carlo simulations to parameter estimates obtained using real data, existing m odels can be validated or refined. Theoretically, if injury events can be patterned from past earthquakes and incorporated into a comprehensive m odel, more realistic outcom es can be predicted for future earthquakes. 341 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Significance of the Project. Estimates of projected casualties associated with earthquakes do not typically employ real data. Conclusions based on models from simulated data may be misleading. The specific problem that will be addressed by this research involves improving methodology regarding estimation of injuries and death due to earthquakes by employing epidemiologic investigational techniques to obtain geographic locations of patients w ho sought care at emergency departments for injuries resulting from the 1994 Northridge earthquake. In the United States, few researchers have attempted to gather actual data relating to injuries that might be attributable to earthquakes for two main reasons. One: medical documentation after catastrophic disasters is secondary to caring for the health needs of victims of disasters. Two: medical records are highly confidential and gaining access to them is difficult for non-governmental researchers. Rather than use real data, many researchers prefer to employ simulated data in Monte Carlo models': 34 5 6 7 * 9 1 0 " 1 2 13 14 Simulated data is a starting-point for estimating casualties, but this current methodology can be improved upon by applying standard epidemiologic investigational techniques. Only by defining the mechanisms and seventies of earthquake-related injuries, and describing how and where people were during the quake, can recommendations be made regarding safer building designs1 5 . By comparing Monte Carlo simulations to parameter estimates obtained using real data, simulations can be refined. Theoretically, if injury events can be patterned from past earthquakes and incorporated into a comprehensive model, more realistic outcomes can be predicted for future earthquakes. A significant contribution to the National Earthquake Hazards Reduction Program (NEHRP) would be to report type and severity of injury by structure, damage sustained, and geologic information (USGS Modified Mercaiii Index (MMI1 6 ). Peak Ground Acceleration (PGA1 7 ), and 2.5 minute soil code"). This would help increase earthquake hazard assessment abilities, enhance or validate existing models, and provide information useful to promotion and design of more effective mitigation strategies. Prgjwt PlkDj The specific hypotheses to be tested are as follows: 1. The number of injuries reported in emergency departments after the Northridge earthquake that occurred in residences that were damaged (as assessed by either a red or yellow tag) is statistically significantly different than those injuries reported in emergency departments that occurred in residences that were not damaged or inspected (as assessed by a green tag or no tag). 2. Reported variation in type and severity of injuries detected in emergency departments after the Northridge earthquake can be explained in part by increased risk associated with 342 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. specific structure types, seism ic activity at the injury location, and dem ographic characteristics (i.e.. age. gender, ethnicity). M ethodology Under the auspices of the Heaith Officer of Los Angeles County (Title 17. Sections 1250 and 2500 of the California Health and Safety Code), the Injury and Violence Prevention Program (IVPP) of Los Angeles County Department of Heaith Services (LAC-DHS) has obtained injury- specific log data from a sample of emergency departments concentrated in the San Fernando Valley for the months of January from 1992. 1993. and 1994. Since these data do not include address and usually omit mechanism of injury, medical record review is necessary to obtain necessary details. By reviewing medical records of those injuries that were reported during and after the earthquake, mechanism o f injury may be obtained, allowing an objective assessment as to whether the injury might be earthquake-related. Known eanhquake-reiated injury site addresses will be provided to EQE Engineering to be linked to their damage database and Northridge geologic database. EQE wiii maximize the number of matches to 80-90% of ail possible matches for those injuries that are explicitly stated as being earthquake-related. Injury site addresses for injuries detected between January 17. 1994 and January 31. 1994 that are not explicitly identified as being earthquake-related will also be provided to EQE for a quick assessment of address match in order to obtain any possible correlations. EQ E will contribute expen knowledge and experience with both the damage data set as well as the Northridge geologic data1 9 :o :i — 2 2 2 4 2 5 2 4 2 7 2129 M . EQE will provide IVPP with the Building and Safety Inspection Tag status (red. yellow, or green) reflecting damage sustained, structure type, year built, and geologic information including MMI-1 1 . PGA2 2 , and USGS 2.5 minute soil code2 3 for ail matched data. Data collection and management will follow previously established guidelines (see Appendix 1). Emergency department log data have been grouped by injury type, body location of injury, disposition of patient, mode of arrival of patient, and patient age and ethnicity (when available) (see Appendix 2 for partial list of coding schemes). An instrument to be used for medical record abstraction has been designed and pilot-tested (see Appendix 3). Contacts have been established at 21 facilities, including 4 trauma centers. When possible, data will be obtained in machine- readable format. LAC-DHS assures confidentiality of all medical records, therefore individual identifiers will be dropped and data wiii be presented in summary form only. Data analysis will use a t-test to determine whether the number of injuries reported in structures that were damaged is statistically significantly different than the number of injuries reported in non-damaged or non-inspected structures (hypothesis #1). To test hypothesis #2. a multivariate logistic model will be employed to determine whether structure type, degree of damage, seismic activity, and available demographic characteristics (i.e.. ethnicity, age. gender) are predictive of injury. Estimates of risk of injury due to demographic characteristics or exposure to 343 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. environmental elements can be obtained through odds ratios (exponentiation of the model's parameter estimates > . Continuing efforts will focus on obtaining injury data from all facilities in Los Angeles Couty, including those not concentrated in the San Fernando Valley. An undocumented amount of paramedic transport and emergency vehicle traffic was diverted to more distant facilities. Other efforts address ongoing emergency department surveillance in order to respond to future disasters in a timely manner and to provide baseline injury data on a regular basis. Requests for this activity have been submined to other federal agencies such as the Centers for Disease Control and Injury Prevention in Atlanta. Final Report and Dissemination. Injury data will be presented in summary form by structure type, degree of damage, seismic activity, and census tract. Additionally, these data will be mapped by appropriate geographic boundaries (i.e.. census tract, health district). High risk groups can be identified and categorized demographically. by mechanism of injury, and by geologic environment in order to target appropriate sub-populations for specific interventions and risk reduction. Population trends can be monitored in the event of future earthquakes and these data may also serve as a comparison for other urban communities. These data will be made available to other researchers (with individual identifiers dropped to ensure confidentiality of medical data), policy makers, seismologists, engineers, geologists, health care providers, and community-based organizations in the interest of promoting community health and preventing injuries that might occur in future earthquakes. IVPP may also disseminate information through press releases regarding preventable injuries, ongoing reporting in the form of the Los Angeles County Department of Health Services' Public Heaith Letter (distributed to 24.000 health care providers), the Southern California Hospital Council, and the Southern California Trauma Society. Publication o f methodology and results of analysis is a key priority for the researchers. Papers will be submitted to appropriate journals and professional societies for publication and presentation. Billie Weiss. M.P.H.. was one of the principal investigators, and Maya Mahue. M.S.. was the lead epidemiologist for a collaborative study that is near completion addressing injuries resulting from the 1994 Northridge earthquake. This study was funded in part by the Centers for Disease Control and Injury Prevention in Atlanta. This consortium involved four components: a) assessment of injuries presenting at emergency department facilities (performed by IVPP), b) assessment of injuries presenting at emergency department facilities that were admitted to a hospital (performed by the Southern California Injury Prevention Research Center), c) random sample telephone survey of individuals in Los Angeles County to estimate the proportion of individuals that sustained injuries and the proportion of those injured that sought medical care 344 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (performed by University of California Los Angeles Department o f Community Health), and d) linking patients admitted with earthquake-related injuries to structural data (performed by University o f California Los Angeles Department of Engineering and EQE International Engineering). The Los Angeles County Department of Health Services IVPP has an extensive history and expertise in measuring and monitoring the health of the Los Angeles population and is a well- recognized leader in the field of injury epidemiology and surveillance. The scope o f work covered by IVPP under this protocol involved review and analysis of available emergency department log data in order to estimate the impact of injuries on emergency department facilities after the Northridge earthquake. Preliminary results are depicted in Appendix 4. Through this consortium, a working relationship has been established with EQE International Engineering. Since the volume of data obtained through emergency department logs is larger than any other source. EQE has expressed an interest and committment to provide IVPP with pertinent structural and geologic information in exchange for injury data to enhance their current casualty models (see Appendix 5). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 345 Institutional Qualifications. LAC-DHS has available mainframe computing resources, personal computing hardware, statistical analysis and modeling software, relational data management software, desktop publishing software, mapping software, internet access and publishing, slide-making software and hardware, laser printing (color and black/white), ample office space and associated equipment including fax and telephone service, and an extensive public health library with access to all major on-line literature indices. In addition. LAC-DHS has biostatisticians, statistic analysts, and academic faculty available for technical assistance and consultation. All resources necessary to complete the research are available at LAC-DHS within the Injury and Violence Prevention Program. IVPP has established contacts over the past 9 months with key administrative staff at a sample of 19 facilities concentrated in the San Fernando Valley. Of those facilities, data has been obtained from 12 facilities. Regular telephone verification of the contacts' status in administrative roles is conducted. Some facilities are linked by mainframe to LAC-DHS. therefore if the data is housed electronically, transfer is convenient. IVPP attends consortium meetings on a monthly basis with representatives from EQE, UCLA Department o f Community Health. UCLA Department of Engineering. UCLA-Harbor Medical Center. Los Angeles County Community Based Outreach and Assessment Team, and the Southern California Injury Prevention Research Center. In addition, communication between IVPP and the Center for Environmental Design Research (CEDR - University of California at Berkeley) occurs on a regular basis (monthly). The CEDR has served as liaison between many engineering, geophysics, and public health researchers. IVPP is approached regularly as a body for dissemination of injury information (including earthquake-related injuries) by many types of agencies including the local Red Cross. Los Angeles County Emergency Operations Center, local colleges and universities (including University of California Los Angeles and University of Southern California), individual researchers that focus on earthquakes, law enforcement, the media, and community-based organizations. One of the traditional roles o f the health department is to provide summary information to researchers for special studies, and IVPP has been performing this function since its inception in 1991. Project Management Plan. (See following page) 346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. T his proposal is n o t a con tin u ation p roject, th erefore no funding has been previously p rovided by U SG S. References 1 .L itan. I.R . P hysical d am age a n d h u m an loss: th e econom ic im pact of earth qu ake m itigatioa m easures in M em p h is, Tennessee: C en tra l U n ited S tates E arth q u ak e C on sortiu m : (M em p h isV . Proceedings. 1993 N ational E a rth q u a k e C onference: 1993.1.571-379. 2.0kada, S . S tu d y on th e evalu ation of seism ic casualty risk potential in dw ellings: P a rt I - In d oor zoning m ap on dwelling space safety. J o u rn a l of S tru ctu ra l a n d C onstruction Engineering ( 1993). T ransactions of A 1 J , 4S4.39-49. 3. Y in, Z . A m ethod to p red ict a n d classify earth qu ake d isasters. E arth q u ak e Research in China f1993V 7:1. 53-68. 4. Em m i. PC. A H orton , C .A . A G IS-based assessm ent of earthquake p roperty dam age an d casualty r isk : S alt L a k e C ounty, U ta h . E a rth q u a k e S p ectra (19931.9:1.11-33. 5. D urkin. M .E . A T h iel, C .C . Im p rovin g m easures to redu ce earthquake casualties. fim htW lkf S p ectra (19921.8:1. 95-113. -u - j Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. b. C obum . AW . Spence. R .J . P om onis. A . F actors determ ining hum an casualty levels in earth q u ak es M ortality pred iction in building collapse. P roceed m gs of th e T en th W orld C onference on E arth q u ak e E ngineering. 19-24 J u ly 1 Q 9 2 . M a d rid (1992). A A Balkem a. R otterd am . V ol 1 0 . 5989-5994 7. M u rak am i. H .O . Sim ulation m odel to estim ate h u m an loss for occupants of collapsed buildings in a n earth q u ak e. Proceedings of th e T en th W orld conference on E a rth q u ak e E ngineering: 19-24 Ju ly 1992. M ad rid (1992). A A B alkem a, R otterd am . V ol 10. 5969-5974 3. Shiono, K . .A ssessm en t m odel for earth q u ak e fatalities. Proceedings of the Second U S-Asia C onference on E ngineering for M itigating N atural H azard s D am age. Y ogvak arta. In d on esia. June 22-26.1992 (1 992), A .N .L . C hiu & A S . D anuatm odjo (eds), Publisher, p lace of publication unknow n. E10-1 -El 0-8. 9. S am ardjieva, E . & O ike, R . M odelling th e num b er of casualties from earthquakes. Jou rn al of N atu ral D isaster Science (1992). 14:1,17-28. 10. M u rak am i. H .O . A sim ulation m odel to estim ate h u m an loss for occupants of collapsed buildings in a n earth q u ak e. Proceedings of th e T en th W orld conference on E arth q u ak e E ngineering (1992), A A B alkem a (R otterdam ), 10,5969- 5974. 11. D u rk in , M .E . St T hiel. C .C . E stim atin g casualties in earth q u ak es: an assessment Proceedings of th e F o u rth Intern ation al C onference on Seism ic Z onation. S tan ford U niversity. A ugust 25-29.1991 (1991), E a rth q u ak e E ngineering R esearch Institu te. O ak lan d , CA III. 293-300. u > o o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 12. Shiono. K , R rim gold, F , & O h ta, Y. M eth od for th e estim ation of earthquake fatalities an d its ap plicability to th e global m acro-zonation of hum an casualty risk . P roceedings of the F ou rth In tern ation al C onference on Seism ic Z onation. S tan ford U niversity. A ugust 25-29.1991 (1991). Ill, 277-284. 13. Steinbrogge, K . V . C om m ents on th e h istory of earth q u ak e casualty estim ation. W ork shop o" nnw kitm g earth qu ake casualties for planning a n d respon se: A silom ar C onference C enter. Pacific, CA. D ecem ber 4-6.1990.1 -5 14. Stojanovski, P . & D ong, W . S im ulation m odel for earth q u ak e casualty estim ation. R isk M anagem ent Softw are, Inc., 149 C om m onw ealth D r., M enlo P a r k . C A 94025. I S. N oji, E . T he role of epidem iology in seism ic vu lnerab ility red u ction : A n interdisciplinary ap p roach . TheC usac Joumil(l994), 8-9. 16. Dewey, J.W ., R eagor. B.G., D engler, L .. & M oley, K . Intensity distribution an d isoseism al m aps for th e N orth ridge C aliforn ia earth qu ake of J a n u a ry 17,1994. U .S. Geologic Survey O pen F ile R ep ort ( 1995195-97. 17. U nprocessed d ata from U niversity of South ern C aliforn ia (Los Angeles). U nited S tates G eologic Survey, an d th e C alifornia D ivision of M ines an d G eology (1996). 18. E vernden, J .F . & T hom pson, J .M . P redictive m odel for im p ortan t ground m otion p aram eters associated w ith la rg e an d great earth q u ak es. U nited States G eologic Survey. B ulletin 1838 (1988). 19. E guchi, R . & C hang, S . Losses associated w ith building dam age in M em phis. Proceedings of the 11 th W orld C onference on E arth q u ak e Engineering. Ju n e 23-28,1996, A capulco, M exico. V O Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. 20. Tobui. T.. Davis. J.. Eguchi. R... Nathe, S Northndge earthquake case study Fifth International Conference on Seismic Zonation. October 17-19. 1995. Nice, French Riviera, France. 2 1. Eguchi. R. Mitigating nsks to infrastructure systems through natural hazard reduction and design. Proceedings of the International Symposium on Public Infrastructure Systems Research. CK Choi & J Penzien. Eds., Sept. 25-27, 1995 Seoul. Korea. 22. Eguchi, R., Goltz. J , Seligson, H. The application of new technology Northndge Earthquake. Lifeline Performance and Post Earthquake Response. Anshel J. Schiff Ed Technical Council on Lifeline Earthquake Engineering, Monograph No. 8. August 1995 23. Eguchi. R. Seligson. H., Goltz. J. Rapid post-earthquake damage assessment for lifeline facilities. 6th U S.-Japan Workshop on Earthquake Disaster Prevention for Lifeline Systems, iulv 18-19,1995, Osaka City, Japan. 24. Blais, N., Seligson. H. Petrow, A. J. Use of rapid damage assessment and geographic information systems for emergency response in the Nonhridge earthquake. Proceedings of the 11th World Conference on Earthquake E ng in eerin g . Ju n e 2 3 -2 8 , 1996, A capulco, M exico. 25. Chang, S. & Shinozuka, M. Life cycle cost analysis with natural hazard nsk. National Seismic Conference on Bridges and Highways: “Progress in Research and Practice". December 10-13,1995, San Diego, Federal Highway Administration. California Department of Transportation. 2 6 . Chang, S. Building stock age distribution and seismic risk assessment 42nd North American Meetings o f the Regional Science Association. Cincinnati. Ohio, November 9-12.1995. u » C O O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 27. Jones, B. & Chang, S. Economic aspects of urban vulnerability and disaster mitigation. Urban Disaster Mitigation, The Role of Engineering and Technology. F. Y. Cheng & M.S. Sheu, Eds., Oxford: Elsevier Science Ltd., 1995,311-320. 28. Seligson, H.. Eguchi R., Goltz, J. Seismic vulnerability assessment for Southern California. Proceedings of the Fifth U S National Conference on Earthquake Engineering Chicigo. Illinois, July 10-14,1994. 29. Seligson, H. & Yanev, P. Post-earthquake damage inspection o f buildings - California practice and the Loma Prieta experience. International Seminar on Post-Earthquake Emergency Damage and Usability Assessment of Buildings. Athens, Greece, September 22-24,1993. 30. Seligson. H., Eguchi, R., Tierney, K . A methodology for assessing the risk of hazardous materials release following earthquakes - A demonstration study for the Los Angeles area. Fourth U.S.-Japan Workshop on Fm hginkf Disaster Prevention for Lifeline Systems. Los Angeles. CA. August 19-21.1991. 3 1. Dewey, J.W., Reagor, B.G., Dengler, L , & Moley, K. Intensity distribution and isoseismal maps for the Northndge California earthquake of January 17,1994. U.S. Geologic Survey Open File Report ( 1995) 95-97. 3 2 . Unprocessed data from University of Southern California (Los Angeles). United States Geologic Survey, and the California Division o f Mines and Geology (1996). 33. Everoden, J.F. < f c Thompson, J.M. Predictive model for important ground motion parameters associated with large and great earthquakes. United States Geologic Survey. Bulletin 1838 (1988). Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Appendix 1. Protocol for Medical Record Data Abstraction ► The medical record abstraction instrument should be completed as thoroughly as possible based on the information in the medical record. ► Check-boxes are provided for most fields. If specific information is requested and available, or if 'Other' is the selected response, please print (as clearly as possible) the specific information in the space provided. ► The IVPP medical record abstraction tool equates missing to unknown, with the exception of data relating to intention of injury. If the intention of injury is denoted specifically in the medical record as being of undetermined intent, please check that option on the instrument If there is no mention of intent please check the option 'Unknown'. For information not found on the medical record, choose the option 'Unknown'. » Upon completion of abstraction, check the field that indicates that the specific medical record has been abstracted. Upon completion of abstraction for that day, return all the medical records to the administrator or contact person for that facility. NEVER REM OVE M EDICAL RECO RDS FROM THE FACILITY. ► Medical records are confidential. Although abstractors have signed statements ensuring that confidentiality will be maintained, one cannot emphasize the importance o f this matter. Abstractors should keep close track of all abstracted forms and of all medical records. Breaching confidentiality is grounds for dismissal and may result in other serious civil action. u > tn N » Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Data Management and Quality Control Emergency Department Log Data Transfer Protocol ► Logical checks are performed including verification of the file structure so that the cumulative file structure remains consistent, verification that the transfer data file contains material for dates reflected in the file name, and that the number of records copied was complete. Data Transfer Process 1 . Determine that the data source is from the same facility as the destination file. 2. Record the destination filename on the data transfer log. 3. Record the source filename, laptop identification, and initials of the data entry person in the data transfer log. 4. Verify that the file structure of the incoming file is the same as the destination file. 5 Record the number o f records from the destination file (prior to appending) in the data transfer log. 6. Record the number o f records from the incoming file in the data transfer log. 7. Append the field data (from the floppy diskette) to the destination file. 8. Verify that the total number of records equals the sum o f the destination file prior to appending. 9. Perform the following global entries: a) data entry person's initials, b) facility code. 1 0 . B ack-up th e new ly ap pend ed destination file. Cleaning and Coding of Data ► Data was cleaned by inspecting all entries for injury type, ensuring that field-specific data were within acceptable ranges, and that no duplicate entries were apparent » Obvious non-injuries are housed in a separate file. Chest pain without mention of injury was not considered an injury, nor were non-injury specific infections, post-operative bleeding and pain, noo-industrially related carpal tunel syndrome, and non-specific pain. Clinician's diagnosis supercedes patient complaint Data Cleaning Process 1 . Perform a global change to convert all the characters to upper case. 2. Code missing numeric data u '9 and missing character data as *Z\ 3. Generate a list of missing values for the log date, log number, and medical record numbers. 4. Investigate duplicate entries, Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Appendix 2 Data Coding Process Based on the previously defined abbreviations, data was grouped and coded The original entries were maintained and a code field for each variable was added, or a code created in SAS. Some codes contain more than one abbreviation, but the grouping is logical for the current study. In future studies, the data may be re-coded based on the original entries if the current grouping is deemed undesirable. The following codes are associated with body location of injury: l=Head/Neck 5=Chest 2=Upper Extremity 6=Abdomen 3=Lower Extremity 7=Trunk 4=Back 8=Unknown S'ote: In most analyses, 5 IChest) and 7 (TrunkI are considered together. In preliminary analyses, the groupings for injury type were collapsed as follows: Bum/Chemical Exposure: Bum. Chemical/Toxic Exposure Crush Follow-Up. Reecheck Injury. Wound Dislocation: Fracture. Sprain/Strain Animal Bite/Scratch: Derangement. Foreign Body; Human Bite/Scratch; Ingestion of Item; Insect Bite/Sting; Nerve Injury; OD: Drug/ETOH; OD ETOH; OD: Non-Prescnption Drug; OD: Other. OD: Prescription Drug; OD: Street Drug, OD: Unknown; Other. Pain; Perforation; Rape/Sexual Assault; Rupture: Soft Tissue Injury; Whiplash: Superficial/Minor: Abrasion/Cut/Scratch; Avulsion/Tear. Bruise/Contusion; Hematoma; Laceration; Puncture Wound Trauma: Amputation; Blunt Trauma; Concussion: Severed; Trauma Unknown: Unknown Crush: Follow-up: Non-Specific: Orthopaedic: Other: U l L A Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. The following codes are associated with ethnicity/race: I = White, Non-Hispanic 2=White, Hispanic 3=African-American 4=Natwe American 5=Asian-Pacific Islander 8=Other 9=Missing IO=Unknown Nole: Most emergency department logs did not include this level o f detail. If'e allow the opportunity to distinguish between missing and unknown, but most o f the data is missing. f t Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Appendix 3 LOS ANGELES COUNTY DEPARTMENT OF HEALTH SERVICES INJURY AND VIOLENCE PREVENTION PROGRAM EMERGENCY DEPARTMENT FORM Facility Name Abstracted By Sequence No __________________ Code Abstraction Date I Medical R ecord No ID/PF No _______ Billing No _______ A. DEMOGRAPHIC INFORMATION 1 Name (Laat). 2 Street A ddress 3 City (Number) (Direction) 6 Telephone No (______)______ •________ a Sex 01 □ Female 02 □ Male 900 Unknown 1 1 Race/Ethmc<y 01 □ W hte, Non-Hispamc 02 0 W M e, H ispantc 03 □ African American 12. Payment Type. (Check all that apply) 01 □ Private Insurance (Type: _ 0 2 0 Self Pay/Cash QSOMedt-Cal 04 □ Medicare (F irst!. (Middle). (Name) 4 State (Street Type) (Apt J/RmM) 5 Zip Code ________________ 7 Social Security No . 9 Age __________ 10 Birthdate: 04 □ Native American 06 □ A sian-PactAc Islander 88 □ Other (S p ecify ________ 98 □ Unknown 05DHM O 08 □ Worker's Compensation 07 OHosp4e* Employee 88 □ Other (Specify________ 98 □ Unknown in O n Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. B. EMERGENCY DEPARTMENT VISIT 13 Date of EO Arrival I 14 Time of ED Arrival 15 Tima of ED Erdl: (Miliary Time) 16 Mode of ED Arrival: 01 □ Saif (drive or walk-in) 02 □ Private vehicle (someone else brought in) 03 □ Police (Miliary Time) 17. Disposlion from ED: 01 □ Discharged home 02 □ Discharged to PMD 03 □ Admitted 04 □ Exprred/Morgue 06 □Incarcerated 06 □ Left w/out being seen (Walk-out) 04 □ Ambulance (Provider/Rescue# 06 □ Helicopter (Prouder/Rescue# _ 88 □ Other (Specify.______________ 98 □ Unknown 07 □ Transferred to another fa d ty (Fee 08 □ Left against medical advice (AMA) 88 □ Other (Specify._________________ 98 □ Unknown Incident-Specific Information IB. Date of Incident: I I 20. Property Use: 01 □Hom e 02 □ Residential Insttution 03 □ Place of Recreation/Sport 04 □ Public Building (Non-School) 06 □ Street/S idewalk OSaFreewayWighway 19. Tima of Incident: 0 7 a H o sp ta l 0 6 0 School O B □ Factory/Warehouse 10 □ IndustnaVConstruction S le 88 □ Other (Specify________ 98D Unknow r (MWary Time) 21. Location of Incident: (Street Name/Cross Streets/location) 2 2 W as pabart ertncated from incident location? 01 □ Y » 02 □ No 23. Did the ando* occur on the job? 0 IQ Y c e 0 2 Q N o 99 □UbkacMD (C «y) 99 □ Uiksown new U ) c /» iliNsician DiagDiKisilCD-') C ixle A Dcscrqilunl £ c la £ % a § J C 'J 1 t s e f i B o z 8 > □ t B 0 9 } t 1 n w * 5 1 5 - o Z 8 > □ " 8 B a 9 $ f 1 1 1 □ □ o r * * o s & > O « O 1 * I II □ □ 0 0 & > X C “ 8 i ill 3 - b a o o - 3 S S Ul B = 3 " 8 ? £ 3 - m , a □ □ ' - N « g o o e 3 6 c 358 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. II External Cause of Injury (regardless of intent) 01 □ MV accident - Traffic |Golo 321 02 □ MV acadml - Ncnlrallic |G oto 321 03 □ MV accident - Other (____________ 04Q Firearms |Go to 331 05 0 Falls |Go to 34) 06 □ PoumBg |Go to 331 07 □ Nmaal/Envmnniaiul (Go to 36| 08 □ Cutting or picrcmg object (Object: _ _j (Goto 321 09 □ Sbucfc by/agaad object ■ Household item 10 □ Strut* by/agam* object - Other ( ________ 11 □ Cau^u tn/betvieeri object (Object: I h o w n m i y S u b m e r a t i i 1 3 □ S u l l o c a n o n 14 □ Stixifpilaun/llmgmg 13 □ Fire/Bums 1 6 □ F ig h t/B r a w l 17 □ Machinery 18 □ llimiadc/Assault 19 □ Sutctde/Sutctde Attempt _ ) 20 □ Child Abuse 2 1 □ R^ie/Sexual Assault _) 22 0 Overextrucn J 88 □ O lh a ( Specify:_________ 9 9 □ L U tn o w n J 12 □ 32. Motor VdtideInjuredPcnaa; (From 31.uniats01.02 *td031 01 □ Mater vchide operator 03 □ Motcrcydisl 02 □ Malar v etude pasaaigcr 04 □ Pedal cydut 33. TypcafWc^ran: |Fram31.option041 01 □ HmdgmiptsuA/revolver 03 □ Hintng Rifle 02 □ 9iatpn 04 □ MiliUry Rifle 34. Circum*mcetf Fall: (Frt*t»31.opUcn03| 01 □ Fall while in t m p t h 02 □ Fall from ladder/scaffolding 03 □ Fall from *am/*aps 04 □ Fall front buildhig 03 □ Fall from slgipngArgipng 03 □ Pedeautan 88 □ Other (Speedy: 99 □ Utknown 03 □ Am i I t Wcaptn 88 □ Other (Specify: _ 99 □ (Jhknmwn 06 □ Fall from puimg/dtoviig 07 0 Fall from bed 08 □ Fall from natural ate (hill, etc) 88 □ Other (Specify:_____________ 99 □ IJhkntavn 33. Poutnmg Events: (From 31. optim06| Check the man appliceblemd describe ago*. 01 □ Drugs (Dlidi):_________________________________________________ 02 □ Dmgi (Metbcmal): 03 □ Akxdial:________ 04 □ Chennai:_______ 0 3 0 Gases:__________ 0 6 0 Food:__________ 0 7 0 Pi**:__________ ui U l VO 08 O Insect/Animal Bitc/Sfag:. 8SO Other___________ ^ 99 0 bhknown R e » H * 4 itrvai Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. April 10. 1996 R H k ■ Stiffty • Dnign EnftMffing Mi. Billie Weiii Director of Injury and Violence Prevention Program Department of Health Services County of Los Angeles 313 N. Figueroa Street Lot Angeles. CA 90012 Subject: USGS Proposal on Collection o f Northndge Earthquake Casualty Data Dear Billie: EQE International is pleased to submit this letter of intent to participate in the subject study should Los Angeles County be awarded its research grant from the USGS. As you know, we have been actively involved in the collection of building damage data resulting from the January 17,1994 Notthridge earthquake. Much of this information was collected under contract to the State of California through its Office of Emergency Servicer As part of the proposed study, EQE will make available this data in a format that allows correlation of injury data from the earthquake with building damage information. The complete scope*of-work for our effort is attached as a supplement to this letter. o\ o Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. In addition to this »cope-of-work, we have assigned person-hourt and estimated costs associated with this effoit. In total, our budget is for $9,000. Our schedule for completion will be commensurate with the major milestones set for other tasks within your proposal. As part of this effort, we understand that aggregated data on number of injuries or ir\jury rates (on a census tract level) will be provided to EQE for its use in enhancing its current casualty models. We appreciate the opportunity to be a part of your proposal effort and hope that you are successAil in receiving your award. Should you have any questions regarding any of the above, please ftel free to contact either Hope Seligson, Nail Blais or myself. Sincerely, EQE International, Inc. (ft<2- , Ronald T. Eguchi cc: Hope Seligson, CAPR Vice President and Center Director Neff Blais, CAPR Center for Advanced Planning and Research Maya Mahue, LA Co U ) 0\ LtkatorcTM ’ c n • 1*101 V onK inM »Avtmu*. Suto400 ■ Irvfcw.CA VZ71S-IQU USA • TakptmwOltilSM JOS • FAX(7IOttt4M9 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. United States Df pan Ten of the Interior Or. jB illie W eiss Los 'Angeles County Dept of Health Services Injury & Violence Prevention Program 313 ;N . Figueroa. R o o m 127 Los Angeles. C A 90012 I D ear! Dr. Weiss: i Enclosed are the following: I a. Award 1434-HQ-97-GR-03163: | b. Standard Form (SF) 269A. Financial Status Report; j c. S F 270. Request for Advance or Reimbursement: I d. Standard Form (SF) 3881. Payment Information Form - A C H Vendor i Paym ent System: and ' e. Assurances. U.S. GEOLOGICAL SI ,'.VEY R o a v , V i t g i n i , 2 2 0 9 V TnReply Refer To: Mail, StoD 205A MB 2 5 i«L Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Participation in the Automated Clearing House (A C H ) System program has becom e mandatory. The Payment Information Form (SF 3881) permits the Government to wire transfer payments to your financial Institution. I f you are not already a participant in the A C H Vendor Paym ent System, you must complete the "C om pan> Information" section of the form. Your financial institution must complete the "Financial Institution Information" section of the form. O nce the |form has been completed, either you or your financial in s titu t'o r must forward che completed form to the U.S. Geological Survey at the address in the !"Agency Information" section of the form. All payments for th". award willj be processed in th is manner. Plea|se have the Assurances signed and returned to the undersigned within 10 calendar days from receipt of th is le tte r. If you have any question* . please call; m e at 703-648-7382. i yjicerely I ' m u i i K I k a d t i i 'Pansy RflYeatts A Contrawing Officer 5 Enclosures Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. OAFA/PBA (Revj. 1 0 /8 9 ) i GRAN' \ W w R D U.S. Geological Survey 1 RECIPIENT: 3 Los AngeTes County Department of Health Services 1 Injury and Violence Prevention Program ® 313 N . Figueroa - R o o m 127 -n Los Angelas. C A 90012 c c d ; ^ t 2. A P P L IC A T IO N /P R O P O S A L TITLE: Risk and Loss Associated with Geological. Societal. and Structural Hazards Related to F.arthquakes in Urban Los Angeles County, dated April 11. 1996, am ended February 19, 1997. 1 3. A W A R D A M O U N T : S60.000 Q . c a o 3 T " 5 4. A W A R D N U M B E R : 1434->4Q-97-Gft-03l63 5 . P A Y M E N T M E T H O D : See attachment A i 6. B U O G E T P E R IO D : 6/1/97 through 5/31/94 g ; ;. P R O J E C T P E R IO D : 6/1'V/ through 5/31/98 | 8. U S G S A D M IN IS TR A TIV E C O N T R A C T IN G O FFIC E R : | Margaret S. Broadwater ° U.S. Geological Survey - S' Office of Acquisition & Federal Assistance. M S 205A 3 12201 Sunrise Valley Drive f Reston, V A 20192 1 Telephone Noi.: 703-648-7484 1 3. U S G S P R O J E C T OFflCEft: Dr. John Sim s U.S. Geological Survey Geologic Division -- M S 905 12201 Sunrise Valley D^ive Reston. V A 20192 Telephone No.: 703-648-6722 u > £ Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. 10. S PE C IAL C O N D ITIO N S : ' A. The Recipient’s application, bearing the t l t k and date shown in Biocc 2. is 1* - orporated here'n by reference. B . Thei Principal Investigator for this award .s B illie P . Weiss (213) ’;-'Ti-7785. C . This awjrd shall be administered in accerdai .* .e with Attachment A. ^ c i a l Term s and Conditions, and Attachment B. General Provisiors. 11. A W A R D A U T H O R IT Y : Public Law 95-124 ! Public Law 104-208 W A U T H O R IZ E D U S G S G O V E R N l £ N T OFFICIAL: U t o\ U t r REQUISITION NUMBER: 97-7107-0390 12: A C C O U N T IN G A N D A P P R O P R IA T IO N D A T A : 97-7160-09201 O C 4120 $60,000 APPENDIX 2: Pilot Study Medical Record Abstraction Instrument Real-Time Facility Survey Form Post-Event Facility Survey Form 366 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. LOS ANGELES COUNTY DEPARTMENT OF HEALTH SERVICES INJURY AND VIOLENCE PREVENTION PROGRAM EMERGENCY DEPARTMENT FORM Facility Name: Code Medical Record No.: Abstracted Bv: Abstraction Date: / / ID/PF No.: Sequence N o: Billina No.: A . D E M O G R A P H I C I N F O R M A T I O N 1 . Name: ( L a s t ) ( F i r s t ) ( M i d d l e l 2. Street Address. (Number) (Direction) (Name) (Street Typo) (Apt #/Rm 0) 3. City 4 State: 5. ZioCode; 6 . Telephone No . (__ __) 7. Social Security No.: 8. Sex: 01 □ Female 02 □ Male 99 □ Unknown 9. Aqe: 10. Birthdate: / / 11. Race/Ethnicity: 01 □ White, Non-Hlspanic 02 □ White, Hispanic 03 □ African American 12. Payment Type: (Check all that apply.) 01 □ Private Insurance (Tvoe: 02 □ Self Pay/Cash 03 □ Medi-Cal 04 □ Medicare 04 □ Native American 06 □ Asian-Pacific Islander 88 □ Other (Soecifv 1 99 O Unknown ) 0 6 0 HMO 06 □ Worker's Compensation 07 □ Hospital Employee 88 □ Other!Soecifv: ) 99 □ Unknown U ) On Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. LAC-DHS IVPP Earthquake Injury ED I Med Record Form (Continued) B . E M E R G E N C Y D E P A R T M E N T V I S I T ......... - ~ ’ I 13. Dale of ED Arrival:____ / _____ / __________14 Time of ED Arrival:________________15. Time of ED Exit.______________ (Military Time) (Military Time) 16. Mode of ED Arrival: i 01 D Self (drive or walk-in) 04 O Ambulance (Provider/Rescue#:___________________ ) 02 □ Private vehicle (someone else brought in) 06 □ Helicopter (Provider/Rescue#:____________________) 03 □ Police 88 D Other (Specify:________________________________ ) 99 □ Unknown 17 Disposition from ED: 01 □ Discharged home 02 □ Discharged to PMD 03 □ Admitted 04 □ Expired/Morgue 06 □ Incarcerated 06 □ Left w/out being seon (Walk-out) 07 □ Transferred to another facility (Fac : 08 □ Left against medical advice (AMA) 88 □ Other (Specify:_________________ 90 Q Unknown Incident-Specific Information 18 Date of Incident: / / 19. Time of Incident: 20. Property Use: 01 □ Home 02 □ Residential Institution 03 □ Place of Recreation/Sport 04 □ Public Building (Non-School) 06 □ Street/Sidewalk 06 □ Freeway/Highway 21. Location of Incident._________________ (M ilitary Time) 07 □ Hospital 08 □ School 09 □ Factory/Warehouse 10 □ Industrial/Construction Site 88 □ Other (Specify:___________ 99 Q Unknown (Street N am e/C ross Streets/location) 22. W as patient extricated from incident location? 01 n Yes 02 □ No 23. Did the incident occur on the job? 01 □ Yes 02 □ No (City) 99 □ Unknown 9 9 0 Unknown 24. Complaint (Abbreviated & Describe): 0 1 02 03 u > 0\ 00 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. LAC-DHS IVPP Earthquake Injury ED I Med Record Form (Continued) Circumstances of Injury: 25. Physician Diagnosis (ICD-9 Code & Description): 0 1 _____________ . ________( D e s c r i b e : _______________________ 0 2 ____________ . ________( D e s c r i b e : _______________________ 0 3 ____________ . ________( D e s c r i b e : 99 □ Unknown 26. Pre-existing Medical Conditions: (Describe) 0 1 02 00 99 □ Unknown 27. Earthquake-relatedness was noted? 01 Q Yes 02 □ No Intarv-Speclflc Information 28. W as the Injury sports related? 01 O Yes 02D N o 99 □ Unknown 29. Location of Injury on Body: (Check all that apply.) 01 □ Head/Neck 0 4 0 Back 02 □ Upper Extremities 06 □ Chest 03 □ Lower Extremities 06 □ Abdomen 07 □ Trunk 88 □ Other (Specify:. 99 □ Unknown ui O n R ep roduced with perm ission of the copyright ow ner. F urther reproduction prohibited w ithout p e rm is s io n . LAC-DHS IVPP Earthquake Injury ED / Med Record Form (Continued) 30. Intention of Injury: (Check all that apply.) 01 □ Unintentional 04 □ Legal Intervention 88 □ Other (Specify:_________________________________) 02 □ Suicide/Self-inflicted 05 □ Late Effects 99 □ Unknown 03 □ Homicide/Assault 06 □ Undetermined Intention 31. External Cause of Injury (regardless of intent): 01 □ MV accident - Traffic (Go to 32] 02 □ MV accident - Nontraffic (Go to 32] 03 O MV accident - Other (____________ 04 □ Firearms (Go to 33] 06 □ Falls (Go to 34] 06 □ Poisoning (Go to 36] 07 □ Natural/Environmental (Go to 36] 08 □ Cutting or piercing object (Object:. J (Go to 32) 09 □ Struck by/against object - Household Item 10 Q Struck by/against object - Other (___________________ 11 □ Caught in/between object (Object:_________________ 12 □ Drowning/Submersion 32. Motor Vehicle Injured Person: (From 31, options 01,02 and 09] 01 □ Motor vehicle operator 03 □ Motorcyclist 02 □ Motor vehicle passenger 04 □ Pedal cyclist 33. Type of Weapon: (From 31, option 04] 01 □ Handgun/plstot/revolver 09 □ Hunting Rifle 02 □ Shotgun 04 □ Military Rifle 34. Circumstance of Fall: [From 31, option 06] 01 □ Fall while In transport 02 □ Fall from ladderfscaffolding 03 □ Fall from stalrs/steps 04 □ Fall from building 06 □ Fall from slipping/tripping 13 □ Suffocation 14 □ Strangulation/Hanging 1 5 0 Fire/Burns 16 □ Fight/Brawl 17 □ Machinery 18 □ Homicide/Assault 19 □ Suicide/Suicide Attempt 2 0 0 Child Abuse 21 □ Rape/Sexual Assault 22 □ Overexertion 88 □ Other (Specify:________ 99 o Unknown 05 0 Pedestrian 88 □ Other (Specify:. 99 o Unknown 06 □ Assault Weapon 88 □ Other (Specify: _ 99 □ Unknown 06 □ Fall from pushing/shoving 07 O Fall from bed 08 □ Fall from natural site (hill, etc.) 88 □ Other (Specify:_____________ 99 0 Unknown ut > 4 O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. LAC-DHS IVPP Earthquake Injury ED I Med Record Form (Continued) 35. Poisoning Events: [From 31, option 06] Check the m ost applicable and describe agent. 01 □ Drugs (Illicit):_________________________________________________________________________________________ 02 □ Drugs (Medicinal):____________________________________________________________________________________ 03 □ Alcohol:_____________________________________________________________________________________________ 04 □ Chemical:__________________________________________________________________________________ ________ 06 □ G ases:_____________________________________________________________________________________________ 06 □ F ood_______________________________________________________________________________________________ 07 □ Plant:_______________________________________________________________________________________________ 06 □ Insect/Animal Bite/Sting:_______________________________________________________________________________ 6 6 O Other:______________________________________________________________________________________________ 99 □ Unknown 36. Natural/Environmental Events: (From 31, option 07] 01 □ Exposure, excessive cold 04 □ Exposure, chemical 07 □ Abandonment/Neglect 02 □ Exposure, excessive heat 06 □ Lack of food 86 □ Other (________________________________) 03 □ Exposure, radiation 06 □ Lack of water 99 □ Unknown Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Real- Time Facility Survey Form i OFPARIMENT OF HEALTH SERVICES UBRGENCV OPERATIONS CENTER COMMUNICATIONS SUPERVISORIAL DISTR IC T 1 MAP 676EI 8MM 594A4 63507 636IH 63403 594A4 63503 600J6 594E6 635A4 PARAMEDIC RECEIVING _ REDOINET HOSPITALS BEVERLV HOSPITAL CALIFORNIA MEDICAL CHI CHILDRENS HOSPITAL-LA " EL A DOCTORS HOSPITAL GARFIELD MEDICAL CTR. GOOO SAMARITAN HOSP. KAISER-LOS ANGELES LAC/USC MEDICAL CENTER POMONA VALLEV HOSPITAL QUEEN otANGVHWP PRES WHITE MEMORIAL SUB-TOTAL RG 3 1 1 T 5 I ?' i s' i i TELEPHONE (213)725-4225 {213)746-2411 (2i 3)660-2450 (213)266-5514 (616)307-2125 (213(977-2423 (213)667-401? (213)226-670^ (714)685-9500 (213)413-3000 (213)268 5000 & L , i> NON-PARAMEDIC RECEIVING [m a p REDOINET HOSPITALS RG TELEPHONE 640C3 LANTERMAN STATE HOSP. 5 (714(595-1221 634C7 ORTHOPEDIC HOSPITAL I (213)742-1000 X - 634C2 ST.VINCENT'S MED. CTR 1 (213)484-7301 594B7 TEMPLE COMMUNITY HOSP. 1 (213)362-7252 SUBTOTAL — lOTAl THISSHEE1. I S L I T -4 N > (H i k _Lfc. DO A : l l: i X 15= . i irzx: Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Real-Time Facility Survey Form i •artm ent o f h ealth se r v ic e s e m e r g e n c y o p e r a t io n s c e n t e r COMMUNICATIONS SUPERVISORIAL DISTRICT I MAP PARAMEDIC RECEIVING HEAR ONLY RG TELEPHONE /> 6J7C4 GTR. EL MONTE HOSPITAL S (s iq s t o - ttt/ 636A4 MONTEREY PARK HOSPITAL S (810)670-5/51 63SFS SANTA MARTA HOSPITAL 1 (213)266-8500 ____ SUDTOTAi: J jL L Q t. ‘ OjQf) • / u j w o t 0 0 tM /} £T NON PARAMEDIC RECEIVING MAP HEAR ONLY RG TELEPHONE 601A4 CASA COLMA 5 (714)503-7521 634D1 CIGNA-LOS ANGELES 1 (213)484-3120 63SB3 NORRIS CANCER HOSPITAL 1 (213)224-6600 634G2 PACIFIC ALLIANCE HOSP. 1 (213)624-8411 676G4 PICO RIVERA COMMUNITY 3 (310)048-1121 635B3 USC UNIVERSITY HOSPITAL 1 (213)342-8500 SUBTOTAL: IOTAL IIIIS SHEEI u » ■ > 1 u > Post-Event Facility Survey Form northridge earthquake data review Hoscitai __________ __ _ _____________________ Contact Person______________________ Teieonone_________ Mosoitai Disaster Resccnse Group (drcte ona): * 2 3 4 5 6 7 8 sumoar of oauaras wen aannauaKe mmma ruunse/Snessee t£HB8 n F 1 1 iPH January i7 4:30 a.m .-8:00 a.m. 8:00 a.m .- 4 00 P.m. 4 00 p .m .-i2:00 miamgrtt January 18 12:01 a.m .-8:00 a_m. 8:00 a.m .-4:00 p.m. 4:00D.m. -12:00 micmgnt 12:01 a.m. - 8:00 a.m. January 19 at 8:00 a.m. - January 20 at 8:00 a.m January 20 at 8.00 a.m. - January 21 at 8:00 a.m. January 21 at 8:00 a.m. - w>anuary 22 at 8:00 a.m. Grana total: Please FAX or mail tnis fc los Angeies County EMS Agency 19951 Manner Avenue Suite 100 Torrance. CA 90503-1672 Attn. Miguel Ascarrunz -AX: 3101370-9332 T^enie you. for your ccooeraaon. &L -SC ■3- 4L. SL. 4E2 £ - 4 L 4 - £ L 4 . arm ov Merer 25 to: Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX 3: Statistical Tests for Assuming Earthquake Attributabilitv 375 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table A3-1. Pre-Earthquake Weekday Inlurv Visits Compared to Post-Earthauake Weekday Injury Visits Treated at 3 Hospitals in Los Anaeles Countv. January 1994. Facility M eans (Std. Dev.) Weekday______ Pre-Event________Post-Event p-value* Hospital &' Hospital ’ C' Hospital 'D' Tuesday 15.5 29 n.s.* (2.12) (12.73) Wednesday 14 19 n.s.* (1.41) (14.14) Thursday 16 11.5 n.s.* (7.07) (3.54) Friday 23.5 16.5 n.s.* (15.33) (6.36) Saturday 29.33 16.5 p < 0.05 (6.66) (0.71) (decrease) Sunday 21.33 14 n.s.* (2.08) (4.24) Monday 13.67 11 n.s.* (2.08) (1.41) Tuesday 50 59.5 n.s.* (4.24) (13.43) Wednesday 42.5 49.5 n.s.* (2.12) (0.71) Thursday 36.5 54 n.s.* (6.36) (0) Friday 40.5 44 n.s.* (4-95) (7.07) Saturday 40.67 48 p < 0.05 (5.51) (2.83) (increase) Sunday 39.67 32 n.s.* (7.77) (9.90) Monday 49 55 n.s.* (3) (0) Tuesday 19.5 29 n.s.* (3.54) (4.24) Wednesday 18.5 25.5 n.s.* (9.19) (10.61) Thursday 16.5 23 n.s.* (7.78) (2.83) 376 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table A3-1. (ContinuedI Means (Std. Dev.) Facility__________ Weekday______ Pre-Event________Post-Event p-valueA Hospital 'D' (Continued) F rid a y 2 6 .5 2 5 n . s . ' (3 .5 4 ) (4 .2 4 ) S a tu r d a y 2 3 2 6 p < 0 .0 5 (1) (0) ( in c r e a s e ) S u n d a y 22 3 4 .5 n .s.* (7) ( 3 .5 4 ) M o n d a y 17 2 1 n . s . ' (5 .2 9 ) (0) A F r o m m o d ifie d 2 - s a m p le t- te s t ( a s s u m e s u n e q u a l v a r ia n c e s ) : t = ( x ,- W ((s2,/n,)+(sVn2 )) d f= v v=((s2,/n,)+(s; 2 /n2 ))/[((s2l/n,)2 /(n1 -1 ))+((*?22/n 2) 2/( n 2-1))] * n o t s ta tis tic a lly s ig n ific a n t, p > 0 .0 5 _____________________________ 377 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-2. Differences between Frequencies of Injuries Treated at 3 Emeraenc\ Departments in Los Angeles County: 1 Week Post-Northridae Earthquake Compared to W eekday Averages for the R est for January. 1994 Hospital W eekday Date Frequency Comparison Average p-valueA Hospital 'B ' Tues. 1/18/94 38 17.0 <0.05 Wed. 1/19/94 29 12 3 <0.05 Thurs. 1/20/94 9 15.3 n.s.* F ri 1/21/94 12 22 7 <0.05 Sat. 1/22/94 16 26.3 <0 05 Sun. 1/23/94 17 18.8 n.s.* M on. 1/24/94 12 13.0 n.s * Hospital ’ C’ Tues. 1/18/94 69 500 <0.05 Wed. 1/19/94 50 44.7 n.s.* Thurs. 1/20/94 54 423 n.s.* F ri. 1/21/94 39 43.3 n.s.* Sat. 1/22/94 50 420 <0.05 Sun. 1/23/94 39 36.0 n.s.* M on. 1/24/94 52 50.0 n.s.* Hospital ’ D’ Tues. 1/18/94 32 21.7 <0.05 Wed. 1/19/94 33 18.3 <0.05 Thurs. 1/20/94 25 18.0 n.s.’ F ri. 1/21/94 28 25.0 n.s." Sat. 1/22/94 26 23.8 <0 05 Sun 1/23/94 32 233 n.s.* M on. 1/24/94 19 177 n.s.* A 1-sample t-test: t=(*-g) / (s/( /n)) * Not statistically significant, p > 0.05 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-3. Mechanisms of Injuries for 3 Emergency Departments: 1/17/94 Versus Average from Other Mondays. January 1994. ui v O Mechanism Hospital Frequency Comparison Group Average Comparison Group Std Dev t-test* d.f. p < 0.05** Motor Vehicle Collisions ■ B * 10 2.50 0.58 -25.98 3 increase ’C’ 1 6.75 3.50 3.29 3 decrease D’ 0 1.00 0.00 undefined ... ... Firearm and violent injuries 'B * 0 0.00 ... ... ... . . . 'C * 0 1.25 0.50 5.00 3 decrease 'D ' 2 1.25 1.50 -1.00 3 n.s. Falls 'B' 33 3.25 0.50 -119.00 3 increase •c 15 12.50 2.38 -2.10 3 n.s. 'D ' 30 5.75 2.22 -21.87 3 increase Poisonings •B * 6 1.00 0.82 -12.25 3 increase 'C 1 2.50 2.65 1.13 3 decrease 'D ' 2 3.00 1.15 1.73 3 decrease Cut by or pierced by •B’ 60 1.50 0.58 -202.65 3 increase ’C 26 8.00 3.37 -10.69 3 increase D’ 20 2.75 1.48 -23.26 3 increase Struck by or against 'B' 54 1.75 1.26 -83.05 3 increase 'C' 23 5.25 2.63 -13.50 3 increase ’D’ 18 2.50 2.38 -13.02 3 increase Caught in or between object(s) *B ’ 0 0.25 0.50 1.00 3 decrease ■c 0 1.00 0.82 2.45 3 decrease ’D' 3 0.50 0.58 -8.66 3 increase Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-3. (Continued) Mechanism Hospital Frequency Comparison Group Average Comparison Group Std Dev t-test* d.f. p < 0.05** Fire or burn ■ B ’ 1 0.00 0.00 ’C 0 1.50 0.58 5.20 3 decrease 'D ' 1 0.00 . . . . . . . . . . . . Overexert ion 'B * 1 0.75 0.96 -0.52 3 n.s. 'C' 1 1.75 1.26 1.19 3 decrease 'D' 3 0.25 0.50 -11.00 3 increase Other mechanisms ’B‘ 3 0.75 0.50 -9.00 3 increase ’O ' 2 7.00 1.26 7.95 3 decrease 'O’ 8 1.00 0.82 -17.15 3 increase Unknown mechanisms B ' 39 1.00 1.15 -65.82 3 increase ’C’ 7 3.00 1.63 -4.90 3 increase 'D ' 9 . . . . . . . . . . . . . . . * t=(x-„)/(s/(v/n)) * * 3 degrees of freedom, 1-tailed critical value: -2.353 U J 00 o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-4. Mechanisms of Injuries (or 3 Emergency Departments: 1/18/94 Versus Average from Other Tuesdays. January 1994. Mechanism Hospital Frequency Comparison Group Average Comparison Group Std Dev t-test* p < 0.05** Motor Vehicle Collisions 'B ' 3 5.33 3.06 1.32 decrease ’C’ 2 5.67 3.06 2.08 decrease 'D' 3 3.67 2.08 0.55 decrease Firearm and violent injuries B * 1 0.33 0.58 -2.00 n.s. ’C’ 0 2.33 1.15 3.50 decrease ’D' 0 1.33 1.53 1.51 decrease Falls ’B ' 6 3.33 2.08 -2.22 n.s. ’C’ 14 10.67 0.58 -10.00 increase D' 15 3.67 1.53 -12.85 increase Poisonings B’ 1 2.67 0.58 5.00 decrease 'C 3 2.33 1.15 -1.00 n.s. 'D ' 2 1.33 1.53 -0.76 n.s. Cut by or pierced by 'B' 8 2.67 1.15 -8.00 increase ■ C ’ 16 6.33 1.53 -10.96 increase D’ 4 3.67 2.89 -0.20 n.s. Struck by or against ’B ' 13 1.00 0.00 increase ’C ' 15 8.33 1.53 -7.56 increase ’D ' 3 2.67 2.08 -0.28 n.s. Caught in or between object(s) ’B ' 0 0.00 . . . . . . . . ’C 0 1.67 2.08 1.39 decrease 'D’ 1 0.33 0.58 -2.00 n.s. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-4. (Continued) Mechanism Hospital Frequency Comparison Group Average Comparison Group Std Dev t-test* p < 0.05** Fire or burn ’B ' 1 0.00 'C ' 2 1.33 1.53 -0.76 n.s. ’D’ 0 0.67 1.15 1.00 decrease Overexertion * B ' 0 0.33 0.58 1.00 decrease ’C' 1 1.33 0.58 1.00 decrease ’D’ 0 1.33 1.53 1.51 decrease Other mechanisms 'B * 4 0.67 0.58 -10.00 increase ’C’ 3 8.00 4.36 1.99 decrease ’D' 3 1.67 0.58 -4.00 increase Unknown mechanisms ’B‘ 1 0.67 0.58 -1.00 n.s. ’C’ 13 2.00 1.00 -19.05 increase ’D' 1 1.33 1.15 0.50 decrease * t= (x-„)W n )) * * 2 degrees of freedom, 1-tailed critical value: -2.92 ut 00 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-5. Mechanisms o f Injuries for 3 Emergency Departments: 1/19/94 Versus A veraae from Other Wednesdavs. January 1994. Mechanism Hospital 1/19/94 Frequency Comparison Group Average Comparison Group Std Dev t-test* p < 0.05** Motor Vehicle Collisions 'A ' 6 1.67 1.15 -6.50 increase 'D ' 4 2.67 1.53 -1.51 n.s. Firearm and violent injuries ’A’ 0 0.00 0.00 . . . . . . . . 'O ' 0 0.33 0.58 1.00 n.s. Falls 'A ' 7 2.67 2.08 -3.61 increase ’D’ 9 4.33 2.08 -3.88 increase Poisonings ’A ' 1 1.67 0.58 2.00 n.s. 'O’ 2 0.67 1.15 -2.00 n.s. Cut by or pierced by 'A ' 4 2.00 1.00 -3.46 increase D * 6 3.33 2.52 -1.84 n.s. Struck by or against 'A ' 7 1.67 2.08 -4.44 increase D’ 4 1.33 0.58 -8.00 increase Caught in or between object(s) 'A ' 0 0.00 0.00 . . . . ___ D’ 0 0.33 0.58 1.00 n.s. Fire or burn 'A ' 1 0.00 0.00 . . . . . . . . ’D' 0 0.00 0.00 . . . . . . . . Overexertion 'A ' 1 0.33 0.58 -2,00 n.s. D’ 1 1.00 1.00 0.00 n.s. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-5. (ContinuedI Mechanism Hospital 1/19/94 Frequency Comparison Group Average Comparison Group Std Dev t-test* p < 0.05** Other mechanisms 'A ' 1 1.67 1.15 1.00 n.s. D * 4 3.67 1.53 -0.38 n.s. Unknown mechanisms ’ A ’ 1 0.67 0.58 -1.00 n.s. D’ 3 0.67 0.58 -7.00 increase * t=(*-„)/(s/e/n)) * * 2 degrees of freedom, 1 -tailed critical value : -2.92 o o Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-6. Mechanisms o f Injuries for 3 Emergency Departments: 1/21/94 Versus A verage from Other Fridays. January 1994. Mechanism Hospital 1/21/94 Frequency Comparison Group Average Comparison Group Std Dev t-test* p < 0.05** Motor Vehicle Collisions 'A ' 1 5.67 3.51 2.30 n.s. Firearm and violent injuries 'A ' 0 0.67 0.58 2.00 n.s. Falls 'A ' 6 5.00 1.73 -1.00 n.s. Poisonings 'A ' 1 1.67 0.58 2.00 n.s. Cut by or pierced by ’A’ 1 2.00 1.00 1.73 n.s. Struck by or against 'A ' 1 3.00 1.73 2.00 n.s. Caught in or between object(s) ’A ' 0 1.00 0,00 . . . . . . . . Fire or burn 'A ' 0 0.33 0.58 1.00 n.s. Overexertion 'A’ 0 1.00 1.73 1.00 n.s. Other mechanisms ’A ' 2 1.00 1.73 -1.00 n.s. Unknown mechanisms 'A ' 0 0.33 0.58 1.00 n.s. * t= (* -,0 W n )) *• 2 degrees o( freedom, 1-tailed critical value: -2.92 U ) 00 c / > Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-7. Mechanisms of Injuries for 3 Emergency Departments: 1/22/94 Versus A veraae from Other Saturdays. January 1994. Comparison Comparison 1/22/94 Group Group Std Mechanism Hospital Frequency Average Dev t-test* p < 0.05** Motor Vehicle Collisions B’ 1 4.25 0.96 5.88 increase C’ 2 5.25 2.63 2.14 n.s. 'D ' 2 4.25 2.08 1.87 n.s, Firearm and violent injuries ’B ' 0 0.50 0.58 1.50 n.s. ’C 5 0.50 0.58 -13.50 increase ’D' 1 1.00 1.53 0.00 n.s. Falls ’B ' 4 8.75 2.75 2.99 decrease ’C’ 18 12.00 1.41 -7.35 increase D’ 5 7.00 1.53 2.27 n.s. Poisonings B’ 0 1.75 1.71 1.77 n.s. ’C 2 2.50 1.29 0.67 n.s. D’ 1 2.75 1.53 1.98 n.s. Cut by or pierced by •B’ 4 2.50 1.73 -1.50 n.s. ’C’ 3 7.75 3.50 2.35 n.s. ’D ' 6 1.75 2.89 -2.55 increase Struck by or against 'B' 3 3.25 2.50 0.17 n.s. 'C 12 6.50 2.65 -3.60 increase ’D ' 7 2.75 2.08 -3.54 increase Caught in or between object(s) B’ 1 0.50 1.00 -0.87 n.s. ’C’ 1 1.75 0.50 2.60 decrease D’ 0 0.50 0.58 1.50 n.s. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table A3-7. (Continued) Comparison Comparison 1/22/94 Group Group Std Mechanism Hospital Frequency Average Dev t-test* p < 0.05** Fire or burn 'B' 1 1.00 1.15 0.00 n.s. ’C’ 1 0.50 0.58 -1.50 n.s. D’ 0 0.25 1.15 0.38 n.s. Overexertion ■ B ' 0 0.50 0.58 1.50 n.s. 'C ' 1 0.75 0.50 -0.87 n.s. 'D’ 0 0.25 1.53 0.28 n.s. Other mechanisms ’B ' 1 1.75 0.50 2.60 decrease ■ C ’ 5 1.75 1.26 -4.47 increase D’ 3 2.75 0.58 -0.75 n.s. Unknown mechanisms 'B' 1 1.50 1.73 0.50 n.s. ’C’ 0 2.75 1.71 2.79 decrease D’ 1 0.50 1.15 -0.75 n.s. * t=(x-/<)/(s/(/n)) * * 3 degrees of freedom, 1 -tailed critical value: -2.353 u » o o >4 APPENDIX 4: Abbreviated Tniury Score And Ininrv Severity Score Guidelines Reprinted with permission from the Abbreviated Injury Scale, 1990 Revision, Association for the Advancement of Automotive Medicine, Des Plaines, IL 388 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. THE ABBREVIATED INJURY SCALE 1 9 9 0 R ev isio n Association for the Advancement of Automotive M edicine 2340 Des Plaines River Road Suite 106 Des Plaines, IL 60018 USA 389 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS Currant M em bers of th e C om m ittee on inluvy Scaling: Thom as A. Gennarem. M.O.. Professor or Neurosurgery. University or Pennsylvania tChairmani Bame P-truceiii. Executive Onectcr. Association ror the Advancement of Automotive Meoione Susan P. Baker. Professor or Keaitn Policy ana Management. Johns Hookins University Howaro R. Chamoion. M.O.. Chief or Trauma Services. W ashington Hosonei Center Steonen A. Deane. M.O.. Senior Lecturer in Surgery. Syonev university. Australia Haroio A. Fenner. M.O.. Ortnooeoic Surgeon. HoOOS. NM □onaid F. Hueike. Ph.O.. Prolessor ot Anatomy a no Ceil Biology. University ot Michigan Eiien j . MacKenzie. Ph.D.. A ssociate Professor or Heaitn Policy ana Management. Jonns Hoekms University Gem M. McGinnis. M.S.N.. Associate Director. Heao Iryury Center. University ot Pennsylvania John A. M om s. jr .. M.O.. A ssistant Professor ot Surgery, vanoerokt Urwversn y John E. Pfess. M.O.. Professor ot Pathology. Indiana U raver si ty Jo nn D. S tates. M.O.. Professor Emeritus of O rthooadics. Univorsny of K s d e i t i r Jo seen j . Teoas. ill. M.O.. A ssociate Professor ot Pediatnc Sixgery. Unworau y of Florida Donaid D. Truneey. M.O.. Professor or Surgery. Oregon Health Sciences University Davio w . Y ates. M.O.. Senior Lecturer. Accident & Emergency Meoione. M anchester UK Former Membera: Robert W. Bryant. Accident investigation. General M otors Research Laooratories. Warren. Ml Robert N. Green. M.O.. LL3. Ccroner. Province ot Ontario. Canaaa Lee N. H am es. American Medicai Association tretireoi. Chicago. IL Michael Henderson. M.O.. Australian Doctors' Fund. Sydney. Austraka a .c . Hermg, M.O.. Secretary. Committee on Tratena. American College of Surgeons (retired!. Chicago. IL Joseon C. Marsh. Research Engineer. Ford Motor Co.. Dearoom. Ml Kermit Morgan. Staff Engineer. American M otors. Detroit. Ml W.D. Nelson. General Motors Techracal Canter. W arren. Ml j . Thom as Noga. NASS Program. National Highway Traffic Safety Administration. U.S. Oeot. of Tranaportadon W .J. Ruby. Staff Engineer. Ford Motor Co.. Dearborn, m i G. A nthony Ryan. M.D.. Road Accident Research Unit. University of f itslairta David C . Viano. Ph.D.. Biomedical Science D epartm ent. General Motors Research Laboratories A m erican Colloqo of Surgeons' Advisors: William F. Blaisdad. M.O.. Professor of Surgery. University ot California. Oaths Charles F. Frey, M.O.. Professor ot Surgery. University ot California. San Francisco Frank R. Lewis. M.O.. Professor ot Surgery. University ot California. San Francisco A m erican A ssociation lor the Surgery of Trauma Advtsac: Eugene Moore. M.O.. Pratessor ot Surgery. University ot Colorado. Denver j P ed iatric Surgical P anel Advisor*: I J. Ale* Haller. M.O.. Professor of Pediautc Surgery. Johns Hookins Urvversrty Burton H. H am s. M.O.. Professor of Pediatric Surgery. New England u * * —1 Center. Boston Michael E. M atiack. M.O.. Professor of Pediatnc Surgery. University of Utah Max L. Ramenotstcy. M.O.. Professor of Pediatnc Surgery. University of PlRSOtagn Davio E. W esson. M.O.. A ssociate Professor ot Pediatnc Surgery, Toronto Sick Childrens Hossitai B o rn e L. Beaver. M.D.. A ssistant Professor ot Surgery. University of Maryland Hospital. Baltimore Ad H oc Advisor*: Leonard M. Parver. M.O.. Associate Professor of Oohthaimoiogy, Georgetown University j Richard C. Schultz. M.O.. Professor ot Plastic and Reconstructive Surgery. University of Illinois | Layout a n d Format: Irene Herzau “Copyright 1990. Association for the AdeaacooMm of 1 nineaaiiii ffidirinr Das Plsiiue, X L M i l , USA 390 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. RULES TO REMEMBER HEAD • Skull fracture ♦ brain injury ♦ LOC * code fracture and injury only e LOC is coded only when no anatomical information easts, or in the rare instance when the LOC yields a higher seventy than does the injury • Self-reported LOC with no evidence of head injury and no EMS/medical personnel corroboration is disregarded Neurological Deficits are not present pre-tnjury, and last more than a transient period (is., minutes): e hemiparesis. hemiplegia, weakness, ssnsory loss, hyposthasia e aphasia, dysphasia, facial weaknesa/palsy (antral) e visual field defects, deviation of both eyes to one side, unequal pupils, fbtad or non-reactive pupils (not due to eye iniunes) Clinical Signs of Basilar Skull Fracture (any of the followM ng are sufficient evidence to code this iniury) • hemotympanum perforated tympanic membrane with blood in the canal e mastoid hematoma (battle signs), ponorbital acchymosts (raccoons ayss) • CSF otorrhea/rhinorrhea U J v O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. CHEST • Lung laceration * Rib fracture ♦ Hemo-pneumothorax ■ code the hemo-pneumo-thorax with the lung laceration and nb fracture separately (wtifhourtta* hemo-pneumo) • Rib fracture * Hemo-pneumothorax ■ code the hemo-pneumo with the hb fracture • Hemo- and/or pneumothorax only ■ find code under Thorecic Cavity, qq)X wb0n no more apedllc anatomical information exists ABDOMEN and PELVIC CONTENTS • Retroperitoneal hemorrhage is coded only if it can be determined to be unrefefled to thoracic or abdominal iniuries already coded. Injuries re/afed to retroperitoneal hemorrhage include: • injuries to the pancreas, duodenum, kidney. aorta, vena cava, mesenteric vessel e pelvic or vertebral fractures e Lumbar spine injuries are found in Spine e Rib cage injuries sre found in Thorax e Pelvis is found in Lower Extremities SKIN If the astensked injury (or in(unes): e accompanies injury to a deeper structure ■ code in the ectual body region e is the only injury for the body region • find code under the body region, assign to External e are minor, occur in multiple body regions, and are the only injuries ■ find single code for all injuries under External, assign to External Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Estimating Blood Loss - A number of injuries to the skin, vessel lacerations, brain lesions, and internal organs are de scribed in terms of blood loss by volume. The following table should help in assail ing blood loss when information in the hospital chart is not specific, and in coding these injuries in children. U l o u > ( WEIGHT 20% BLOOD LOSS | POUNDS KG CC 220 100 1500 165 75 1125 110 50 750 55 25 376 22 10 ISO L_ 11 5 76 J D IAG RA M O F NINES Reonmea witn permission 01 A merican Burn Association ana American College 0 1 Surfeons. BACK 18 ) 18 394 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. INTRODUCTION The appropriate classification of injuries by type and severity is fundamental to the study of injury etiology. Scales for categorizing injuries are grouped into two types: scales which assess the patient's physiological status, w hich may change over the duration of the injury’s treatment period, and those which describe the injury in terms of its anatomical location, specific lesion and relative severity. Origin of the AIS The need for a standardized system for categorizing injury type and severity arose in the mid 1960s coincident with the first generation o f multidisciplinary motor vehicle crash investigation teams. These teams, typically consisting of specialists in engineering, medicine, anatomy/physiology, and crash investigation, were organized to collect epidemiological data on crashes to enable evaluation of vehicle design in relation to injury incidence and mechanisms. Under the joint sponsorship of the American Medical Association, the Association for the Advancement of Automotive Medicine (formerly the American Association for Automotive Medicine) and the Society of Automotive Engineers, a committee representing these specialties, with the assistance of about 3S consultants, produced the first Abbreviated Injury Scale published in 1971.' The original AIS incorporated the pioneering work on scaling o f DeHaven2 at Cornell University, as well as scales from individual researchers predominantly in the United States and Europe. The AIS, although elementary in scope, became the standard for crash investigation teams funded by the U.S. Department of Transportation as well as university-based and industry-affiliated teams in the USA, Europe and Australia. W V O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. In 1973, the A A AM assumed the lead role in injury scaling and, through its Committee on Injury Scaling, became the parent organization ot the AIS. Details of the 1974 and 1975 modifications to the scale are discussed elsewhere.14 In 1976 the first AIS dictionary was published listing more than 500 injury descriptions.1 Several years later, the 1980 revision incorporated major improvements in coding brain injuries, as well as changes in injury scaling practices related to injury v. outcome, bums, skin lesions, and overall seventy assessment. These clarifications are discussed in the introduction of The Abbreviated Injury Scale 1980 Revision.6 The AIS was originally developed for impact injury assessment. The evolution o f trauma care systems and trauma registries in the 1980s fostered a need for expanding the AIS to facilitate the coding of penetrating trauma. New injury descriptions written in acceptable clinical language were introduced in the 1985 revision. These additions mainly concerned the vascular system and integumentary (skin) injuries. AIS 85 also expanded the range o f injuries and severity, particularly in the thorax and abdomen, to encourage more definitive injury coding than was previously possible.7 Thisedition, AIS 90, represents another major update of the system. Specific improvements are discussec below. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. The Purposes and P h ilo so p h ie s of th a AIS The AIS was developed to provide researchers with a simple numerical method for ranking and comparing injuries by severity, m d to standardize the terminology used to describe injuries. Since 1971, the need for greater sopiusucation of these goals has driven several revisions of the AIS. Throughout these revisions, the scope of injuries has been broadened, not only to include an expanded list of injury descriptors, but ilso to include injuries other than those that occur in a vehicular environment. Increasing the sophistication o f the description of the injuries has allowed the AIS to be utilized in more data collection efforts than ever before. Whereas early versions of the AIS were principally suited for large scale vehicular data, the most current revisions are now also useful to medical researchers involved in clinical circumstances. Because o f its responsiveness to these needs, the AIS has become accepted worldwide and has facilitated comparative injury research. ■ '- j Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Despite the changes that have occurred in the revisions, the AIS attem pts to remain true to the basic principles that were involved in its genesis. These principles have dictated the utility, as well as the limitations, for which the AIS has been useful. First, the AIS is based on anatomical injury and in this way differs from other systems that depend on physiological parameters. The consequence of this principle is that there is only a single AIS score for each injury for any one person, whereas in scales that depend on physiological measures, many scores are possible for a single patient depending on how the person's physiology changes over time. Second, the AIS scores injuries and not the consequences of the injuries. This principle has been employed so that the AIS can be used as a measure o f the severity of the injury itself and not as a measure of impairments or disabilities that result from the injury. A t the AIS has progressed, immediate consequences o f several injuries have been included as part o f certain injury descriptors in order to specify injury severity more precisely. Examples involve th rb nin (toss o f consciousness), blood vessels or solid organs (amount o f hemorrhage), and chest (pneumothorax). Third, the AIS is not simply a ranking o f expected mortality from injury. Were this the case, there would be no way to distinguish the majority o f minor and moderate injuries since they pose little or no threat to life . Although empirical data show that the AIS correlates well with the probability of death at the serious and life-threatening levels (AIS j> 3), other factors are also considered in AIS severity. These include potential for mortality, as well as the diagnostic certainty, rapidity, duration, complexity and expected effectiveness o f resolution with or without existing therapy. These factors are difficult to quantify but must be considered since severity is continually redefined by the progress of medicine. It is anticipated that the AIS will continue to be refined according to these basic principles o f its structure. u > 'O 00 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. A ssessm ent of Multiple Injuries The Abbreviated Injury Scale (AIS) is a consensus derived, anatomically based system that classifies individual injuries by body region on a 6-point ordinal severity scale ranging from AIS 1 (minor) to AIS 6 (currently untreatable). The AIS does not assess the combined effects of multiply-injured paticnto. The Maximum AIS (MA1S), which is the highest single AIS code in a patient with multiple injuries, h a s been used by investigators to describe overall severity. Us usefulness remains important in me sir vehicle injury research concerned with vehicle design changes. In trauma research, however, the MAI b > was found lacking due to its nonlinear relationship with the probability of death. Also, death rates vor y significantly within each AIS value for the most severe injury depending upon the AIS value for the second most severe injury. Baker s Injury Severity Score t ISS) published in 1974 gives a much better fit between overall severity and probability of survival.'9 The ISS is the sum of the squares o f the highest AIS score in three different body regions. [See page 10 for instructions on how to calculate the ISS.] ICP-AIS Compatibility In 1986 MacKenzie et al published their ICD-AIS conversion table, its developmental work and potential uses.1 0 The conversion table translates ICD-9CM coded discharge diagnoses into AIS body regions and severity codes. With AIS 85 it was possible, applying a list of assumptions, to match satisfactorily a number of AIS injury diagnoses and ICD rubrics. The extent to which this will be possible with AIS 90 is yet to be determined. U ) g Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. AIS 9 0 IM PROVEM ENTS_________________________________________________________ The im provem ents in AIS 90 result from almost two decades of clinical and research applications o f the system. T h ese im provem ents are discussed briefly. Coding Guidelines Injury data collection can be hampered because o f problems with the scales, the informatian available to the coder, or the coders themselves. The standardization o f injury terminology and the expansion o f the AIS from its original 75 injury descriptions in 1971 to over 2,00 0 to accommodate both blunt and penetrating trauma has dim inished the inadequacies o f earlier scales. Further, the AAAM offer training seminars to those responsible for extraction and interpretation of injury information. The problems with inadequate injury information are more difficult to solve than those of scales and coders. For example, an autopsy report of "multiple blunt trauma resulting in death" does not provide any specific information on the injuries and is virtually useless for injury coding purposes. Though not as grossly inadequate, even hospital chans can be deficient in detail or can give contradictory in formation from one piece of the hospital record to another. AIS 90 includes specific rules within the dictionary itself to solve some coding dilemmas such as when there is a choice o f descriptions or body regions to which an injury can be assigned, or when clinic)! diagnoses can be used. Synonym s and parenthetical descriptions are used extensively to allow thecode to appropriately match the injury description in the hospital chart with one in the AIS dictionary. These coding rules together with coder training should improve intra- and interrater reliability. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Penetrating Injuries AIS 85 introduced some descriptions that allowed coding of penetrating injuries, such as gunshot and stab wounds. In addition, clinical terminology that routinely is used to describe penetrating injuries to the vascular system, thoracic and abdominal organs was included. Coding experience since 1985 using these descriptors suggested improvements, especially in the terminology. In AIS 90, penetrating injury descriptions are compatible, to the extent possible, across all body regions. The AIS codes for the vascular injuries reflect empirical clinical research findings of the last several years, notably from the Major Trauma Outcome Study (MTOS)." P g d ia ttiiL la iu iifla Age can be an important variable in relation to injury severity. It is well documented that an older patient will have a higher probability of unfavorable outcome than a healthy younger person given thesame injury severity. Very young children may similarly be worse off. Several years ago, Baker convened a group of pediatric trauma surgeons to review all of the injury descriptions in AIS 85 and their AIS severity codes to determine which did not apply to a pediatric population. Of the more than 2,000 diagnoses, these trauma specialists agreed that all but about 1 adequately reflected relative severity of injuries in young children. These exceptions related to the su e of brain hematomas, blood loss m severe lacerations, or internal bleeding, by volume, due to abdominal or thoracic injuries. The Committee on Injury Scaling concurred in these proposed changes and in few cases felt that the changes applied to all age groups. These revisions are incorporated into AIS 90. ■ U o Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Expanded injury List The AIS does not measure impairment or disability. A scale that would compteHMWhfcrAIS andl provide a link between injury severity and societal costs is fundamental. Several salfes hive been suggested or are underway.1,14 A framework for constructing an impairment scale has recently been proposed by States,,s and work on its development has been undertaken by the AAAM Committee on Injury Scaling. The criteria for such an impairment scale are being deliberated. In anticipation of this new scale, the list of injuries fcr the AIS has been expanded to accommodate the addition o f an impairment severity code. Even when the AIS code is the same for a number of different injuries to an organ, the relative impairment o f those injuries may be quite different; thus the need for more definitive injury diagnoses. Numerical Iniurv Identifier AIS 85 introduced a unique 6-digit code for each injury diagnosis to assist in computerization o f data. The addition o f injury descriptions in AIS 90, especially in the brain and extremities, has required a more flexible numerical system than that used in 1985. In AIS 90, each injury description is assigned a unique 6-digit numerical code in addition to the injury severity score. As summarized in the diagram below, the first digit identifies the body region; the second digit identifies the type of anatomic structure; the third and fourth digits identify the specific anatomic structure or, in the case of injuries to the external region, the specific nature o f the injury; the fifth and Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. sixth d ig its id en tify th e level o f in ju ry w ith in a sp ecific b o d y reg io n an d a n ato m ic stru c tu re . T h e d ig it ;o the rig h t o f the d e c im a l p o in t is the A IS sco re: a> a > 3 3 < - * O O 3 3 W la — * * * * o W ) < /) 5 ) o o « a a '5 up oc O g e g o o M > 1 ) * * O ■ a o 2 > w ◦ > c ac n i" o q t - < </>< _j < U U UU LLI . U •fc. o U ) Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. T he follow ing convenuons are used in assigning the n um encs to specific injury descriptions: 1. Body Region 3 . Specific Anatomic Structure or Nature 1 Head 2 Face 02 Skin - Abrasion 3 f.'ecK 04 Contusion 4 Thorax 06 - Laceration 5 Abdomen 08 • Avulsion 1 6 Spine 10 Amputation 7 Upper extremity 20 Burn 30 rfush 8 Lower Extremity w u w i u a i i 40 Deglovmg 9 Unspecified 50 Injury - NFS 6 0 Penetrating 90 Trauma, other than mechanical Hudi L Q fi 02 Length of LOC 2 . 04, 0 6 ,0 8 Level of Consciousness Type of Anatomic Structure 10 Concussion l Whole Area 2 Vessels Soine 3 Nerves 02 Cervical 4 Organs unci, muscles/lie.) 04 Thoracic 5 Skeletal (mcl. joints) 06 Lum bar 6 Head • LOC y s a t i s . .Merces. Qraani.. Bants. Joints are assig n ed co n secu tiv e tw o digit num bers beginning w ith 0 2 . -u S Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 4 . Level -U o i/i Specific injuries are assigned consecutive two-digit numbers beginning with 02. To the extent possible, within the organizational framework of the AIS, 00 is assigned to an injury NFS as to severity or where only one injury is given in the dictionary for that anatomic structure. 99 is assigned to an injury NFS as to lesion or severity. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. External Injuries Pnor to AIS 90, external iniunes to the skin were coded under the External section no matter where they occurred on the body. This practice presented problems to researchers in evaluation o f certain vehicle design changes or public policy measures; for example, the effectiveness of mandatory seat belt use laws in reducing facial injuries. External facial injuries of AIS <_ 2 could not be located in the entire pool of injuries across body regions in the External section. hi AIS 90 external injuries have oeen dispersed across body regions so that they can be easily located The unique numerical code shouid facilitate easy access to specific injuries. One complication dl changing this practice is the potential effect upon the Injury Severity Score which uses External as one cC its six body regions. Specific rules have been written on how to calculate the ISS when external injurie> are involved so as not to compromise the overall severity assessment [see page 11]. Brain Injuries Analysis of large data bases sucn as the MTOS supported the Injury Scaling Committee's conclusion that senous brain injuries (AIS >_ 3) were undercoded when compared to injuries in other body regions. To correct this inconsistency, .he Brain section in AIS 90 has been substantially expanded to include brain contusions with a range from AIS 3 to AIS 3 that account for size, location and multiplicity of the lesions. The volume, size and location descriptors for cerebral and cerebellar hematomas have al| been revised to more adequately reflect the relative severity of these injuries in the real world. Die terminology to describe these injuries is clinically more acceptable. Two major additions are section on intracranial vessels and cranial nerves. AIS 90 allows the coding of basilar skull fracture based upon clinical signs in the absence of document, anatomical information. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Terminology Modifications Vascular injuries have been clarified, both in terms of language and severity, in the Neck, Thorax and Abdomen sections. Changes in abdominal organ injury descriptors make the AIS more compatible vriri current injury classifications used in clinical settings and in trauma research. These improvements will be especially useful in coding penetrating injuries. Alterations have also been made in the descriptors for penetrating injuries that do not involve deep anatomical structures, and for external lacerations. CONCLUSION_________________________________________ The AIS was originally developed to be used by crash investigators to standardize data on the frequency and severity of motor vehicle related injuries. Its use has been extended to epidemi ological research, trauma center studies to predict survival probability, patient outcome evaluation and health care systems research. It also features in studies to assess societal costs o f injuries. The AIS is the foundation for the Injury Severity Score, and will likely remain the basic system for future methods to assess overall injury severity. The AIS has been universally accepted. Accu rate and consistent application of the AIS, therefore, is fundamental to sound injury data collection globally. 4 k © •si Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. R e f e r e n c e s 1. Rating the severity of tissue dam age: I. The Abbreviated Injury Scale, JAMA 215:277-280, 1971. I. D eH aven H, The site, frequency and dangerousness of injury sustained by 800 survivors of light plane accidents. Crash Injury Research, Department of Public Health and Preventive Medicine, Cornell University M edical College, New York, July 1952. 3. States JD, Huelke DF, HamesLN, 1974 AMA-SAE-AAAM revision of the Abbreviated Injury Scale (AIS), AAAM Proceedings 18:479-505. 4. Joint Committee on Injury Scaling of SAE-AAAM-AMA, The Abbreviated Injury Scale (197S revision), AAAM Proceedings 19:438-466. 5. The Abbreviated Injury Scale (AIS) 1976 revision, including Dictionary, American Association for Automotive Medicine (now Association for the Advancement of Automotive Medicine), Des Plaines, 1L. 6. The Abbreviated Injury Scale 1980 Revision, American Association for Automotive Medicine (now Association for the Advancement of Automotive Medicine), Des Plaines, IL. 7. The Abbreviated Injury Scale 1985 Revision, American Association for Automotive Medicine (now Association for the Advancement of Automotive Medicine), Des Plaines, IL. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 8. Baker SP, O'Neill B, H addon W , Long WB, The Injury Severity Score: A method for describ- ing patients with multiple injuries and evaluating emergency care. J Trauma 14:187-196, 1974. 9. Bull JP, The Injury Severity Score of road traffic casualties in relation to mortality, time of death hospital treatment time and disability, Accid Anal & Prev 7:249-255, 1975. 10. NlacKenzie EJ, Steinwachs DM, Shankar BS, Classifying severity o f trauma based on hospital discharge diagnoses: Validation of an ICD-9CM to AIS-85 conversion table, Medical Care 27:412- 422, 1989. 11. Champion HR, Copes WS, Sacco WJ, Major Trauma Outcome Study: Establishing national norms for trauma care. (Accepted for publication in the Journal of Trauma.) 12. Hirsch AE, Eppinger RH, Impairment scaling from the Abbreviated Injury Scale, AAAM Proceedings 28:209-224, 1984. 13. Gustaisson H, Nygren A, Ting vail C, Permanent medical impairment among road traffic victims and risk of serious consequences, RSC, SAE International Congress and Exposition, 1986. 14. The Development of an Impairment Index for Traumatic Injuries, in progress at Johns Hopkins University, Baltimore, MD. 15. States JD and Viano DC, Injury impatrmentand disability scales to assess the permanent consequences of trauma, Acc Anal & Prev 22:151-160, April 1990. 16. Gennarelli TA, Champion HR, Sacco WJ, Copes WS and Alves WM, Mortality of patients with head injury and extracranial injury treated in trauma centers, JTiauma 29:1193-1202, September 1989. •u o V O Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. U SIN G TH E D IC TIO N A R Y Form at The AIS dictionary is divided, tor convenience, into nine sections in the following order: Head (Cranium and Brain); Face; Neck; Thorax; Abdomen and Pelvic Contents; Spine; Upper Extremity; Lower Extremity; External and Other. These sections are different from the six body regions used to calculate the ISS described below. These differences should be carefully noted to avoid errors in assigning injuries to the appropriate body region for calculating the ISS. Within each section, except the Sflins, and External. Bums. Other Trauma, injury descriptions are alphabetized by specific anatomical part and are categorized in the following order: Whole Area, Vessels, Nerves, Internal Organs, . Skeletal. In addition. Upper and Lower Extremities have a section on Muscles. Tendons. Ligaments Inmostcases, the severity level in each anatomical category goes from least severe to most severe. The Anatomical Index which follows the Dictionary lists all of the injury descriptions in AIS 90 in alphabetical order, the body region in which the injury is located, and the page on which it can be found. Each injury description has been assigned a unique 7-digit numerical code (see pages 4-5). The single digit to the right of the decimal point is the AIS number, according to the following severity code: AIS Code Descrmtion 1 Minor 2 Moderate 3 Serious 4 Severe 5 Critical 6 Maximum 9* Unknown at © o f th e copyright owner. Further reproduction prohibited without permission. 7 3 CD " O O C l C o CD Q . ■ o CD C / ) c n o' o Signs and symbols are used throughout the dictionary to help the coder. Examples o f each follow: Brackets [ ] give specific instructions or direction. Example. Afveoktr ridge (bonej fracture with or without injury to teath [do not code teeth separately where these occur simultaneously! Parentheses ( ) give synonym s or further descriptive information. Example. Pancreae laceration maasive (avulsion; complex; rupture; stellate; tissue lossl Boxed Information gives coding guidelines. Example. Lung contusion NFS This diagnosis should be coded only if there is history of chest trauma and a physician's diagnosis is documented by x-ray, CT, MRI, surgery or autopsy. Clinical pulmonary dysfunction is insufficient evidence of a codeablc injury. Diagonal means and/or. i.e., one or more o f the descriptors must be present. Example. Tibia fracture NFS open/displaced/comminutad •tee General Coding Rule 9 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. C alculating th e Iniurv S ev erity S co re (ISS) A. General rules The ISS is the sum of the squares of the highest AIS code in each of the three most severely injured ISS body regions. The six body regions of injuries used in the ISS are: 1. Head or neck Fece J. Chest 4. Abdominal or pelvic co n ten ts « « • Extremities or pelvic girdle 5. External ■ u Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. •u U ) Head cr neck injuries include injury to the brain or cervical spine, skull or cervical spine fractures. Facial injuries include those involving m outh, ears, eyes, nose and facial bones. Chest injuries and injuries to abdominal or pelvic co ntents include all lesions to internal organs in the respective cavities. Chest injuries also include those to th e diaphragm , rib cag e, and thoracic spine. Lumbar spine lesions are included in the abdominal or pelvic area. Injuries to the extremities or to the pelvic or shoulder girdle include sprains, fractures, dislocations, and am putations, except for the spinal column, skull and rib cage. External injuries include lacerations, contu sions, abrasions, and bum s, independent of their location on the body surface. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Note again that these ISS body regions do not necessarily coincide with the sections used in the AIS. For example, the AIS Spine section is divided into three ISS body regions: cervical in ISS Head or Neck, thoracic in ISS Chest, and lumbar in ISS Abdominal or Pelvic Contents. rhe following example should licip in understanding the ISS calculation. ' ISS BO DY REGION IN JU R Y AIS CODE H IG H E ST AIS A U ^ h e a d /n e c k : Cerebral contusion Internal carotid artery, complete transection 140602.3 320212.4 4 16 FACE: Ear laceration 210600.1 1 CHEST: Rib fractures left side, ribs 3-4 450420.2 2 ABDOMEN: Retroperitoneal hematoma 543800.3 3 9 EXTREMITIES: Fractured femur 851800.3 3 9 EXTERNAL: I i i Overall abrasions 910200.1 1 0 \ SS - 34 j ISS scores range from 1 to 75. A score of 75 results in one of two ways, either with three AIS 5 injuries or with at least one AIS 6 injury. Any injury coded AIS 6 is automatically assigned an ISS of IS. However, the coder is instructed to code all the injuries in that patient even though the ISS will not ba altered by additional injuries. It is not possible to calculate an ISS on a patient who has any code 9 injury; hence, the need to press for detailed injury information. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. B. Coding Injuries to the Skin In AIS 85. minor and moderate (AIS 1-2) injuries 10 the skin and penetrating injuries were coded under the External section and assigned to the External body region for calculating the ISS. These injuries have, now been dispersed across body regions to help in locating them and are marked with asterisks. (See External Injuries on page 6 for rationale.) This change in practice should not affect the overall severity assessment. The following rules should be applied in calculating the ISS scores involving external injuries: * If the asterisked injury is the only injury in a body region, locate it under the body region in which it occurs but assign it to ISS body region, External. * If the asterisked injury accompanies an injury to a deeper structure, code the asterisked injury under the body region in which it occurs. The injury to the deeper structure in that body region will take precedence over ntg. external injury for ISS. * If minor (AIS 1) external injuries occur in multiple body regions but are the only injuries, code as asinale injury under External (e.g., multiple contusions would be coded as Contusion, code 910400.1, under External body region). Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission General Coding Rules Instructions are included throughout the AIS dictionary to help coders to make appropriate decisions concerning specific injury diagnoses. These are not repeated here. A number or coding principles, however, apply across body regions. These rules which follow should be learned and applied diligently. Injuries described as "probable." "possible," "impression of," or "rule out" should not be coded unless they are substantiated in the medical record. Foreign bodies are not injunes and therefore are not coded. 3. The AIS does not assign codes to conse quences of injury (e.g., blindness) but rather to the injury per se (e.g., optic nerve avul sion). 4. Surgical procedures and other treatment in terventions should not be used to determine the severity of an injury. a Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. 5. AIS 6 is used only for injuries specifically as signed severity level 6 in the AIS. The use of AIS 6 is not an arbitrary choice simply because the patient died. 6. The "crush" injury description is used only when the injury meets the criteria in the dictionary. 7. Bilateral injuries are coded separately for organs such as the kidneys, eyes, ears, and extremities unless the dictionary specifically allows for coding as a single injury (e.g., lung injuries). Maxillae, mandibles, the pelvis and rib cage are coded as single struc tures. 8. An open fracture, by definition, means that the skin overlaying the fracture is lacerated. The external laceration is implicit in the code for open fracture and is not coded separately. 9. AIS 90 uses "not further specified” (NFS) to allow for coding injuries when detailed infor mation is lacking. Injury unspecified means that an injury has occurred to a specific organ or body part, but the precise injury type is not known. For example, a kidney injury could be a contusion or a laceration, but this information may not be available. In this example, the Iddne; injury is coded as NFS. 99 is assigned to an injury NFS as to lesion or severity. [See Numerical Injury Identifier, page 4.] Severity unspecified means that a specfic in jury (e.g., laceration) has occuned, but the level of severity is not specifically given or ii unclear. In this example, the injury should be coded as laceration NFS. To the ex ten possible within the organizational frameworl of the AIS, 00 is assigned to an injury NFS a. to severity. [See Numerical Injury Identifier page 4.] Use of NFS should not be confused with cod. 9 which is assigned in those cases wher trauma has occurred, but no information i available regarding specific organ or region For example, "blunt abdominal trauma” i assigned a code 9. 1 0 . If there is any question about the severity o f an injury based upon ail available docu mented information, code conservatively (i.e., the low est AIS code in that injury's cate- gorv;. 11 . Estimating Blood Loss — A number o f injuries to the skin, vessel lacerations, brain lesions, and internal organs are described in terms o f blood loss by volum e. T he follow ing table should help in assessing blood loss w hen in formation in the hospital chan is not specific, and in coding these injuries in children. ( WEIGHT B L O O D LO SS ] =OUNOS <G c c 2 2 0 100 • 5 0 0 165 75 1 1 2 5 110 50 7 5 0 55 25 3 7 5 22 10 1 5 0 V 11 5 in 418 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX 5: Memorandum from EOE International 419 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. EQ E IN TERNA TIO N A L, INC. 4590 MacArthur Blvd.. Suite. 400 Newport Beach. CA 92660-2027 phone 949/833-3303: fax 949/833-3392 e-mail: HAS@ eqe.com MEMORANDUM Date: October 30. 1998 To: Maya M ahue-Giangreco Injury and Violence Prevention program. Surveillance Section County of Los A ngeles. Departm ent of Health Services From: Hope Seligson Copies: Charlie Huyck. Ron Eguchi Subject: Northndge Earthquake Injury Data: Match to the Los Angeles County Assessors and EQ E/O ES Damage Databases (EQE Project Num ber 250658.01) Summary: EQE International is pleased to provide to the County of Los Angeles. Department of Health Services (DHS), the results of the address match between the casualty data collected by DHS, and the Los Angeies County Assessors Database and the EQ E/O ES Damage Database. DHS provided location information on 3.436 injury records to EQE International to be matched against the complete Los Angeles County Assessor's Tax Roll for the year 1993. The results of this match are being returned to DHS along with pertinent structure information including: year built, construction class, square footage, etc. Matching efforts, utilizing automated matching as well as "hand-matching" and manual review, resulted in 2.049 matches (60%). However, a reduced amount (19%) of the injury records is considered “earthquake-related”. The overall match rate for the earthquake-related injuries to Assessor's data is 71%. Table 1 presents a breakdown of match results by injury type. in addition, the injury data was matched to the EQE/OES Damage Database developed dunng the Northndge Earthquake response. Information pertaining to the Building and Safety Inspections following the Northndge Earthquake are being supplied for 377 records (11% of total database. 18% of data matched to Assessor's database) which had an Assessor's Parcel Number assigned as a result of the Assessor Database match, and were subsequently matched to the damage database. The match rate for earthquake-related injunes is somewhat higher than the overall database, reaching 17%. 420 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. MEMORANDUM Z J 10/99 =aae 2 of 7 TABLE 1: Match Resuits by Injury Type Number of Records Percent Match to Assessors Database Percent Match to Assessors Database. Geocoded (w/ Lat & Long) ! 1 - Cleany EQ-reiated 237 76% 76% 2 - Not EQ-related 2783 57% 56% 3 - Assumed EQ-relatea 405 68% 65% i 4 - Indirectly EQ-relatec 11 55% 55% i Total 3.436 : Results of the Los Angeles County Assessor's Tax Roll Address Match: EQE International receives 3.436 recoras collected by DHS. The database provided location information (i.e. street address) for mjunes resulting from the Northndge Earthquake. The goal of this project was to match the data received from DHS to the Los Angeles County Tax Assessor's Roll, and the EQE/OES Damage Database, in order to identify the structure type and conditions where the respondent was located, and where the injury was reported to have taken place. The DHS injury record data were attached to the Los Angeles County Assessor's Database, which contains approximately 2.25 million parcel records. The results of the address match allowed EQE to develop a table that contains all of the site-specific information from the County Assessor's Tax Rolls. Overall, nearly 60% of the total number of records supplied were matched to an address within the Los Angeles County Tax Assessor's Roll. Table 2 defines the fields provided from the Assessor's data, and the gives a brief descnption of the field contents. EQE is providing the data in an Excel file entitled DHS_ASSR XLS (Table: D H S A sse sso r D am age Match). Condominiums: While the majonty of matches resulted in the selection of unique parcel sites. 112 street address matches resulted in the selection of multiple parcels. These records are denoted with a "Match Flag ’ of 2 (condominiums) or 3 (split parcels). Most of these multiple matches (90) actually represent condominium buildings’, and aggregation of the key fields (i.e.. square footage and number of units) results in a reasonable representation of the actual building. For each condominium record, the "Square Footage Building" represents the aggregated area of all units within the building. "Square Footage Unit" represents the actual area of the unit in question (when unit number was provided with the injury data) or a representative unit area (when the unit number was not provided with the injury data). For non-condominium structures. "Square Footage Unit" and "Square Footage Building" are equal. Condominium buildings represented in the current database range from 4 unit buildings with approximately 3600 square feet of floor space, to buildings with more than 200 units and Note: The Los Angeles County Tax Assessor assigns each condominium a unique parcel number for tax purposes. For condominiums that have a common street address and unit numbers, multiple matches will ensue. For condominiums, which have unique street addresses, only the unit in question w ill appear. 421 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. MEMORANDUM 2/10/99 =age 3 cf 7 more than 200.000 square feet of floor space. The remaining multiple matches resulted from other issues, such a s split parcels. Please note that some condominium units are given a unique street address. In this case, the address match will not return multiple parcels: rather the results will be the return of the singular condominium unit Results of the EQE/OES Damage Database Parcel Number Match: Once the injury data had been matched to the Los Angeles County Tax Assessor's Roll, it was possible to refer to the Northndge Earthquake dam age database developed by EQE International and the Governor's Office of Emergency Services. This database represents the best comoiiation of Building and Safety Inspection records from many of the local (unsdictions directly impacted by the earthquake. The damage database contains nearly 115.000 distinct Building and Safety dam age inspection records. Each of these inspection records had previously been address matched to the Los Angeles County Tax Rolls, and the parcel number recorded. Of the 2.049 addresses matched to the County Tax Rolls. 18% (377 records) appeared in the dam age database. Pertinent dam age information for these 377 records is therefore being made available DHS. including: OES Damage ID. Building and Safety Placard (tag), estimated damage, and estimated percent damage. See Table 2 for field definition and descnption. Reference Maps: EQE International is pleased to provide two Geographic Information Systems (GIS) maps which present the DHS injury data in relation to the ground shaking from the Northridge Earthquake. To develop these maps. EQE utilized the latitude and longitude from the Tax Assessor's database, when available. In this manner. EQE was successful in obtaining a latitude and longitude for 59% (2.011 records) of the complete DHS injury database, and 69% of the earthquake-related injury records (449 records). The first map shows the official Modified Mercaili Intensities (MMI) for the Northndge Earthquake, developed by the USGS (Dewey. 1995), overlain with the DHS Injury database. Likewise, the DHS Injury data were placed over the contours of Log of PGA developed by OES-GIS using uncorrected data from the U.S. Geologic Survey (USGS) and the California Division of Mines and Geology (CDMG), as documented in the EQE/OES Northndge Earthquake Report of Data Collection and Analysis . Part A (EQE/OES. 1995). Three additional map views will be developed to provide a more detailed look at selected areas with high concentrations of injunes. These areas will be identified with consultation from DHS. References: Dewey. J.W.. B.G. Reagor, L. Dengler, and K. Moley (1995). “Intensity Distnbution and Isoseismal Maps for the Northridge. California. Earthquake of January 17, 1994." U.S- Geological Survey Ooen-File Report 95-92. Evemden. J.F. and J.M. Thomson. (1988), “Predictive Model for important Ground Motion Parameters Associated with Large and G reat Earthquakes." United States Geological Survey Bulletin 1838. 422 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. MEMORANDUM 2710/99 P a g e 4 of 7 EQE International. Inc. a n a th e G overnors Office of Em ergency Services (1995). The Northndoe Earthquake of Jan u ary 17 1994: Report of Data Collection and Analysis. Part A: Damage and Inventory D ata. Irvine. California Attachments: Los A ngeles County A sse sso rs Tax Roll Use C ode Definitions E vem aen Soil C ode Definitions Modified Mercalii Intensity Map for the Northridge E arthquake Contours of Log of Peak Ground Acceleration for the Northridge E arthquake 423 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. MEMORANDUM 2 ' 1 0 / 9 9 = a g e 5 o f 7 Table 2 DHS Address Match Oatabase • Field Definition and Description Field Name Description ENTNUM DHS's Unique identifier. ADD num Address Number provided by DHS. ADD DIR i Street Direction (compass Direction) provided by DHS. ADD NAME Street Name provided by DHS. ADD STYP Street Suffix provided by DHS. ADD APT ! Apartment or unit number (where applicable) provided by DHS. CITY City or community name orovided by DHS. LEGAL_CITY i The legal name for the City (or the name of the city in which the community resides) (i.e. Northridae is within the Citv of Los Angeles). LEGAL CNTY The legal name for the County. ZIP Zip Cooe orovided by DHS. FULL ADD Full Address provided by DHS. Earthquake j I Code indicating wnether the injury was earthquake-related, provided by DHS (i=earthguake-reiated. 2 = not earthquake-related. 5 = assumed earthauake-related. and 9=indirectly earthquake related) FaciCode Facility code mdicatinq hospital provided bv DHS. TIMEPD Code indicating time of injury provided bv DHS. ISS Code indicating iniury seventy provided bv DHS. Parcel Number ! Tax Assessor Roll Number for each situs. Situs Address Assessor's street address for the parcel. Situs Zip Code Parcel Zip Code. Use Code This code differentiates the major use of the parcel (Residential. Commercial) and the detailed use within the parcel (Single Family, Duplex. Service Station, etc.). Attachment I provides a complete listing of use definitions Structure Type This field identifies the generalized construction of the improvement. A=Stee! Frame B= Concrete Frame C= Block/Bnck/Other Concrete D= Wood Frame 424 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. MEMORANDUM -'1 0 /9 9 3 a g e o of 7 Table 2 (Continued) DHS Address Match Database - Field Definition and Description Field Name Description Year Built Year tne improvement was constructed. Numoer of Units Numoer of Residential Units, or office units in the improvement. Square Footage Unit i i i i i This field identifies the actual square footage for single family dwellings and other non-condominium Duildings. and TYPICAL square footage of a unit within the improvement for condominiums. Please refer to text in tnis memo that further descnbes the method of handling condominiums. Square Footage ■ Building This field contains the total square footage for the improvement. This number should equal the Unit Square Footage for Single Family Dweilinqs and other non-condomimum buildinqs. Latitude Latitude of parcel Lonqitude Lonqitude of parcel Evem den Soil Code i Evemoen Regional Soils applied to the County A ssessors Tax Rolls by EQEfOES. See Attachment II for soil code definitions (Evemden. 1988V MMI Value The USGS Modified Mercalli Intensity (MMI) as mapped for the parcel in question. This value represents the shaking intensity expenenced at the improvement. (Dewey. 1995)3 C ensus Tract The 1990 C ensus tract the parcel resides in. This is supplied to allow for the application of Census data. C ensus Block Group The 1990 C ensus Block Group for the parcel. Log(PGA) - Interpolated Value Value of Log(pga) interpolated between contours as drawn 0y OES-GIS (EQE/OES. 1995). Interpolated PGA PGA (in units of g) computed from interpolated Log(PGA), above. Value Street Number Situs Street Number from the A ssessors File. Street Direction If included in the street name, the com pass onentation of the street (North. South. East. West. etc.). Street Name Situs Street Name from the A ssessors File. City from Assessor Situs Citv Name from the A ssessors File. Match Flag i The type of match resulting from the A ssessors File. I 1= A unique match on address 2 = A match to multiple parcels (Condominiums) 3 = A matcn to multiple parcels which were distinct and unique (split parcels) OES Damage ID A unique identifier relating the A ssessors Parcel data to the OES Damage Database. ; Evem aen. J.F. and J.M. Thomson. (1988). "Predictive Model for Important Ground Motion Param eters Associated with Large and Great Earthquakes." United States Geological Survey Bulletin 1838. ; Dewey. J.W.. B.G. Reagor. L. Dengler. and K. Moley (1995). "Intensity Distribution and Isoseismal Maps for the Northndge. California. Earthquake of January 17. 1994." U.S. Geological Survey Open-File Report 95-92. 425 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. MEMORANDUM Z J 10/99 ^ age 7 of ~ Table 2 (Continued) DHS Address Match Database - Field Definition and Description Field Name Description Tag The Building and Safety Placard applied to an inspected structure. : g=Green Tag (Safe to Occupy) y=Yellow Tag (Limited Access Only) ; r=Red Tag (Unsafe) u=Unfcnown Placard (Placard not identified to OES when reported). Estimated Damage Building and Safety Inspector's estimated damage in terms of repair . cost, when available. Percent Damage Percent damage to the structure estimated by the Building and Safety : Inspector, when available. 426 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. MEMORANDUM 211 0 / 9 9 A tta c h m e n t I L o s A n g e les C o u n ty A s s e s s o r 's T ax Roll U s e C o d e D e fin itio n s 427 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. _______________ 0000 RESIDENTIAL 00 (OPEN)______________________________ 010V VACANT tAND__________________ 01 SINGLE 3 r d d i q i t - 0 4 th d i g i t 1=Pool 4=Thcrapy Pool C=Condominium 0 = P lan n ed Res Oevelopment E=Condo C o n v er s io n f H o o p c r a t ive H=0wn-Your-Oun 02 DOUBLE, DUPLEX, OR TWO UNITS 03 THREE UNITS (ANT COMBINATION) 04 TOUR UNITS (ANT COMBINATION) 05 fIVE OR MORE APARTMENTS OR UNITS. COOPERATIVES OR OWN-TOUR-OWN PROJECTS NOT SEPARATELY PARCELLED. 3 r d d i g i t 4 th d i g i t C=4 S t o r i e s 1=Pool o r l e s s 9 = 0 th c r 5=5 S t o r i e s Improvements or more Only A=Coopernt ive B=0un-Your-0wn ^C o n d o m in iu m M=Modular -U to 00 06 MODULAR HOMES 3 rd d i g i t 0 = S in g le R es id e n c e 1=Mult i p l e R esid e n ce 4 th d i g i t 0=None 1=Pool 4=T herapy Pool C=Condominium 0 = P lan n e d Res. D evelopment ^ MOBILE HOMES 3rd d i g i t 4 th d i g i t 0 = S in g le 0 = A ssesse d by RP R e s id e n c e (P erm anent Tdn.) 1 - M u l t i p l e P= Asscssed by PP R e s id e n c e (No Perm anent f d n . ) 08 ROOMING HOUSES__________________________ 09 MOB HE HOME PARKS 3 r d d i g i t - 0 4 th d i g i t 1=Pool Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 1000 ‘ COMMERCIAL 10 (OPEN) 100V VACANT LAND 10 COMMERCIAL 3 r d d i o i t 0=Open =Miscel Ia neous Commercial 2=Art i s t in R es id e n c e 1 ICG STORES 12 STORE COMBINATION (WITH OfflCE OR RESIDENTIAL) 3 rd d i q i t 0 = S to re S O f f i c e C o m b in a tio n 1= S tore & R e s i d e n t i a l C o m b in a tio n 13 DEPARTMENT STORES 3 rd d i o i t 1= 0 is c o u n l D e partm ent S t o r e s 2 r f l u i l d i n g S u p p l i e s ( B u i l d e r s E m p o riu n s , e t c . ) !=Home F u r n i s h i n g s ( B a r k e r B r o t h e r s , e t c . ) - : R e t a i I-W arehouse C o m b in a tio n (L evi t z ) 5='Jarehouse S t o r e ( P r i c e Club, e t c . ) K SUPERMARKETS ! " d d i q i t C=Superm arkct- 12000* or more 1 • S u perm arket - 6000* th ro u g h 11999* 2=Small Food S t o r e s - L ess t h a n 6000* •u N ) V O 15 SHOPPING CENTERS (NEIGHBORHOOD, COMMUNITY) 16 SHOPPING CENTERS (REGIONAL) 17 OFFICE BUI 1 0 1NGS 3 r d d i g i t 1=Loft ty p e B u i I d i n g s 2 - O f f i c e and R e s i d e n t i a l 18 HOTEL AND MOTELS 3 r d d i g i t 0 = H o te ls - u n d e r 50 rooms 1= H otels- 50 rooms and over 2=M otels- u n d e r 50 u n i t s 3= H o tels - 50 u n i t s and o v e r 4= M o te (/H o te l and A partm ent C o m b in a tio n s - Under 50 u n i t s 5= M o tc l/H n te l and Apartm ent C o m b in a tio n s - 50 u n i t s and over 19 PROFESSIONAL BUILDINGS 3 r d d i g i t 1=Medical D e n ta l B u i l d i n g 2 = V e t e r i n a r y H o s p i t a l s , C l i n i c s Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. 2000 *C»*CRCIAL 20 (OPEN) 21 RESTAURANTS, COCKTAIL LOUNGES 3 rd d i g i t 0 = R c s t a u r a n t s , C o c k t a i l lo u n g e s , Taverns 1= f a s t food- Ualk Up 2 = f a s t food- Auto O r i e n t e d 22 UHOLESAIE AND MANU fACTURING OUTLETS 23 BANKS, SAVINGS t LOANS 24 SERVICE SHOPS RADIO t T.V. REPAIR REFRIGERATOR SERVICE PAINT SHOPS ELECTRIC REPAIR LAUNDRIES 25 SERVICE STATIONS 3 r d g i g i t 0 = f u l I S e r v i c e 1= S elf S e r v i c e 2 = S t a t i o n w ith Car Wash 26 AUTO, RECREATION EOPT., CONSTRUCTION EOPT., SALES & SERVICE 3 r d d i g i t 0=Auto S e r v i c e Shops (Body S fe n d e r Commercial G a r a g e ) 1=uscd Car S a le s 2=New Car S a le s t S e r v i c e 3 : C a r Wash 4=Car Wash - S e l f S e r v i c e Type 5 = R e c r e a t i o n Equipment S a l e s I S e r v i c e (Cam pers, M otor Homes, B o a t s ) 6 = f ar m and C o n s t r u c t i o n Equipment S a l e s t S e r v i c e 7=AUTO SERVICE CENTERS (NO GASOLINE) 27 PARKING LOTS (COMMER CIAL USE PROPERTIES) 3 r d d i g i t 0 = L o t s - p a t r o n or Employee 1=1 o t s - Commercial p a r k i n g 2 = P a r k in g S t r u c t u r e s - P a t r o n or Employee 3 = P a r k in g S t r u c t u r e s - Corrmercial p a r k i n g 28 ANIMAL KENNELS 29 NURSERIES'OR GREENHOUSES • f o r im proved p r o p e r t i e s , 4 t h d i g i t d e s c r i b e s th e number of s t o r i e s in th e m ain s t r u c t u r e ( w i t h t h e e x c e p t i o n of l i f t s o r c o n d o m i n i u m ) . See S e c t i o n 4 . 3 B. 2 t h r u 5 - t o i n d i c a t e t h e 0 o f s t o r i e s from 2 t h r u 5 6= t o i n d i c a t e 6 t h r u 13 s t o r i e s 7= t o i n d i c a t e 14 t h r u 20 s t o r i e s C=Condominiuns 8 = to i n d i c a t e 21 t h r u 30 s t o r i e s L = Lift ( e n t e r e d by L i f t Desk S e c t i o n ONLT) 9* t o i n d i c a t e o v e r 30 s t o r i e s NOTE: The 2nd d i g i t "G" in 0100 through 0900 s e r i e s i n d i c a t e s " L i f t e d " im provem ents. ■ P . w O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 3000 ‘ INDUSTRIAL 30 (OPEN) 300V VACANT LAND 30 INDUSTRIAL 3 r d d i q i t 0=0pen ^ M i s c e l l a n e o u s I n d u s t r i a l 2 = A r t i s t i h ' f l e s i a e n c e 31 LIGHT MANUFACTURING SMALL EQUIPMENT MANUFACTURING SMALL MACHINE SHOPS INSTRUMENTS MANUFACTURING PRINTING PLANTS 32 HEAVY MANUFACTURING 33 UAREH0USING, DISTRIBUTION, STORAGE 3 r d d i q i t 0 = W a r e h o u s i n g , D i s i r i b u t i o n under 10,000* 1=Uarehousi n g ,0 i s t r i b u t i on 10 ,0 0 0 ' 24,999* 2 = U a r e h o u s i n g , 0 i s t r i o u t i o n 25,000* th ro u g h 5 0 ,0 0 0 * 3 = U a r e h o u s i n g , D i s t r i b u t i o n o v er 5 0 ,0 0 0 * 4-Pub! ic S t o r a g e (B ek tn g .L y o n s) 5 = P u b lic S to r a g e - M in i w arehouse 34 FOOO PROCESSING PLANTS 3 rd d i q i t 0=Meat 1-B everage 2=Other 35 MOTION PICTURE,RADIO AND 4000 IRRIGATED FARM TELEVISION INDUSTRIES 40 (OPEN) 4010 PRIVATE RURAL PUMPING PLANT 3 rd d i q i t 41 FRUITS £ NUTS 0 = S tu d io s 42 VINEYARDS ^ T r a n s m i s s i o n F a c i l i t i e s 43 VINE £ BUSH FRUITS 2=Microwave R e l a y Towers 44 TRUCK CROPS 36 LUMBER YAR0S 45 FIELD CROPS 37 MINERAL PROCESSING 46 PASTURE 47 DAIRIES 3 rd d i q i t 48 POULIRY, EiC. 1=Cement,Rock £ G rave l P l a n t s 49 fEEO LOTS 2 = P e t r o l e i n R e f i n e r i e s , Chemical P l a n t s 38 PARKING LOTS (INDUSTRIAL USE PROPERTIES) 39 OPEN STORAGE 3 r d d i g i t W r u c k i n g Ccm panies, Term inals 2 = C o n t r a c t o r S t o r a g e Yards ' F o r improved p r o p e r t i e s , 4 t h d i g i t d e s c r i b e s t h e number of s t o r i e s i n th e main s t r u c t u r e ( w i t h t h e e x c e p t i o n o f l i f t s o r c o n d o m in iu m s ). See S e c t i o n A . 3 6 0=one S to r e y 2 t h r u S=to i n d i c a t e t h e t o f s t o r i e s from 2 t h r u 5 6 = to i n d i c a t e 6 t h r u 13 s t o r i e s 7s t o i n d i c a t e 14 t h r u 20 s t o r i e s fl=to i n d i c a t e 21 t h r u 30 s t o r i e s 9 m o indicate over 30 sto rie s Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 5000 DRT FARMS 50 (OPEN) 51 FRUITS & NUTS 52 VINEYARDS 53 f IELD CROPS 54 TASIURE 55 UMBER • PINE 56 IIMBER • FIR 57 TIHBER - REDUOOO 58 DESERT 59 UAS1E 6 0 0 0 'R E C R E A T IO N A L 60 OPEN 61 THEATERS 3 r d d i q i t 0=M ovie- In d o o r 1=Movie- D r i v e - I n 2=Legi t i m a t e T h e a te r 62 WATER RECREATION 3 r d d i q i t W e e Owned Boat S l i p 63 BOULING ALLEYS 64 CLUBS, LOOCE HALLS,FRATERN ORGANIZATIONS 7000 ‘ INSTITUTIONAL 65 ATHLETIC ANO AMUSEMENT FACILITIES 3 r d d i g i t 0 = A u d i t o r i i m s , Stadium s Ampni t h e a t e r s 1=Amusement F a c i l i t i e s 2=Com m ercial Swimming Pool S c h o o l s 3 = G y m n as iu n s, H e alth Spas 4=Dance H a l l s 5 = T e n n i s C o u r t s , C lubs P r o Sh o p s 66 GOLF COURSES 3 r d d i q i t 1-Non P r o f i t 2 = T h re e p a r 3 = H i n i a t u r e 67 RACE TRACKS 3 r d d i q i t W o r s e S t a b l e - P r i v a t e 60 CAMPS 3 r d d i g i t W r a i l e r an d Camper P a r k s ( o v e r n i g h t ) 69 SKATING RINKS 3 r d d i q i t 0 = l c e 1 = R o l l e r 70 (OPEN) 71 CHURCHES 3 r d d i q i t 1=Church P a r k i n g L ots 72 SCHOOLS (PRIVATE) 73 COLLEGE, UNIVERSITIES (PRIVATE) 74 nOSPITALS 3 rd d i q i t ^ C o n v a l e s c e n t H o s p i ta l :., N u rs in g Homes 75 HOMES FOR AGED & OTHERS 76 OPEN 77 CEMETERIES,MAUSOLEUHS.MORJUARIES 3 rd d i q i ts 0=C em e teries,M ausoleum s 1- M o r t u a r i e s , F u n e r a l Homes 78 OPEN 79 OPEN * For improved p r o p e r t i e s . 4 t h d i q i t d e s c r i b e s t h e number of s t o r i e s in th e main s t r u c t u r e ( w i t h e x c e p t i o n of l i f t s or co n d o m in iu m s ). See S e c t i o n 4 . 3 B. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 3225 . e l t a r e and S o c ia l S e r v ic e s 8000 MISCELLANEOUS “326 P o s ta l f a c i l i t y ::827 l i b r a r y 30 OPEN 3B23 : . u r t B u i l d i n g , J a i l 81 uinnr 3829 “ ill ta r y Post COMMERCIAL I MUTUAL: PUMPING PtANIS STATE ASSESSED PROPERTY 8330 ^ - U l i c S c h o o l . G e n e r a l 8351 C :lle g e 32 MINING 3332 n ig h School 8833 L1emcntary School 83 PETROLEUM 1 GAS 8834 School A d m in is tr a tio n C enter 8835 School S e rv ic e Center 84 PIPELINE, CANALS 8840 Recreat ion,General 85 RIGHT Of UAY 3841 P u b l ic park 8842 Art C e n t e r , Museum 86 UATER RIGHTS 8843 Publ ic Swiitming Pool 8844 ■(sorts Stadium 87 RIVERS I LAKES 8845 Beach 8346 r o r s e Stable 8000 GOVERNMENT OUNED PROPERTIES 234 7 4-,usemcnt Ride ("900'' PARCELS) 3848 Ball f i e l d ( L i t t l e L e a g u e , e t c . ) 8800 (OPEN) 3849 Tauth f a c i l i t y ( S c o u t s , e t c .) 8S0V VACANT LAND 8810 R ights of Uay, General 8850 water R elate d f a c i I i t i e s , General 8811 S t r e e t s , Road, Highway 8851 Small Boat Marina 3852 Boat S l i p 8812 f u t u r e S t r e e t , AI ley, e t c . 8853 Boat Mooring 8813 Power tra n s m is s io n Lines 8854 Pier,W harf 3814 sewers, U t i l i t i e s 2855 flo o d Control D rainage 3856 1- r i g a t i o n • R e l a te d u > u > 8857 0am 8858 R e s e r v o i r , Tank Underground S to rag e 8859 W atershed 3860 T r a n s p o rta t io n,General 8861 Harbor 1 R elate d 8862 A i r p o r t , General 8863 A i r p o r t , ! Hanger 8864 A i r p o r t , l i e - Down 8865 A i r p o r t , f ix c d - Based O p e rato r 8866 Rapid t r a n s i t , B u s , e t c . 2870 C oncession on P u b lic P r o p e r ty 8871 food Concession 3872 Souvenir Shop 8873 P a rk in g Lot Lease 8874 Off ic e Space Lease 8890 1 Comnunity Redevelopment 8891 P u b li c Housing 8899 Government P r o p e r t y and P o s s e s s o r y I n t e r e s t Not C l a s s i f i e d in Any of Above 8900 Dump S i t e s i!820 Caverrvncnt S e r v ic e s , General 3821 C ily riatl, A d m i n i s t r a t i o n Center 8822 A u x ilia ry and Regional Center 8823 P o li c e and f i r e S t a t i o n 3324 u t i l i t i e s O f f i c e ( P o w e r , w a t e r , e t c . MEMORANDUM 2/10/99 Attachment II Evemden Soil Code Definitions Soil C o d e Site Geology A Granitic and M etamorphic Rock B Paleozoic Sedim entary Rock C Early M esozoic Sedim entary Rock D C retaceous-E ocene Sedim entary Rock E Undivided Tertiary Sedim entary Rock F O ligocene-Pliocene Sedim entary Rock G Pliocene/Pleistocene Sedim entary Rock H Tertiary Volcanic Rock I I Q uaternary Volcanic Rock J Q uaternary Sedim entary Deposits (Shallow G roundwater) K.L.M O ther Q uaternary Sedim entary Deposits Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 434 APPENDIX 6: Final Report from EOE International 435 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Engineering Xi~k m Stiiety • Design February 16.1999 Maya Mahue-Giangreco Injurv and Violence Prevennon Program. Surveillance Section COUNTY OF LOS ANGELES. DEPARTMENT OF HEALTH SERVICES 313 N. Figueroa Street. Room 127 Los Angeles. California 90012 Subject: Final Report: Los Angeles County Northridge Earthquake Hospital Injury Database Enhancement and Analysis. EQE Project Number: 250658.01 Dear Mava: EQE International is pleased to provide to the County of Los Angeles. Department of Health Services (DHS). the results of the Northridge earthquake hospital injury' database enhancement and analysis. The principle intent of this project w as to determine available building and building damage data corresponding to locations of documented injunes in the Northridge earthquake. The major tasks of the project included: • Performing address review and matching of 3,436 hospital records from the Los Angeles County DHS injury database to the Los Angeles County Assessor's database and the OES/GIS Northridge Earthquake Inspection database. This database included baseline injury scene addresses prior to the earthquake, as well as earthquake-related injuries. • Association of matched addresses with Northridge ground motion maps (Modified Mercalli Intensity and peak ground acceleration) and developm ent of 8-1/2" x 11" color regional ground motion maps for presentation. In addition, smaller scale maps of heavily impacted service planning areas were also developed. • Limited comparison of documented earthquake-related injuries from the LA County injury database to injury' models used in EPEDAT. (EPEDAT is the Early Post- Earthquake Damage Assessment Tool, real-time earthquake loss assessment software developed by EQE for the California Office of Emergency Services). The goal of this sub-task was to plot Northridge injury data relative to existing casualty estimation models. INTRODUCTION Following the Northridge Earthquake, a diverse group of researchers created a small consortium to document and study earthquake-related injury data. The researchers included epidemiologists torm the Los Angeles County Department of Health Services and the Southern California Injury Prevention Research Center at UCLA, public health M acA rthtir Boulevard. Suite 4**0 a \e w p o r t Beach. CA *^2660-2027 L‘SA a Telephone H33-3303 • r AX (949) 442-9485 436 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. M aya M ahue-Giangreco Z ,'1 6 /9 9 Page 2 researchers from what would shortly become the Center for Public Health and Disaster Relief at UCLA, and engineers specializing in loss estimation from EQE International. The general goal of the research was to capture perishable data on injuries and deaths caused by the earthquake, as well as identify factors contributing to the risk of injury. Further, it was hoped that by integrating data on injuries and deaths with earthquake hazard and building damage data, we would advance the state-of-the-art and better quantify injury risk in future earthquakes. From the perspective of engineering-based loss estimation, there is room for much improvement in the estimation of earthquake-related injuries. Current models, based on data from previous earthquakes, relate generic injury rates to building damage states (described by a range of damage, e.g., 10 - 30%), regardless of building construction. It is our hope that additional in-depth study of injuries in earthquakes, such as Northridge, will allow researchers to refine existing injury models and facilitate improved estimates of casualties in future earthquakes. The current study represents the first steps to such an improvement; systematically analyzing injury data relative to earthquake hazard parameters and building construction information. 1.0 ADDRESS MATCH RESULTS DHS provided location information on 3,436 injury scenes to EQE International to be matched against the complete Los Angeles County Assessor's Tax Roll for the year 1993. The injury database included baseline injury data for the two weeks prior to the earthquake ("pre-earthquake"), in addition to data for the two weeks following the earthquake ("post-earthquake"). Post-earthquake injuries have been further categorized by DHS, to identify whether they were earthquake-related. Matching efforts were concentrated on those injuries identified as post-earthquake and earthquake-related. The results of the address match were returned to DHS, along with pertinent structure information including year built, construction class, square footage, etc. In addition, geocoding information (latitude and longitude) was available for 98% of matching records. The distribution of DHS injury data, and the results of the address match, are summarized in Table 1. .A s shown in the table, matching efforts, utilizing automated matching as well as "hand-matching" and manual review, resulted in 2,049 matches (60%). However, a reduced number of the injury records (650 records, or 19%) are considered "earthquake-related" (denoted with an asterisk in Table 1), and the match rate on these records was higher than the overall match rate, reaching approximately 71%. In addition, the injury data were matched to the EQE/OES Damage Database developed during the Northridge Earthquake response. Information pertaining to the Building and Safety Inspections following the Northridge Earthquake have been supplied to DHS for 377 injury records that had an Assessor's Parcel Number assigned as a result of the Assessor Database match, and were subsequently matched to the damage database. Of these 377 records, 64% (242) are associated with the post-earthquake injury records. The damage database match results for post-earthquake, earthquake-related injuries totaled 108 matches (24% of post-earthquake, earthquake-related injuries matched to the Assessor's database, and 17% of all post-earthquake, earthquake-related injuries). 4 3 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. M aya M ahue-G iangreco 2 /1 6 /9 9 Page 3 Table 1: DHS Injury Data Time Period ! Earthquake- Related (ER) Number of Records Address Match Results to Assessor's Database Percent Matched to Assessor's Database Match Results for OES/EQE Damage Data : Pre-Earthquake ‘ Clearlv E R 3 3 100% 0 1 Pre-Earthquake i NotER 1.261 834 66% 135 : Pre-Earthquake i Sub-Total i 1,264 837 66% 135 Post-Earthquake i Clearlv E R * 234 178 76% 38 ! Post-Earthquake i Not ER 1,522 753 49% 134 Post-Earthauake i Assumed E R * 405 275 68% 68 Post-Earthquake 1 Indirectiv E R * 11 o 55% 2 j Post-Earthquake ! Sub-Total 1 2,172 1.212 56% 242 ! TOTAL I 3,436 2,049 60% 377 Mote: * = Post-earthquake injuries considered to be "earthquake-related", and the focus of the present analysis. 1.1 ADDRESS MATCH TO THE LOS ANGELES COUNTY ASSESSOR’S DATABASE: The database provided by DHS indicated location information (i.e. street address) for injuries resulting from the Northridge Earthquake. The goal was to match the data received from DHS to the Los Angeles County Tax Assessor's Roll and the EQE/OES Damage Database, to identify the structure type and conditions where the injury was reported to have taken place. The address information in the DHS injury data was compared to the Los Angeles County Assessor's Database, which contains approximately 2.25 million parcel records. The results of the address match allowed EQE to develop a table that contained the site- specific information from the County Assessor's Tax Rolls for each reported injury location. Overall, nearly 60% of the total number of injury records supplied were matched to an address within the Los Angeles County Tax Assessor's Roll. Table 2 defines the fields provided from the Assessor's data, and gives a brief description of the field contents. In addition, a variety of geographic data previously associated with the parcel was also provided, including Modified Mercaili Intensity and peak ground acceleration experienced in the Northridge Earthquake, local soil conditions, census tract, and census block group. EQE has provided the data in an Excel file entitled DHS_ASSR.XLS (Table: DHS A ssessor Damage Match). 4 3 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mava M ahue-G iangreco 2 /1 6 /9 9 Page 4 1.1.1 Condominiums: While the majority of matches resulted in the selection of unique parcel sites, 112 street address matches resulted in the selection of multiple parcels. These records are denoted with a " Match Flag" of 2 (condominiums) or 3 (split parcels). Most of these multiple matches (90) are associated with condominium buildings1 , and aggregation of the key fields for all units within the identified structure (i.e., square footage and number of units) results in a reasonable representation of the building. For each condominium record, the "Square Footage Building" represents the aggregated area of all units within the building. "Square Footage Unit" represents the actual area of the unit in question (when unit number was provided with the injury' data) or a representative unit area (when the unit number was not provided with the injury data). For non-condominium structures, "Square Footage Unit" and "Square Footage Building" are equal. Condominium buildings represented in the current database range from 4 unit buildings with approximately 3600 square feet of floor space, to buildings with more than 200 units and more than 200,000 square feet of floor space. The remaining multiple matches resulted from other issues, such as split parcels. It should be noted that some condominium units are given a unique street address. In this case, the address match will not return multiple parcels; rather the resuits will be the return of the singular condominium unit. Note: The Los A ngeles C ounty Tax A ssesso r assigns each condom inium a unique parcel num ber for tax p u rp o ses. For condom inium s th at have a com m on stre e t ad d ress and unit num bers, multiple m atch es will e n su e . For condom inium s, w hich have unique s tre e t ad d resses, only th e unit in q u estio n will appear. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. M ava M ahue-Giangreco 2 /1 6 /9 9 P ag e 5 Table 2: DHS Address Match Database - Field Definition and Description Field Name ! Description 1 ENTNUM | DHS's Unique Identifier. A D D.NUM 1 Address Number orovided bv DHS. ADD_DIR I Street Direction (compass Direction) provided bv DHS. ADD NAME Street Name provided bv DHS. ADD _STYP Street Suffix provided bv DHS. A D D . APT Apartment or unit number (where applicable) provided bv DHS. c m ' Citv or communitv nam e provided bv DHS. LEGAL.CITY The legal name tor the City (or the name of the city in which the communitv resides) (i.e. Northridge is within the Citv of Los Angeles). LEGAL.CNTY The legal name for the Countv. ZIP Zip Code provided bv DHS. FULL .ADD Full Address provided bv DHS. Earthquake Code indicating whether the injury was earthquake-related, provided by DHS (l=earthquake-reiated, 2 = not earthquake-related. 5 = assum ed earthquake-related, and 9=*indirectly earthquake related) FaclCode Faciiitv code indicating hospital provided bv DHS. TIMEPD Code indicating time of injurv provided bv DHS. ISS Code indicating injurv severity provided bv DHS. Parcel Number Tax Assessor Roll Num ber for each situs. Situs Address Assessor's street address for the parcel. Situs Zip Code Parcel Zip Code. Use Code This code differentiates the major use of the parcel (Residential. Commercial) and the detailed use within the parcel (Single Family, Duplex. Service Station, etc.). Attachment I provides a complete listing of use definitions. Structure Type This field identifies the generalized construction of the im provem ent A=Steel Frame B= Concrete Frame C= Block/Brick/Other Concrete D= Wood Frame Year Built Year the improvement was constructed. Number ot Units Number of Residential Units, or office units in the improvement. Square Footage Unit This field identifies the actual square footage for single family dwellings and other non-condominium buildings, and TYPICAL square footage of a unit within the improvement for condominiums. Please refer to text in this mem o that further describes the method of handling condominiums. Square Footage Building This field contains the total square footage for the improvement. This number should equal the Unit Square Footage for Single Family Dwellings and other non-condominium buildings. Latitude Latitude of parcel Longitude Longitude of parcel 440 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ntava M ahue-G iangreco 2 .'1 6 /9 9 Page 6 Table 2: DHS Address Match Database - Field Definition and Description (Cont) Field Nam e Description Evemden Soil Code Evemden Regional Soils applied to the County Assessor's Tax Rolls by EQE/OES. See Attachment II for soil code definitions (Evemden. 19881- MMI Value The USGS Modified Mercalii Intensity (MMI) as mapped for the parcel in question. This value represents the shaking intensity experienced at the improvement (Dewev. 1995V5 Census Tract The 1990 Census tract the parcel resides in. This is supplied to allow for the application of Census data. Census Block Group The 1990 Census Block Group for the parceL Log(PGA) - Interpolated Value Value of Log(pga) interpolated between contours as drawn bv OES-GIS (EQE/OES. 1995). Interpolated PGA Value PGA (in units of g) computed from interpolated Log(PGA), above. Street Number Situs Street Number from the Assessor's File. Street Direction If included in the street name, the compass orientation of the street (North. South. East. West. etc.). Street Name Situs Street Name from the Assessor's File. Citv from Assessor Situs Citv Name from the Assessor's File. Match Flag The type of match resulting from the Assessor's File. 1= A unique match on address 2 = A match to multiple parcels (Condominiums) 3 = A match to multiple parcels which were distinct and unique (split parcels) OES Damage ID A unique identifier relating the Assessor's Parcel data to the OES Damage Database. Tag The Building and Safety Placard applied to an inspected structure. g=Green Tag (Safe to Occupy) y=Ye!low Tag (Limited Access Only) r=Red Tag (Unsafe) u=Unknown Placard (Placard not identified to OES when reported). Estimated Damage Building and Safetv Inspector's estimated damage m terms ot repair cost, when available. Percent Damage Percent damage to the structure estimated by the Building and Safety Inspector, when available. ; Evernden. J . F . and J.M . Thomson. (1988). "Predictive Model for Important Ground M otion Param eters A ssociated with Large and G reat E arthquakes." United States Geological S u rvey Bulletin 1838. 3 Dewey. J . W . . B.G. Reagor. L. Dengler, and K. Moley (1995). "Intensity Distribution and Isoseismat M aps for the Northridge, California, Earthquake of January 17, 1994," U.S. Geological S u rvey Open-FUe Report 95-92. 4 4 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mava M ahue-Giangreco z / i 6 /9 9 Page 7 1.2 Address Match to the EQE/OES Damage Database: Once the injury data had been matched to the Los Angeles County Tax Assessor's Roll, it was possible to refer to the Northridge Earthquake damage database developed by EQE International and the Governor's Office of Emergency Services. This database represents the best compilation of Building and Safety Inspection records from many of the local jurisdictions directly impacted by the earthquake. The damage database contains nearly 115.000 distinct Building and Safety damage inspection records. Each of these inspection records had previously been address matched to the Los Angeles County Tax Rolls, and the parcel number recorded. Of the 2,049 addresses matched to the County Tax Rolls, 18% (377 records) appeared in the damage database. Pertinent damage information for these 377 records has therefore been made available DHS, including: OES Damage ID, Building and Safety Placard (tag), estimated damage, and estimated percent damage. See Table 2 for field definition and description. 2.0 MAP DEVELOPMENT To depict the injury locations relative to indicators of hazard in the Northridge earthquake, ten (10) project maps were developed for DHS. EQE utilized the latitude and longitude previously associated with the Tax Assessor's database to locate injuries. In this manner, EQE obtained a latitude and longitude for virtually every record successfully matched to the Assessor's database; 59% (2,011 records) of the complete DHS injury database, and 69% of the earthquake-related injury records (449 records). In addition, DHS provided census tract data on Service Planning Areas (SPAs) that was used to create a map layer of SPA boundaries. These SPA boundaries are shown on each of the project maps. Several of the project maps show the official Modified Mercalli Intensities (MMI) for the Northridge Earthquake, developed by the USGS (Dewev, 1995), overlain with the DHS Injury database. The DHS Injury data were also placed over contours of peak ground acceleration developed by Dr. David Wald of the United States Geological Survey (USGS). These contours were developed from accelerograph recordings compiled from a variety of sources, including the California Division of Mines and Geology (CDMG), Los Angeles Department of Water and Power (LADWP), Southern California Edison (SCE), the University of Southern California (USC) and the USGS. EQE provided DHS with four (4) hard copy maps and one (1) overhead of each of the ten project maps on December 30.1998. The final project maps include regional views as well as close-up views of two areas, the San Fernando area and Southwestern Los Angeles, as follows: 1 . Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) - Regional Extent 2. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Modified Mercaili Intensity - Regional Extent 442 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Maya Mahue-Ciangreco 2/ 16/99 Page 8 3. Injuries Recorded at Emergence' Departm ents betw een 1 /1 7 and 1 /3 1 fo llo w in g the 1994 Northridge Earthquake (M w 6.7) and M odified Mercaili Intensity - San Fernando Extent 4. Injuries Recorded at Emergence' Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw o.7) and Modified Mercaili Intensity - Southwestern Extent 5. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Peak Ground Acceleration Contours (%g)- Regional Extent 6. Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Peak Ground Acceleration Contours (% g) - San Fernando Extent 7 . Injuries Recorded at Emergency Departments between 1/17 and 1/31 following the 1994 Northridge Earthquake (Mw 6.7) and Peak Ground Acceleration Contours (%g)~ Southwestem Extent 8. Injuries Recorded at Emergency Departments between 1/1/94 and 1/16/94 - Regional Extent 9 . Injuries Recorded at Emergencv Departments between 1/1/94 and 1/16/94 - San Fernando Extent 10. Injuries Recorded at Emergence' Departments between 1/1/94 and 1/16/94 - Southevestem Extent 3 .0 IN JU R Y MODEL COMPARISON This final task involved a limited comparison of earthquake-related injuries from the Los Angeles County injury database, to EQE's regional casualty models used in EPEDAT (the Early Post-Earthquake Damage Assessment Tool software). EPEDAT was developed for the California Office of Emergency Services to estimate regional earthquake impacts, such as building damage and casualties, for emergency response and planning purposes. EPEDAT utilizes enhanced versions of two published casualty algorithms to estimate a range of casualties expected on a regional basis. 3.1 EPEDA T Casualty M odels Most published casualty algorithms are simplified relationships between injury rates and building damage states (i.e.. the ATC-13 injury model suggests a minor injury rate of 3 per 1,000 occupants, or .003, for buildings within the moderate damage state, defined as 10 - 30% damage). The first algorithm used in EPEDAT was taken from ATC-13 (1985), while the second was taken from Whitman (1974). Both of these methodologies estimate injury and death rates from building damage category, although it should be noted that the two 443 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. M ava Nlahue-Giangreco 2 /1 6 /9 9 P age 9 algorithm s define the range of d am age in each state differently. W hile th ese casu alty estim ates are mean death and injury’ rates for any building w ithin a g iv en d a m a g e state, each dam age state includes a considerable range of possible dam ages. To ad eq u ately reflect the range of possible injuries w ith in a g iven dam age state, and the lik ely increase in casu alties toward the upper end o f the range, a probability distribution u tiliz in g the m ean casualty rate was applied to each in d iv id u a l bu ild in g dam age algorithm co n sid ered w ith in EPEDAT. The beta probability distribution w as selected, because th ese distributions were exam ined in ATC-13 and sh o w n to be superior for a p p roach in g the problem of m odeling uncertainty in earthquake dam age. The resulting b eta-m od ified algorithm s relate injury and death rates to predicted dam age for a g iven b u ild in g type, at various levels of M odified Mercaili Intensity. 3.2 Selection o f Injury Data for Model Comparison The pu rp ose of this task was to com pare, to the extent possible, recorded injury data to regional injury m odels. It should b e n oted that the injury m odels are b ased o n injury’ rates, estim ating the number of p eop le injured relative to the population ex p o sed . A ccordingly’, im plem entation of the com p arison w ou ld require an estim ate o f the ex p o sed p op ulation. It w as necessary, then, to w ork w ith in an area w here w e co u ld estim ate both the total population exposed, as w ell as be reasonably sure w e had captured all earthquake-related injuries. EQE d ev elo p s and maintains a variety o f detailed building inventory data files for u se in EPEDAT applications. Because p op u lation data is also required for casu alty m o d elin g , an area w ith detailed building inventory tabulated by’ census tract w as required. For the current com parison, a recently d ev e lo p e d database for the City of Los A n g eles w a s u tilized as the starting point for the b u ild in g and population exposure estim ates. This database w as assem bled from 1994 Los A n geles County' .Assessor's data, aggregated into EPEDAT building classes by cen su s tract. Because o f the database lim itation (i.e., m u st be w ithin the City’ of Los A n geles), injuries presented at H ospital B could n ot be an alyzed as it is beyond city lim its. M aps o f injury concentration for the rem aining h ospitals (A , C and D) were review ed to id en tify a reasonable study area. As the data from H osp ital A are considered in com p lete. H osp itals C an d D w ere used to identify the stu d y area. To ensure a reasonable com p arison o f m od el injury rates to actual injury rates, it is im portant to analyze areas w h ere w e have con fid en ce in both the estim ate of the ex p o sed population, and that the total num ber of d ocu m en ted injuries includes all earthquake-related injuries. Based, in part, on the review of the m ap o f injury concentration, w e have lim ited the analysis to census tracts within 5 m iles of H ospital C or D, and are therefore assu m in g that all injuries occurring w ithin 5 m iles o f H ospital C or D w ere treated at o n e o f the hospitals. That is, w e have assum ed that the current injury' database represents the "universe'' of injuries in all census tracts w ith in 5 m iles of Hospitals C an d D. W hile this a ssu m p tion is required to continue the analysis, it has a number of lim itations. First, a significant number of injury records w ere n ot successfully matched to the A ssessor's database, and therefore w ill not appear o n the m aps and w ill be om itted from this 4 4 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mava M ahue-G ianereco I,' 1 6 /9 9 Page 10 com p arison . Second, the injury database is lim ited to injuries recorded at the em ergency dep artm en ts o f just tour hospitals. Injunes treated at o th er h ospitals, clinics, or First A id stations, w ill n ot be included. Therefore, it is likely that th e "universe" of injuries actually u n d erestim ates the total number of injuries. A s iniury m o d els vary by structural type, a co m m o n structural type w ith a significant p op u lation exp osu re and num ber of associated in ju n es w a s selected; w ood fram e, sin gle fam ily d w ellin g s. The final analysis su b set o f injury data m a y be described as follow s: 1. Injuries occurred in the post-earthquake tim e p eriod (e.g., Tim epd = 2) 2. O n ly earthquake-related injuries are in clu d ed (e.g.. Earthquake = 1 (clearly earthquake-related). Earthquake = 5 (a ssu m ed earthquake-related), and Earthquake = 9 (indirectly earthquake-reiated)) 3. The injury record m ust have been su ccessfu lly m atch ed to the A ssessor's database, and the parcel record m u st also h a v e b een associated w ith a census tract location. 4. T he centroid of the census tract m u st be w ith in 5 m iles o f either H ospital C or D, a n d be within the C ity of Los A n geles. (A total o f 49 candidate census tracts w ere identified). 5. O n ly injuries occurring in sin gle fam ily, lo w -rise w o o d frame d w ellin gs w ere in clu d ed (e.g.. Use C ode = 0101 or 0100, sin g le fam ily dw elling, and Structure T yp e = D, W ood Frame.) The resu ltin g su b set of injuries included 72 of the 459 earthquake-related injuries that w ere su ccessfu lly m atched to the A ssessor's database (o u t o f 650 total). These m ay be further cla ssified by structure age, as considered by EPEDAT; pre-1950, and 1950 and later. M ost of the injuries in the subset occurred in pre-1950 d w ellin gs, reflecting the p red om in an t a g e of residential construction in the so u th w estern Los A ngeles stu d y area. Table 3 p ro v id es a breakdown of the analyzed injury data according to earthquake hazard (M odified M ercaili Intensity) and injury severity. Table 3 - Injury Data Included in M o d el C om parison Injurv Severitv 1 i ! l M ild M oderate i Serious U n k n ow n Total 1 ; MMI I Pre- I 1950+ Pre- j 1950+ | Pre- 1950 : Pre- 1950+ Pre- j 1950+ ! TOTAL 1950 1 1950 1 j 1950 1950 1950 i I VI 7 1 2 2 1 0 1 0 0 1 1 0 10 | 2 I 121 vii : 20 1 6 5 1 1 | 1 0 1 2 0 28 1 7 i 35 i VIII : 23 I 0 2 1 0 1 0 0 1 0 0 25 I 0 I 25 Total 1 50 1 8 9 1 1 | 1 0 1 3 0 63 I 9 ! 72 445 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mava Mahue-Ciangreco 2 /1 6 /9 9 Page 11 3.3 Determination o f E x p o se d Population W hile census data p ro v id es inform ation on the total p o p u lation resid in g w ith in a given census tract, im p lem en tation o f the injury' m odels requires segregation o f this data into a variety of structures a n d structure ty'pes. That is, the total n u m b er o f occupants w ill be d ivid ed am ong resid en ts o f sin g le fam ily dw ellings, apartm ent b u ild in gs, and condom inium s. For the purpose of the cu rren t com parison, occupant lo a d in g as d eterm in ed w ith in the EPEDAT database has b een u tilized . EPEDAT allocates resid en tial occupants through a com bination of cen su s d ata and published occupancy' algorith m s. ATC-13 (1985) provides a generic occu p a n cv algorithm that indicates that p erm an en t residential d w ellin gs w ill have a typ icai n igh ttim e occupancy of 3.1 p erson s per th ou san d square feet. H ow ever, this d o es n ot con sid er variations in p op u lation d en sity. That is, one w ou ld expect more o ccu p a n ts per thousand square foot in an inner-city area of m ulti- fam ily apartm ents (e.g.. East Los .Angeles), than in a m ore suburban neighborhood d om inated by larger sin g le fam ily h om es (e.g., Beverly H ills). To take th ese variations into account, EPEDAT u tilizes norm alization factors that scale the estim ated num ber of occupants to the actual n u m b er reported by the census data. For a g iv en cen su s tract, the total num ber of residential square feet is used to predict a prelim inary occupancv from the ATC-13 m odel. T his n u m b er is then scaled up to m atch the cen su s data, and the resulting m ultiplier u sed to adjust estim ates for in d ivid u al cla sses o f b u ild in gs w ithin the tract. W hile there may' be so m e error associated w ith these assu m p tion s, it reflects an im provem ent over the u n m o d ifie d occupancv algorithm . For each of the 49 cen su s tracts w h o se centroids fell w ith in 5 m iles o f either H ospital C or D, the number and total square footage of w ood frame, low -rise sin g le fam ily dw ellings w as determ ined. T hese sq u are footage estim ates w ere u se d w ith the cen su s tract norm alization factors to estim a te the total num ber of occupants w ith in the selected class of structure. Because of lim itation s associated w ith the n orm alization factors, five census tracts w ere rem oved from the data set. The final resulting data se t in clu d ed 8 census tracts at MMI VT, 21 cen su s tracts at MMI VII, and 15 cen su s tracts at MMI VIII. It is interesting to note that the 1990 census data indicates that there are 192,796 people living w ithin the 44 rem ain in g cen su s tracts. U sing the EPEDAT algorithm , 143, 668 (75%) are assum ed to resid e w ith in sin g le fam ily w ood fram e structures. The rem ainder w ill be distributed a m o n g m u lti-fam ily apartm ent buildings, co n d o m in iu m s, single fam ily dw ellings o f oth er con stru ction types, and other resid en tial structures. O f the 143,668 residents o f sin g le fa m ily w o o d frame structures, ap p roxim ately 115,000 (80%) live in pre-1950 structures, w h ile the rem aining 20% live in n ew er (1950 and later) hom es. 3.4 Comparison o f Injury M odels to Northridge Injury Data The population exp osu re data, an d associated injuries are su m m arized in Table 4. As can be seen in the table, m o st (77%) o f the exposed d w ellin gs w ere constructed prior to 1950, and are typically sm aller than the n ew er structures. In a d d ition , m ost o f the identified injuries occurred in th ese o ld er structures. A cursorv rev iew o f injury rates indicates that 4 4 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Slava M ahue-G iangreco 2/ 16/90 Page 12 the older structures exhibit higher average injury rates than the new er structures, w ith mjur\r nsk generally increasing w ith M odified Mercaili Intensity. It sh o u ld be noted that the lack of injury data tor m o d em 11950 and later) w ood frame d w ellin gs at MMI VIII indicates that the current sam p le is too sm all to adequately capture poten tial injuries. Table 4 Summary of Population Exposure Data for Single Family W ood Fram e D w ellings in 44 Selected Census Tracts Structure A ge ! MMI (#C Ts) | I N um ber of ; B uildings Building Area (SF) Average Area (SF) j EPEDAT ; Estim ated Population ' N um ber of DHS Injuries Average DHS Injury Rate Pre-1950 VI (8) 5.278 7.216.146 1367 19.913 1 9 4.5 x 10* Pre-1950 VII (21) 14.059 25.762.078 1.832 50,013 1 23 4.6 x 10* Pre-1950 VIII (15) 8.998 13.233366 1.470 45.165 i 25 5.5 x 10* Total: Pre- j 28,335 46,211,590 1,630 115,091 | 57 , 1950 I 1 ; 1950 and VI (8) 3.150 6,170,694 1,959 11,998 | 2 1.7 x 10* i later I j } 1950 and VII (21) 4.605 12,100,261 2,628 13,071 1 4 3.1 x 10* i later 1 i j 1950 and . VIII (15) 722 1,190.642 1,649 3 3 0 8 1 0 0 i later j i ! Total: 1950 j I 8,477 19,461397 2296 2 8 3 7 7 i 6 i and Later 1 ! i Overall 1 36,812 65.673,187 1,784 143,668 1 63 ! Total I ; i For com parison, injury rates for in d ivid u al census tracts may be plotted relative to the injury m odels currently' utilized w ith in EPEDAT. Figure I presents the in d ivid u al census tract injury rates for pre-1950 sin gle fam ily w ood frame dwellings, relative to both EPEDAT injury m odels. Each op en square sym bol on the chart represents the injury rate com puted for a single census tract. The figure show s that the injury rates estim ated from the current Northridge earthquake su b set fall w ell below the upper bound EPEDAT m odel (EPEDAT/W hitman), and are distributed more closely around the EPE D A T / ATC m odel. Figure 2 provides a sim ilar chart for post-1950 construction. A gain, the com puted Northridge injury rates fall w ell b elo w the EPEDAT/W hitman, and relate m ore closely to the EPEDAT/ ATC curve. It sh o u ld be kept in m ind that it is anticipated that the current data set underestimates the actual num ber of injuries, and hence w ill underestim ate the resultant injury rate. W hen sufficient data exists, it is a lso possible to exam ine the distribution o f injury rates for a given structure type at various intensity levels. Figures 3 and 4 present the distribution of injury rates for pre-1950 dw ellings, and dwellings constructed in 1950 and later, respectively'. Also sh o w n on the figures are the number of data points (n), the resulting mean injury rates, and the estim ated standard deviation. A lth ou gh the overall num ber of data points may' be too sm all to draw definite conclusions, the figures can be 4 4 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. M dva M ahue-O iangreco 2/ 16/09 Page 13 used to identify general trends. .As sh o w n o n Figures 3 and 4, there is no d em onstrated difference between the mean injury rates for w o o d frame dw ellings in MMI VI a n d VII, although pre-1950 dw ellings do sh o w an increase in m ean injury rate b etw een M M I VII and VIII. However, the standard d eviation increases consistently w ith MMI (recogn izin g the limitations of the database under consideration), indicating an increasingly w id er dispersion of injury rates w ith increasing MMI. The lim itations of the data are m ore ob viou s on Figure 4, which dem onstrates that all injury rates for MMI VIII are zero. The mean injury rates for a given MMI m ay be com pared to the lower bound EPEDAT/ATC m odel rates, as sh o w n in Table 5. For pre-1950 w ood frame d w ellin g s, the N orthridge earthquake injury rate is larger than the m odel rate at MMI VI and VII, and sm aller than the m odel rate at MMI VIII. H ow ever, if one exam ines the range d efin ed by the com puted "mean plus one sigma" and the "mean m inus one sigma" injury rates, the m odei rates are mostly w ithin these b ou n d s, or of the sam e order of m agnitude as the bounding values. For exam ple, at MMI VIII. the Northridge earthquake m ean injury rate plus one standard deviation is roughlv of the sam e order of m agnitude as the m o d el rate (0.00131 vs. 0.00196). TABLE 5 COMPARISON OF NORTHRIDGE EARTHQUAKE M EAN INJURY RATES WITH EPEDAT/ATC INJURY M ODEL FOR PRE-1950 WOOD FRAME DW ELLINGS : j Building A ge MMI Sample I Size (n) 1 N orthridge M ean Injurv Rate N orthridge Standard D eviation Northridge m ean m inus one standard deviation | N orthridge i m ean p lu s j one standard 1 d eviation EPEDAT- ATC Injury Rate Pre-1950 VI 7 5.0 x 10-* 3.5 x 1C H 1.5 x 10-* 8.5 x 10-* I 9.0 x 10-5 Pre-1950 VII 21 5.0 x 10-* 4.2 x 10-* 8.0 x 10-s 9.2 x 10-* I 2.8 x 10-* Pre-1950 VIII 15 7.0 x 10-* 6.1 x 10-* 6.1 x 10-* 1.31 x 10-3 1 1.96 x 10-J Conclusions and Recommendations There are many limitations associated w ith the current injury rate com parison, including: • The injury data was lim ited to injuries recorded at the em ergency d epartm ents o f four hospitals, oniv tw o of w h ich w ere used for the m odel com parison. Injuries not presented at em ergency departm ents, such as injuries treated at clinics, or at first aid stations w ill be om itted. Further, only about 70% of injury records w ere successfully matched to the A ssessor's database and were available for m o d el comparison. Accordingly, it is exp ected that the resulting injury rates underestimate actual injury rates. • The population exposure estim ates are based on 1990 census data. The ex p o sed population is distributed am o n g various residential structure types by m eans of a generic occupancy algorithm id en tify in g the number of people per thousand square feet of residential occu p an cv, scaled up or dow n to reflect the total cen su s 448 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. M ava N lahue-G iangreco Zf 16/90 P age 14 tract exposure. While this is expected to be an improvement over the generic algorithm, it does not consider variations in density between single and multi- family structures within the same census tract. • The limited number of data points (e.g., 44 census tracts) are not sufficient to draw definitive conclusions regarding the appropriateness of existing injury models for single family wood frame dwellings, and no other structure types are addressed under the current comparison, Despite the limitations of the current assessment, it is clear that the current approach to validating and improving existing injury modeling techniques is appropriate. The N'orthndge earthquake was the first modem, urban earthquake in the US. for which both injury data and damage data were collected in a systematic way, and it provides an excellent opportunity for assessment and improvement of existing injury modeling techniques. Additional injury model review should be conducted, and would be greadv facilitated by • the collection and review of additional injury’ data (from Northridge and other earthquakes) • examination of additional parameters (injury’ severity’, building damage and loss, building inspection tag, other ground motion parameters) • development of additional building and population exposure databases for areas where Northridge injury data already exists (e.g.. Hospital B ) • extension of the model comparison to other abundant structure types, such as multi-family structures and apartments. • development of guidelines for post-earthquake collection of injury and building damage data to facilitate future analyses and model improvements We have appreciated the opportunity to work with you on this important research topic, and look forward to additional collaboration in the future. Please feel free to call should you have any questions. Sincerely’, EQE INTERNATIONAL, INC. cc: Ronald T. Eguchi, EQE International Attachments: Evemden Soil Code Definitions Los Angeles County Assessor's Tax Roll Use Code Definitions 449 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mava M ahue-Giangreco 2 /1 6 /9 9 Page 15 References: ATC (1985), Earthquake Damage Evaluation Data for California. Applied Technology Council, ATC-13, Redwood City, California. Dewey, J.W ., B.G. Reagor, L . Dengler, and K. Molev (1995). "Intensity Distribution and Isoseismal Maps tor the Northridge, California, Earthquake of January 17,1994," US- Geological Survey Open-File Report 95-92. Evemden. J.F . and J.M. Thomson. (1988), "Predictive Modei for Important Ground Motion Parameters Associated with Large and Great Earthquakes." United States Geological Survey Bulletin 1838. EQE International. Inc. and the Governor’s Office of Emergency Services (1995). The Northridge Earthquake ot fanuarv 17.1994: Report of Data Collection and Analysis. Part A : Damage and Inventory Data. Irvine. California Whitman, R.V., J.M. Biggs, J . Brennan III, C.A. Cornell, R . de Neufville, and E.H. Vanmarcke (1974), Methodology and Pilot Application - Seismic Design Decision Analysis. Massachusetts Insntute of Technology Civil Engineering Report R74-15. 450 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. 0012 Figure 1 - Injury Model Comparison - Single Family Wood Frame Dwellings, Pre-1950 - EPED A T/W H ITM A N 001 A-EPEDAT/ATC 0 008 □ Injuries in Tracis “ 5 m i ol Hosp C or D - Northridge EQ f l) 2 t a 0006 0004 0002 O /l 7 MMI ms ^ Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 2 - Injury Model Comparison - Single Family Wood Frame Dwellings, 1950 and Later 001 0 009 0 008 0 007 0 006 0 0 0 5 0 0 0 4 0 0 0 3 0 002 . 1 - m - EPED AT/W H ITM AN * • -EPEDAT/ATC £ 3 □ Injuries in T racis within 5 mi of Hospitals C or D - Northridge EQ i ^ 1 0 001 o f r 6 ■ u K ) 7 M M I Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 3 - Distribution of Injury Rates for Single Family Wood Frame Dwellings, Pre-1950 33 u e H M a N c • O 0 k . 1 E a Z -u ut VI n = 7, m ean = 0 0005. s i clew - 0 00035 VII n = 21, m ean = 0 0005, s i dev = 0 00042 VIII n = 15, m ean = 0 0007, s i dev = 0 00061 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 4 - Distribution of Injury Rates for Single Family Wood Frame Dwellings, 1950 and Later a u c a « e o 2 E 3 Z V I n = 8, mean - 0 0001. si dev - 0 00010 V II n = 20. mean = 0 0001 si dev = 0 0001b V III n = 15. mean - 0 0 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. M ava M ah u e-G ian g reco 2 /1 6 /9 9 A ttachment I Evernden Soil Code Definitions Soil Code Site Geology A Granitic and M etam orphic Rock B Paleozoic Sedimentary Rock C Early Mesozoic Sedimentary Rock D Cretaceous-Eocene Sedimentary Rock E Undivided Tertiary Sedimentary Rock F Oligocene-Pliocene Sedimentary Rock G Pliocene/Pleistocene Sedimentary Rock H Tertiary Volcanic Rock 1 Quaternary Volcanic Rock J Quaternary Sedimentary Deposits (Shallow Groundwater) K,L,M Other Quaternary Sedimentary Deposits - C t K J \ M ava M ahue-G ianereco 16/99 A t t a c h m e n t II L o s A n g e le s C o u n ty A s s e s s o r 's T a x Roll U s e C o d e D e f in itio n s Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 456 Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. _____________0000 RESIDENTIAL 00 (OPEN)________________________________ 010V VACANT LAMP___________________ 01 S1NCLE 3rd digit-0 Lth digit 1=Pool 4=Therapy Pool Condominium D=Planned Res Development E-Condo Conversion F-Cooperat ive H=0wn*Your-0wn 02 DOUBLE, DUPLEX, OR TW O UNITS 03 THREE UNITS (ANT COMBINATION) 04 FOUR UNITS (ANT COMBINATION) 05 FIVE OR M ORE APARTMENTS O R UNITS. COOPERATIVES O R O'JN-YOUR-OUN PROJECTS NOT SEPARATELY PARCELLED. 3rd digit Lth digit 0=4 Stories 1=Pool or less 9=0ther 5=5 Stories Improvements or more Only A=Cooperative B=Own-Your-Onn C=Condominium K=Modutar ■ C * L /t - J 06 HOOULAR HOM ES 3rd diqit 4th diqit 0=Sing(e 0=None Residence 1=Pool 1=Multip(e 4=Therapy Pool Residence C'Condominiim 0'Planned Res. Development 07 MOBILE HOM ES 3rd diqit 4th diqit 0=Single O'Assessed by RP Residence (Permanent fdn.) 1'Multiple P=Assessed by PP Residence (No Permanent Tdn.) 08 ROOMING HOUSES 09 HOBILEHOME PARKS 3rd digit-0 4th d ig it T*Pool Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. 1000 ‘ C0W1ERCIAL 10 (OPEN) 100 V VACANT LAND 10 COM M ERCIAL 3rd d iait 0=Opcn ^Miscellaneous Comacrcial 2 = Artist in Residence 1100 STORES 12 STORE COMBINATION (WITH OFFICE OR RESIDENTIAL) 3rd diqit 0=Store I Office Combination I^Store I Residential Combination 13 DEPARTMENT STORES 3rd diqit I'Discount Department Stores 2: Bui(ding Supplies (Builders Emporium, e tc .) J^Home Furnishings (Barker Brothers, e tc .) irRetaiI-Warehouse Combination (Levitt) 5=uarehouse Store (Price Club, e tc .) K SUPERMARKETS I 'd diqit C=Supermarkct- 12000* or more 1=Supermarket- 6000* through 11999* 2-Small Food Stores- Less than 6000* ■ p . co 0 0 15 SHOPPING CENTERS (NEIGHBORHOOD, CO M M U N ITY )______________________ 16 SHOPPING CENTERS (REGIONAL) 17 OFFICE BUILDINGS 3rd digit l=Loft type Buildings 2=0ffice and Residential IB HOTEL AND MOTELS 3rd d igit 0=Hotels- under 50 rooms 1=Hote(s- 50 rooms and over 2-Hotels- under 50 units 3=Motels- 50 units and over 4'Hotel/Hoiel and Apartment Combinations- Under 50 units 5=Motel/Hotcl and Apartment Combinations- 50 units and over 19 PROFESSIONAL BUILDINGS 3rd diqit UMedicat Oental Building 2-Veterinary Hospitals, Clinics Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. _________2000 ‘ COWgRCIAL 20 (OPEN)___________________ 21 RESTAURANTS, COCKTAIL LOUNGES 3rd diqit 0=Restaurants, Cocktail Lounges, Taverns 1=fast food' Walk Up 2=Fast Food- Auto Oriented 22 WHOLESALE A N D HANU FACTURING OUTLETS 23 BANKS, SAVINGS 4 LOANS 24 SERVICE SHOPS RADIO I T.V. REPAIR REFRIGERATOR SERVICE PAINT SHOPS ELECTRIC REPAIR LAUNDRIES 25 SERVICE STATIONS 3rd diqit 0=Full Service l=Solf Service 2=Station with Car Wash u n v O 26 AUTO, RECREATION EQPT ., 27 PARKING LOIS (COMMER CONSTRUCTION EOPT., CIAL USE PROPERTIES) SALES & SERVICE 3rd d iqit 3rd diqi t 0=Lots-patron or Employee 0-Auto Service Shops 1=Lots-Commercial parking (Body 4 Fender Commercial 2=Parking Slructures- G a r a g e ) Patron or Employee l=uscd Car Sales 3=Parking Structures- 2=Ncw Car Sales 1 Service Commercial parking 3*Car Uash 28 ANIMAL KENNELS 4=Car Wash - Self 29 NURSERIES'OR Service Type GREENHOUSES 5=Recreation Equipment Sales 4 Service (Campers, Motor Homes, Boats) 6*farm and Construction Equipment Sales I Service 7=AUTO SERVICE CENTERS (NO GASOLINE) ‘ For improved properties, 4th d igit describes the number of sto ries in the main stru ctu re (with the exception of l i f t s or condominium). See Section 4.3 B. 2 thru 5s to indicate the # of stories from 2 thru 5 6=to indicate 6 thru 13 stories 7* to indicate 14 thru 20 sto ries C^Condaminiuns 8=to indicate 21 thru 30 sto ries L=Lift (entered by Lift Desk Section ONLY) 9* to indicate over 30 stories NOTE: The 2nd digit "G" in 0100 through 0900 series indicates "Lifted" improvements. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 3000 “ INDUSTRIAL 30 (OPEN) 300V VACANT '.AND 30 INDUSTRIAL 3rd diqit 0=Open ^Miscellaneous Industrial 2=Artist in'Residence 31 LIGHT MANUFACTURING SHALL EQUIPMENT MANUFACTURING SHALL MACHINE SHOPS INSTRUMENTS MANUFACTURING PRINTING PLANTS 32 HEAVY MANUFACTURING 33 UAREH0USING, DISTRIBUTION, STORAGE 3rd d iqit 0=Uarehousing,0istribuiion under 10,000“ 1=Warehous i ng,0 i s t r i bu t i on 10,000* 24,999“ 2=Uarehousin3, 0 istrio u t ion 25,000* through 53,000“ 3=Uarehousing,0istribution over 50,000“ 4=Public Storage (Beking,Lyons) 5cPublic Storage-Mini uarehouse 34 FOOO PROCESSING PLANTS 3rd diqit 0*Meat 1-Beverage 2=Other 35 M OTION PICTURE,RADIO AND 4000 IRRIGATED FARM TELEVISION INDUSTRIES 40 (OPEN) 4010 PRIVATE RURAL PUMPING PLANT 3rd diqit 41 FRUITS I NUTS 0=Studios 42 VINEYARDS ^Transmission F a c ilitie s 43 VINE t BUSH FRUITS 2=Microwave Relay Towers 44 TRUCK CROPS 36 LUM BER YARDS 45 FIELD CROPS 37 MINERAL PROCESSING 46 PASTURE 47 DAIRIES 3rd diqit 48 POULTRY, ETC. UCement.Rock 1 Gravel Plants 49 FEED LOTS 2=Petroleu» Refineries, Chemical Plants 38 PARKING LOIS (INDUSTRIAL USE PROPERTIES) 39 OPEN STORAGE 3rd digit ^T rucking Companies, Terminals 2=Contractor Storage Yards •for improved properties, 4th d ig it describes the number of s t o r i e s in the main structure (with the exception of l i f t s or condominiim). See Section 4.3 B 0*one Storey 2 thru 5=to indicate the » of sto ries from 2 thru 5 6=to indicate 6 thru 13 s to r ie s 7*to indicate U thru 20 s to rie s 8=to indicate 21 thru 30 s to r ie s 9=to indicate over 30 s to r ie s On O V I U J a C C — O « • u a t « c 1 3 5 X « a «> — CJ CJ u c j v i o * * * V 1 ( 1 V I ■ A • a V I a t O t O X l * M CJ « -*■ X £ 0 — 5 5 * - V I a t O - J a » c • U J U J V I a. Ul o X X ► — o «C X 3 U J - J o SJ — « • — « O L V I C l 3 C W S . o 3 • • L v C I £ 1 ? 2 ■ ? • u* u . ( 0 CL 1 5 > a e a . U J 2 » z 3 V I W c 1 V o CJ X Ul V o » Q U J O < a t 3 5 _ V I V* U J * * X u _ VI «H CJ C l L L c_ Q . ? > o CJ O • 3 u m X L . v V L M VI o M ^ z U J n v » u j a x — C J -O a t 3 3 3 x o u . 3 X ( j V I 8 * U J U J •— o < U J > — J ■ — « c o — "D a. v i T 3 — » c « •— > Ul C c . O 3 O w V I U J z ac cr U J — * - 1 3 U J V o fc* 3 u z CJ o z w f Cj C m X — w C x 2 r ° U W fc- O x u . C J K » H C J V I o o f C J w O u C K l ii 5 X CL O X T 3 U J u C J K » C J X II II o — CL o U J CL O w o U _ • i CJ CJ CJ x *W V 3 3 w O v i O r s . fs. INI N » c «*r r< - (A 76 N . £ • « • — — u CJ * J m > U J z o o 8 ^ - i < f N j v» z m « 3 O — I Q C c j o o z < 1/ 1 UJ C J — — c r -i ' 5 C J T > — C # C — w — C j C i- o o m a o X V I ■*- a a O — f \ l -I T J O L O K 1 VI U a ■ 5 T5 1 Z » • N M v i T J a. X *D < «- U K l VI w o* c 9 * 5 C j « i — u o — a c I I I I o s - 1 V I « « * 2 x 2 VI ^ o ^ -5 UJ D c r O C * D U l I - T) < u 3 K l INI - O < a 461 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 8825 welfare and Social Services 8000 MISCELLANEOUS BB26 Postal f a c ility SB27 Iibrary 80 OPEN 3823 C:urt Building, Jail 81 UllLiir 2829 mi Iitary Post COM M ERCIAL t MUTUAL: PUMPING PI A N IS STATE ASSESSED PROPERTY 8830 pjblic School, General 8831 Col lege 82 MINING 3832 nigh School 8833 Elementary School 83 PETROLEUM 1 CAS 8834 School Administration Center 84 PIPELINE, CANALS 8835 School Service Center 8840 Recreat ion,General 85 RIGHT OF U A Y 3841 Pub!ic park 8842 Art Center,Museum 86 UATER RIGHTS 8843 Publ ic Swimriing Pool 8844 Sports Stadium 87 RIVERS t LAKES e845 Beach 8846 norse Stable 8000 G O V ER N M EN T O UN ED PROPERTIES 8847 Amusement Ride (■•900" PARCELS) 8848 Ball field ( L ittle League.eic.) 8800 (OPEN) 8849 Youth f a c ility ( S c o u ts,etc.) esov VACANT LAND 8810 Rights of Uay, General 8850 Water Related f a c i I i t i e s ,General 8811 Streets, Road, Highway 8851 Small Boat Marina 3852 Boat Slip 8812 future Street, A1 ley, etc. 8853 Boat Mooring 8813 Power Transmission Lines 8854 Pier,Wharf 3814 Sewers, U tilitie s 8855 Flood Control Drainage 8856 1-rigation • Related 8857 Dam 8858 Reservoir, Tank Underground Storage 8859 Watershed ' 8860 Transport at ion,General 8861 Harbor t Related j 8862 Airport, General i 8863 Airport,! Hanger , 8B64 Airport.Tie • Oown j 8865 A irp o rt,fU ed - Based > Operator ] 8866 Rapid Transit,Bus,etc. 8870 1 Concession on Public j Property 8871 Food Concession 8872 Souvenir Shop 8873 Parking Lot Lease 8874 Office Space Lease 8890 C ommun i t y Redeve1opment 8891 Public Housing 8899 Government Property and Possessory Interest Not C lassified in Any of Above 8900 Dunp Sites 8820 Coverrvnent Services, General 3821 City nail, Administration Center 8822 Auxiliary and Regional Center 8823 Police and Fire Station 8824 U t ilitie s Office (Power,Water,etc. O n N > APPENDIX 7: Likelihood Ratio Tests: Model Selection Process 463 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission o f th e copyright owner. Further reproduction prohibited without permission. Table A. 7-1. Polvtomous Logistic Regression: Demographic Model Development and Likelihood Ratio Tests. Chl-Sq. -2 ‘ Log Chl-Sq. Chl-Sq. LRT Proportional Odds Model #_________________________ Model Varlable(s)___________________________ n_______Likelihood______ dth_______LRTA ______ p-value________ p-value 1 Gender 641 765 207 1 8.175 0 0042 0 885 2 Ethnicity* 420 548 189 2 2952 02285 0.1479 3 Age (Trend) 634 743.356 1 26.997 0 0001 0.0516 4 Age Grouped 634 742204 5 28149 00001 0 2508 5** Gender, Ethnicity, Age Group, Hospital 418 518443 10 31 706 0 0004 0 2966 6 Gender, Ethnicity, Age Group Trend, Hospital 418 520.129 6 30.021 00001 02502 7 Gender 418 547.567 1 2583 010 8 0.6638 8 Ethnicity 418 547 332 2 2818 0.2443 01471 9 Age Group 418 533 096 5 17.054 00044 0.1909 10 Age (Trend) 418 534 622 1 15.488 0.0001 01336 11 Hospital 418 538 411 2 11 739 00028 06276 12 Deleted Gender from Model 5 418 518 994 9 31.156 00003 0.3458 13 Deleted Ethnicity from Model 5 418 521.719 8 28.431 00004 0 2734 14 Deleted Age Group from Model 5 418 53317 5 16.98 0.0045 0.516 14-t Deleted Age Group Trend from Model 6 418 533.17 5 16.98 00045 0 5 1 6 15 Deleted Hospital from Model 5 418 529.839 8 20.311 00092 0 2018 Compare Model 12 to 5 (Gender | Ethnicity, Age, Hospital) 418 1 0.55 n.s *** Compare Model 13 to 5 (Ethnicity | Gender, Age, Hospital) 418 2 3.275 n.s.*** Compare Model 14 to 5 (Age | Gender, Ethnicity, Hospital) 418 5 14.726 p < 0.025 Compare Model 14-t to 6 (Age Trend | Gender, Ethnicity, Hospital) 418 1 13 041 p < 0.0005 Compare Model 15 to 5 (Hospital | Gender, Ethnicity, Age) 418 2 11.395 p < 0.005 Interactions 16 #5 with AA*Female Interaction 418 518.429 11 31.721 0.0008 0.3355 17 #5 with Hisp*Female Interaction 418 516555 11 33 595 00004 0.3887 18 #5 with Twenty‘Female Interaction 418 516.562 11 33.587 00004 02869 £ Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Tabla A7-1. (Continued) ___ -2 * Log Chl-Sq. Chl-Sq. LRT Proportional O dds M odel* Model Varlable(s) n Likelihood df~ LRTA p-value p-value 19 #5 with Fourty*Female Interaction 418 518.061 11 32089 00007 03142 20 #5 with Fifty‘Female Interaction 418 518.306 11 31.844 00008 0.2677 21 #5 with Age-gt-Sixty*Female Interaction 418 516.106 11 34.044 00004 0.0401 22 *5 with Age-lt-20*Female Interaction 418 518.428 11 31.722 0.0008 0.3754 Compare Model 16 to 5 (AA*Female Interaction) 418 1 0015 n.s.*** Compare Model 17 to 5 (Hisp*Female Interaction) 418 1 1.889 n.s.*** Compare Model 16 to 5 (20's*Female Interaction) 418 1 1.881 n.s.*** Compare Model 19 to 5 (40's*Female Interaction) 418 1 0.383 n.s.*** Compare Model 20 to 5 (50's‘Female Interaction) 418 1 0.138 n.s*** Compare Model 21 to 5 (>=60*Female Interaction) 418 1 2.338 n.s.*** Compare Model 22 to 5 (<20*Female Interaction) 418 1 0 0 1 6 n.s.*** - df=Degrees of Freedom A LRT=Likelihood Ratio Test; for interaction models 16-22, Chi-Square LRT on 1 df - difference in likelihoods from interaction model versus model 5. * Deleted Ethnicity-Other' from Polytomous Model, no observations in ISS=8 group. ** Core Demographic Model, Final Demographic Model ***n.s.= p > 0.05 (not statistically significant)_____________________________________________________________________________________________ O n isi Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-2. Dichotomous Logistic Regression Model Development: Demographic Characteristics of Patients Identified with Earthquake-Related Injuries from 4 Emergency Departments. Los Angeles County. January 17-31.1994. ■ t* O n O N M o d e l# M odel V a ria b le (s) n -2 * L og L ik e lih o o d df~ C h l-S q . LRT* C h i-S q . LRT p - v a lu e 1 G e n d e r 641 6 2 1 .6 5 8 1 8 .039 0 .0 0 4 6 2 Ethnicity* 4 4 0 4 5 1 .8 7 7 3 4 .6 1 0 0 .2 0 2 7 3 A ge (T rend) 63 4 6 0 2 .3 2 7 1 2 4 .3 4 3 0.0001 4 A ge G ro u p ed 6 3 4 6 0 1 .5 6 3 5 2 5 .1 0 6 0.0001 5** G en d er, Ethnicity, A ge G roup, H ospital 43 8 4 2 3 .3 2 8 11 3 2 .195 0 .0 0 0 7 5-t G en d er, Ethnicity, A ge G roup T rend, H ospital 43 8 4 2 4 .8 6 0 7 3 0 .6 6 3 0.0001 6 G e n d e r 43 8 4 5 2 .3 8 3 1 3 .140 0 .0 7 6 4 7 Ethnicity 4 3 8 4 5 1 .0 1 0 3 4 .5 1 3 0 .2 1 1 2 8 A ge G roup 4 3 8 4 3 9 .9 1 4 5 1 5.609 0.0081 9 A ge G ro u p T ren d 43 8 4 4 1 .1 5 7 1 1 4 .3 6 6 0 .0 0 0 2 10 H ospital 4 3 8 4 4 3 .0 1 5 2 12.508 0 .0 0 1 9 11 D elete d G e n d e r from M odel 5 43 8 4 2 4 .1 2 6 10 3 1 .3 9 7 0 .0 0 0 5 12 D elete d E thnicity from M odel 5 4 3 8 4 2 7 .7 2 4 8 2 7 .7 9 9 0 .0 0 0 5 13 D elete d A ge G ro u p from M odel 5 4 3 8 4 3 5 .7 3 0 6 19.793 0 .0 0 3 0 13-t D elete d A ge G ro u p T ren d from M odel 5-t 4 3 8 4 3 5 .7 3 0 6 19,793 0 .0 0 3 0 14 D elete d H ospital from M odel 5 4 3 8 4 3 5 .0 0 4 9 2 0 .5 1 9 0 .0 1 5 0 C o m p a re M odel 11 to 5 (G en d er | Ethnicity, A ge, H ospital) 1 0 .7 9 8 n.s. C o m p a re M odel 12 to 5 (E thnicity | G en d er, A ge, H ospital) 3 4 .3 9 6 n.s. C o m p a re M odel 13 to 5 (A ge | G en d er, Ethnicity, H ospital) 5 12.402 p < 0 .0 5 C o m p a re M odel 13-t to 5-t (A ge T rend | G en d er, Ethnicity, 1 10.870 p < 0 .0 0 5 H ospital C o m p a re M odel 14 to 5 (H ospital | G en d er, Ethnicity, A ge) 2 11.676 p < 0 .005 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-2. -2 * L o g C h i-S q . C h i-S q . LRT p- M odel #________________________ M odel V a rla b le (s)___________________________ n________ L ik e lih o o d _______ d f^ ________ LRTA ____________ v a lu e In teractions 15 M odel 5 with Ethnicity x S ex Interaction 4 38 42 3 .2 4 7 12 3 2.276 0 .0 0 1 3 16 M odel 5 with Ethnicity x A ge Interaction 4 3 8 4 2 3 .3 2 7 12 32 .1 9 5 0 .0 0 1 3 17 M odel 5 with S ex x A ge Interaction 438 4 2 0 .5 2 5 12 34 .9 9 8 0 .0 0 0 5 C o m p a re M odel 15 to 5 (Ethnicity x S ex Interaction) 4 38 1 0.081 n.s. C o m p a re M odel 16 to 5 (Ethnicity x A ge Interaction) 438 1 0 .0 0 0 n.s. C o m p a re M odel 16 to 5 (S ex x A ge Interaction) 4 38 1 2 .8 0 3 n.s. - d f= D e g re e s of F re e d o m A LRT=Llkelihood R atio T est; for interaction m o d els 15-17, C h i-S q u are LRT on 1 df= d ifferen ce In likelihoods from interaction m o d el v e rs u s m o d el 5. •D eleted E th n ic tty -O th e r' from P o ly to m o u s M odel, no o b se rv a tio n s In IS S = 8 group. " C o r e D em o g rap h ic M odel ***n.s.=not statistically significant_____________________________________________________________________________________________________________________ o \ Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A 7-3. Potvtomous Logistic Regression: Injury Characteristics Model Development Model # Model Varlable(s) n -2x Log Likelihood d f- Chl-Sq. LRTA Chl-Sq. LRT p value Chl-Sq. Proportional O dds p-value Body Location - missina values excluded ln=637) 1 Head/Neck 637 767.638 1 4018 0045 0.2016 2 Upper Extremities 637 765.257 1 6 399 0.0114 0.1605 3 Lower Extremities 637 769.963 1 1.692 01933 0.3537 4 Trunk (Back, Chest, Abd, Trunk) 637 768.68 1 2976 00845 0.0256 5 Head/Neck, Upper Ext, Trunk (Lower Ext is Reference Category) 637 760.142 3 11513 00093 0.0481 External Cause - missina. unknown, other, mv. poison, (ire. burn, overexertion values excluded (n=512) 6 Cut/Pierced by 512 575 424 1 21 707 0.0001 0.4766 7 Falls 512 549324 1 47 807 0.0001 0.7866 8 Struck by/Caught in/Caught between 512 590 537 1 6.595 0.0102 08853 9 Falls, Cut/Pierced (Struck by a s reference category) 512 546.661 2 50 47 00001 0 8398 Body Location and External Cause - missing, unknown excluded, specified mechanisms excluded (n=512) 10 Body Location (4 Groups) & Mechanism (3 Groups) 512 539.373 11 Body Location (4 Groups) 512 584.396 Model 10 Compared to Model 9 (Body Location | Mechanism) 512 Model 10 Compared to Model 11 (Mechanism | Body Location) 512 5 57.758 00001 0.2629 3 12.736 00052 0.2143 3 7.288 n.s.** 2 45.022 p < 0.0005 Interactions 12 Model 10 plus Fell-Arm Interaction 512 538.875 6 58.256 0.0001 0.0452 13 Model 10 plus Fell-Head Interaction 512 539.007 6 58.125 0.0001 03411 14 Model 10 plus Fell-Body Interaction 512 538.074 6 59.058 0.0001 02295 15 Model 10 plus Cut-Foot Interaction 512 537.471 6 59661 0.0001 0.3064 16 Model 10 plus Cut-Head Interaction 512 538868 6 58 264 00001 0.2902 17 Model 10 plus Cut-Arm Interaction 512 539366 6 57 765 00001 02155 Compare Model 10 to Model 12 512 1 0.498 n.s.** Compare Model 10 to Model 13 512 1 0.366 n.s.** Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-3. (Continued) M odel#_______________________ Model Varlable(s) Compare Model 10 to Model 14 Compare Model 10 to Model 15 Compare Model 10 to Model 16 Compare Model 10 to Model 17 Demographics. Body Location and External Cause 18 Gender 19 Ethnicity 20 Age 20-t Age Group Trend 21 Mechanism 22 Body Location 23 Hospital 24* Gender, Ethnicity, Age, Body Location, Mechanism, Hospital 24-t Gender, Ethnicity, Age Group Trend, Body Location, Mecanism., Hospital 25 Deleted Gender from Model 24 26 Deleted Ethnicity from Model 24 27 Deleted Age from Model 24 28 Deleted Body Location from Model 24 29 Deleted Mechanism from Model 24 30 Deleted Hospital from Model 24 Compare Model 25 to 24 (Gender ( Ethnic, Age, Body Loc, Mech, Hosp) Compare Model 26 to 24 (Ethnicity | Gender, Age, Body Loc, Mech, Hosp) Compare Model 27 to 24 (Age | Gender, Ethnic, Body Loc, Mech, Hosp) Compare Model 27 to 24-t (Age Trend|Gender, Ethnic, Body Loc, £>. Mech, Hosp) Sjj Compare Model 28 to 24 (Body Loc | Gender, Ethnic, Age, Mech, Hosp) C R tS q T - -2 x Log Chl-Sq. LRT p Proportional n Likelihood d f- Chl-Sq. LRT* value O dds p-value 512 1 1.299 n.s.** 512 1 1902 n.s.** 512 1 0 505 n.s.** 512 1 0.007 n.s.** 330 407.31 1 2.187 0.1392 0.9577 330 408.937 2 0 5 6 0.7556 0.3817 330 391.144 5 18.354 00025 0.5013 330 393232 1 16.265 0.0001 0.0744 330 375.871 2 33.627 00001 08257 330 400.662 3 8.835 0.0316 0.1053 330 403239 2 6.259 00437 0.8506 330 353.317 15 56.181 0.0001 0.1382 330 355.849 11 53.648 00001 0.1543 330 353.318 14 56.18 0.0001 0.2369 330 355.902 13 53.595 0.0001 0.2791 330 363.314 10 46.184 0.0001 0.1856 330 359.028 12 50.469 0.0001 048 9 330 376.445 13 33053 0.0017 0.1233 330 358.705 13 50.793 00001 0.184 330 1 0.001 n.s.** 330 2 2586 n.s.** 330 5 9997 n.s.** 330 1 7.464 p < 0 01 330 3 5.712 n.s.** Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A 7-3. (Continued) Model# Model Varlable(s) -2x Log n Likelihood df~ Chl-Sq. LRTA " "" Chl-Sq. Chl-Sq. LRT p Proportional value Odds p-value Compare Model 29 to 24 (Mechanism | Gender, Ethnic, Age, Body Loc, Mosp) 330 2 23128 p < 0 0005 Compare Model 30 to 24 (Hospital | Gender, Ethnic, Age, Body Loc, Mech) 330 3 5388 n .s “ ~d.f.=Degress of Freedom A LRT=Likelihood Ratio Test; for interaction models 12-17, Chi-Square LRT on 1 df 'Core Model, Final Model “ n.s.=not statistically significant = difference in likelihoods from interaction model versus model 10 ' J o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-4. Dlchotomous Logistic Regression Model Development: Injury & Demographic Characteristics with R espect to Severity of Earthauake-Related injury. M odel # M odel V a ria b le (s) n -2 x L o g L ik e lih o o d d f~ C h l-S q . LRTA C h i-S q . LRT p - v a lu e B odv L ocation - m issin a & unknow n v a lu e s ex clu d ed (n= 637) 1 H ead/N eck, U pper Ext, Trunk (L ow er Ext is R e fe re n c e C ategory) 63 7 6 1 6 .5 5 2 3 11.42 0 .0 0 9 7 E xternal C a u s e - m issina. unknow n, o th er, mv. poison, fire. burn, overexertion v a lu e s ex clu d ed fn s 512) 2 Falls, C u t/P ierced , Slip/Trip, M otor-vehicle collisions, po iso n in g s, overexertion (S truck by a s re fe re n c e categ o ry ) B odv L ocation a n d E xternal C a u s e (n = 553t 555 4 9 0 .8 0 0 6 6 1 .8 6 2 0.0001 3 B ody L ocation 553 538.051 3 13.727 0 .0 0 3 3 4 M ech an ism o f Injury 553 4 8 9 .1 5 5 6 6 2 .6 2 3 0.0001 5* B ody L ocation & M ech an ism o f Injury 5 5 3 4 8 0 .5 8 9 9 7 1 .1 8 8 0.0001 M odel 3 C o m p a re d to M odel 5 (M ech an ism of Injury | B ody L ocation) 553 6 57.461 p < 0 .0 0 0 5 M odel 4 C o m p a re d to M odel 5 (B ody L ocation | M ech an ism o f Injury) In teractio n s 553 3 8 .5 6 5 p < 0 .0 5 6 B ody L ocation, M ech an ism , & Interaction 55 3 4 8 0 .4 9 6 10 7 1 .2 8 2 0.0001 M odel 6 C o m p a re d to M odel 5 (B ody L ocation x M ech an ism Interaction) 553 1 0 .094 n.s.** ■u - j Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-4. (Continued) M odel #_________________ M odel V a ria b le (s)_______________ D em o g rap h ics. B odv L ocation a n d E xternal C a u se 7 G en d er 8 Ethnicity 9 A ge 10 A ge G ro u p T rend 11 H ospital 12 B ody L ocation 13 M ech an ism of Injury 14A A G en d er, Ethnicity, A ge, B ody L ocation, M ech an ism , H ospital 14-t G en d er, Ethnicity, A ge G ro u p T rend, B ody L ocation, M ech an ism , H ospital 15 D eleted G e n d e r from M odel 14 16 D elete d E thnicity from M odel 14 17 D eleted A ge from M odel 14 18 D elete d B ody L ocation from M odel 14 19 D eleted M ech an ism from M odel 14 2 0 D eleted H ospital from M odel 14 C o m p a re M odel 15 to 14 (G en d er | E thnic, A ge, B ody Loc, M ech, H osp) C o m p a re M odel 16 to 14 (Ethnicity | G en d er, A ge, B ody Loc, M ech, H osp) C o m p a re M odel 17 to 14 (A ge | G e n d e r, Ethnic, B ody Loc, M ech, H osp) C o m p a re M odel 16 to 14 (B ody Loc | G en d er, E thnic, A ge, M ech, H osp) -2 x L o g C h i-S q . LRT p- n L ik e lih o o d d f~ C h l-S q . LRTA __________v a lu e 375 386.441 1 2 .3 1 4 0 .1 2 8 2 375 3 8 6 .4 2 5 3 2 .3 3 0 .5 0 6 7 375 371.831 5 16.924 0 .0 0 4 6 375 3 7 3 .2 8 8 1 15.467 0.0001 375 3 7 8 .1 7 8 2 10.577 0.005 375 3 7 7 .4 7 7 3 11.278 0 .0 1 0 3 375 3 4 6 .4 6 8 6 4 2 .2 8 7 0.0001 375 31 8 .9 7 4 20 6 9 .7 8 0.0001 375 3 2 1 .2 1 8 16 67 .5 3 7 0.0001 375 3 1 8 .9 6 9 19 6 9 .7 6 6 0.0001 375 3 2 2 .0 1 2 17 6 6 .7 4 3 0.0001 375 3 2 8 .1 9 3 15 6 0 .5 6 2 0.0001 375 3 2 7 .5 8 3 17 61.171 0.0001 375 3 4 9 .7 2 4 14 39.031 0 .0 0 0 4 375 3 2 6 .7 7 6 18 6 1 .9 7 8 0.0001 37 5 1 0.0 1 4 n.s.** 375 3 3.037 n.s.** 375 5 9 .2 1 8 n.s.** 375 3 8.6 0 9 p < 0.0J Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-4. (Continued) -2 x L og C h i-S q . LRT p- M odel # M odel V a ria b le (s) n L ik e lih o o d d f~ C h i-S q . LRTA v a lu e C o m p a re M odel 19 to 14 (M ech an ism | G en d er, 375 6 30.749 p < 0 .0 0 0 5 E thnic, A ge, B ody Loc, H osp) C o m p a re M odel 10 to 14 (H ospital | G en d er, 375 2 7 .8 0 2 p < 0 .0 2 5 E thnic, A ge, B ody Loc, M ech) ~ d .f.= D eg ress of F re e d o m ,'L R T =U kelihood R atio T est; for interaction m o d els 6, C h i-S q u are LRT on 1 df= d ifferen ce in likelihoods from interaction m o d el v e rs u s m o d el S. * C o re M odel **n.s.=not statistically significant A A F inal M odel Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-5. Potvtomous Logistic Regression: Structural Characteristics. Model Development Model M Model Variables Square Footage 1 Square Footage- Building (Quartiles- Trend) 415 2 Square Footage - Building (Quartiles - Less than 1468 sq. ft. as reference 415 category) 3 Square Footage - Unit (Quartiles - Trend) 415 4 Square Footage - Unit (Quartiles • Less than 1350 sq ft. as reference category) 415 Tagging 5 Tagging (Green Tags a s reference category) 100 6 T egging (Untagged as reference category) 608 Structure Construction Type 7 Wood-Framed Structures (Missing, Other, Steel Frame, Concrete Frame, URM, 641 Special as reference category) Year Built 8 Year Built (5 Categories Based on Code Revisions - Trend) 413 9 Year Built (5 Categories Based on Code Revisions- GE 1988 as reference 413 category) 10 Year Built (4 Categories Based on Shoaf Paper - GE 1976 as reference category) 413 11 Year Built (4 Categories Based on Collapsing Top2C ategoriesfrom M odel#8- 413 Trend) 12 Year Built (4 Categories Based on Collapsing Top 2 Categories from Model # 8 - 413 GE 1976 a s reference category) Structure Use 13 Multifamily Condos/Apts/Coops (Single Family & Duplexes as reference 405 category) 14 Use Coded (All Others as reference category) 641 All Structural Characteristics (Reduced Model! 15 Mx-fam Use, Year Built, Sq Ft Bkfg, 16 Year Built, Sq Ft Bldg, Sq Ft Unit 393 15 Mx-fam Use, Year Built, Sq Ft Bkfg, Sq Ft Unit 393 4k Chl-Sq. Prop. -2 * Log Chl-Sq. Chl-Sq. LRT Odds Assump p Likelihood df LRT p-value________value 511 437 1 1.127 0.2885 03939 511.387 3 1.177 0.7585 0.4608 512564 1 0000 0.9946 0.2058 511.207 3 1357 07156 0.3611 119 086 1 0 05 9 0.8079 0.1798 730.052 2 0757 0 6848 00001 772.133 1 1 249 02638 0.5063 503 803 1 2.300 0.1294 0.7934 500.892 4 5210 0.2664 09018 502.867 3 3235 0.3568 0.3094 504.69 1 1.412 0.2347 0.8099 502.872 3 3230 0.3574 08339 490952 1 2621 0.1054 0.837 773.382 1 0.000 09989 0.3571 472.993 10 11 883 0.293 0.9241 479 304 9 5.572 0 7819 0.9166 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-6. (Continued) Model # Model Variables 17 Mx-fam Use, Sq Ft Bldg, Sq Ft Unit 18 Mx-fam Use, Year Built, Sq Ft Unit 19 Mx-fam Use, Year Built, Sq Ft Bldg Compare Model 16 to Model 15 (Mx-fam use | Year Built & Sq Ft) Compare Model 17 to Model 15 (Yr Bit | Use, Sq Ft) Compare Model 18 to Model 15 (Sq Ft Bldg | Yr Bit, Use, Sq Ft Unit) Compare Model 19 to Model 15 (Sq Ft Unit | Yr Bit, Use, Sq Ft Bldg) Structural Characteristics (Reduced Model) with Interactions 20 Use, Yr Bit, Sq Ft Bldg, Sq Ft Unit, Yr Bit * Use Interaction Compare Model 20 to Model 15 Main Effects. Reduced Model 21 Mx-fam Use 22 Year Built 22-t Year Built Trend 23 Gender 24 Ethnicity 25 Age Group 25-t Age Group Trend 26 Hospital 27* Use, Yr Bit, Gender, Ethnic, Age Gp, Hospital 27-t Use, Yr Bit, Gender, Ethnic, Age Gp Trend, Hospital 27-tt Use, Yr Bit Trend, Gender, Ethnic, Age Gp, Hospital 28 Gender Deleted from Model 27 29 Ethnicity Deleted from Model 27 30 Year Built Deleted from Model 27 30-t Year Built Trend Deleted from Model 27-tt 31 Structure Use Deleted from Model 27 32 Hospital Deleted from Model 27 33 Age Deleted from Model 27 Chl-Sq. Prop. -2 * Log Chl-Sq. Chl-Sq. LRT Odds Assump p n Likelihood df LRT p-value________value 393 476638 7 8238 0.3121 08473 393 477166 7 7.710 0 3589 0.8611 393 473 458 7 11.418 01214 09373 393 1 6.311 p < 0 025 393 3 3645 n s 393 3 4.173 n.s 393 3 0 4 6 5 n s 393 467 771 11 17.105 0.1048 0.9485 393 1 5222 p < 0 025 259 344.174 1 0.686 0.4077 0.303 259 336085 3 8.775 00324 09897 259 339 238 1 5622 0.0177 0.8708 259 341 385 1 3.475 0.0623 0.4932 259 344.847 2 0.013 0.9935 05272 259 337.247 5 7613 0.1789 0.088 259 337.535 1 7.324 00068 0.0573 259 335.528 2 9.332 0.0094 06607 259 320.945 14 23.914 00469 04709 259 321.266 10 23.594 00088 0.5817 259 323822 12 20 978 00507 0.3315 259 322 397 13 22 463 00486 0 3862 259 321 288 12 23572 00232 03 3 8 259 327.31 11 17549 0.0926 0 2284 259 327.31 11 17.549 00926 0.2284 259 322 575 13 22.285 0.0511 0.599 259 324748 12 20.112 0.065 0.4667 259 325 019 9 19.841 0.0189 0.7934 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-5 .1Continued) M odel# Model Variables n -2* Log Likelihood df Chl-Sq. LRT Chl-Sq. Prop. Chl-Sq. LRT O dds A ssum p p p-value value 33-t Age Trend Deleted from Model 27-t 259 325.019 9 19.841 00189 07934 Compare Model 28 to 27 (Gender | Use, Yr Bit, Age, Ethnic, Hosp.) 259 1 1 451 ns*** Compare Model 29 to 27 (Ethnicity | Use, Yr Bit, Age, Gender, Hosp.) 259 2 0.342 ns.*** Compare Model 30 to 27 (Yr Built | Use, Gender, Age, Ethnic, Hosp.) 259 3 636 5 ns*** Compare Model 30-t to 27-tt (Yr Built Trend | Use, Gender, Age, Ethnic, Hosp) 259 1 3.429 n s * ‘* Compare Model 31 to 27 (Use | Yr Bit, Age Gp, Ethnic, Gender, Hosp) 259 1 1 629 ns*** Compare Model 32 to 27 (Hospital | Use, Yr Bit, Gender, Ethnic, Age, Hosp) 259 2 3802 ns*** Compare Model 33 to 27 (Age | Use, Yr Bit, Gender, Ethnic, Hosp) 259 5 407 3 ns*** Compare Model 33-t to 27-t (Age Trend | Use, Yr Bit, Gender, Ethnic, Hosp) 259 1 3.753 ns*** Interaction 34 Model 27 with Use * Year Built Interaction 259 320797 15 24063 0 064 0522 35“ Model 27 with 3-level interaction 259 319135 17 25 725 0.0796 0.5726 36 Model 27 with ethnicity-structure use interaction 259 320 869 15 23991 0.0652 05423 37 Model 27 with ethnicity-year built interaction 259 320.05 15 24810 00526 03772 Compare Model 34 to 27 259 1 0.149 ns*** Compare Model 35 to 27 259 3 1.811 ns.*** Compare Model 36 to 27 259 1 0077 ns.*** Compare Model 37 to 27 259 1 0 896 ns.*** 38 Model 35 with Age Trend 259 319.436 13 25424 0.0203 0.7398 39 Model 35 with Year Built Trend 259 322 703 15 22156 0.1038 0.4797 Reduced Model with Taoaino Information Included with Previously Identified Stat. Sianif. Structural Variables 40 Mx-fam, Yr Bit, Red & Yellow Tags 88 103.699 5 453 9 04747 0.4838 41 Model 38 with Old-MxFam, '43-'60-Mx-Fam, & '61 -'75-MxFam Interactions 88 101.516 8 6.722 0.5669 0.0001 Compare Model 41 to Model 40 88 3 2.183 n.s * * * * LRT=Likelihood Ratio Test; for interaction models 20,34, 35, 39, Chi-Square LRT on 1 df = difference in likelihoods from interaction model versus ^ model 15,27,27,38, respectively. * Core Structural Characteristics, Adjusting for Demographics ** Final Model, retaining interaction because it was previously detected before data reduction.________________________________________________ Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A 7-0. Model Development: Structural Characteristics and Demographics. Model # Model Variables -b. -4 Sauare Footaoe 1 Square Footage - Building (Quartiles - Trend) 415 2 Square Footage - Building (Quartiles - Less than 1468 sq. ft as 415 reference category) 3 Square Footage - Unit (Quartiles - Trend) 415 4 Square Footage - Unit (Quartiles - Less than 1350 sq ft. as reference 415 category) Tagging 5 Tagging (Green Tags a s reference category) 100 6 Tagging (Untagged a s reference category) 641 Structure Construction Tvdo 7 Wood-Framed Structures (Missing, Other, Steel Frame, Concrete 641 Frame, URM, Special a s reference category) Year Built 8 Year Built - Trend 413 9 Year Built 413 Structure Use 10 Muttifamily Condos/Apts/Coops (Single Family & Duplexes as 393 reference category) 11 Use Coded (All Others a s reference category) 641 All Structural Characteristics (Reduced Model! 12* Mx-fam Use, Year Built, Sq Ft Bldg, Sq Ft Unit 393 13 Year Built, Sq Ft Bldg, Sq Ft Unit 393 14 Mx-fam Use, Sq Ft Bldg, Sq Ft Unit 393 15 Mx-fam Use, Year Built, Sq Ft Unit 393 16 Mx-fam Use, Year Built, Sq Ft Bldg 393 -2 * Log Chl-Sq. LRT p- LlXellhood_______ df Chl-Sq. LRT*A _______ value 417 147 1 0 945 0 331 417.07 3 1.021 0.7961 418.077 1 0014 0.9043 416.601 3 1.491 0.6844 97 229 1 0.016 08991 629.672 2 0 0 2 5 0 9875 628.322 1 1.375 0.2409 412.978 1 1.461 02268 411 363 3 3.076 0.38 394.936 1 2.23 0.1353 629.69 1 0.0007 0.9327 385.653 10 11.514 0.3189 391.398 9 5.769 0.7628 388 924 7 8.242 03117 389.831 7 7.335 0.3948 386.335 7 10.832 0.1461 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-9. (ContinuedI Model # Model Variables n Compare Model 13 to Model 12 (Mx-fam use | Year Built & Sq Ft) 393 Compare Model 14 to Model 12 (Yr Bit | Use, Sq Ft) 393 Compare Model 15 to Model 12 (Sq Ft Bldg | Yr Bit, Use, Sq Ft Unit) 393 Compare Model 16 to Model 12 (Sq Ft Unit | Yr Bit, Use, Sq Ft Bldg) 393 Structural Characteristics (Reduced Modell with Interactions 17 Use, Yr Bit, Sq Ft Bldg, Sq Ft Unit, Yr Bit * Use Interaction 393 Compare Model 17 to Model 12 393 Univariate: Reduced Model (Structural Characteristics with Demographic Characteristics) 18 Mx-fam Use 275 19 Year Built 275 20 Year Built Trend 275 21 Gender 275 22 Ethnicity 275 23 Age Group 275 24 Age Group Trend 275 25 Hospital 275 26** Use, Yr Bit, Gender, Ethnic, Age Gp, Hospital 275 26-t Use, Yr Bit Trend, Gender, Ethnic, Age Gp, Hospital 275 26-tt Use, Yr Bit, Gender, Ethnic, Age Gp Trend, Hospital 275 27 Gender Deleted from Model 26 275 28 Ethnicity Deleted from Model 26 275 29 Year Built Deleted from Model 26 275 29-t Year Built Trend Deleted from Model 26-t 275 30 Structure Use Deleted from Model 26 275 31 Hospital Deleted from Model 26 275 32 Age Deleted from Model 26 275 32-t Age Group Trend Deleted from Model 26-tt 275 ■fk -J o o -2 * Log Chl-Sq. LRT p- Likelihood df Chl-Sq. LRT** value 1 5.745 p < 0.025 3 3.271 n.s.A 3 4.178 n.s.A 3 0.682 n .s A 381.315 11 15.852 0.1467 1 4 338 p < 0 05 290 383 1 0677 0.4107 282.172 3 8 888 0 0308 286.044 1 50 1 6 0 0251 287.361 1 3 6 9 9 00544 290.017 3 1.043 0.7909 284.397 5 66 6 2 0.247 284.811 1 6 249 0.0124 281.145 2 9 9 1 5 0.007 266.235 15 24.825 00523 269 491 13 21.568 0.0624 266 654 11 24.406 0.0111 267 523 14 23.537 0.0521 267 412 12 23647 00227 273155 12 17905 0.1186 273.155 12 17905 0.1186 267.505 14 23 555 00518 271.068 13 19.992 0.0954 269127 10 21.933 00155 269.127 10 21.933 0.0155 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-8. (Continued) -2 * Log Chl-Sq. LRT p- Model # Model Variables n Likelihood df Chl-Sq. LRTA A value Compare Model 27 to 26 (Gender | Use, Yr Bit, Age, Ethnic, Hosp.) 275 1 1 288 n .s A Compare Model 28 to 26 (Ethnicity | Use, Yr Bit, Age, Gender, Hosp.) 275 2 1 178 n.s.ft Compare Model 29 to 26 (Yr Built | Use, Gender, Age, Ethnic, Hosp.) 275 3 6.92 n.s.A Compare Model 29-t to 26-t (Yr Built Trend| Use, Gender, Age, 275 1 3.663 n.s A Ethnic, Hosp.) Compare Model 30 to 26 (Use | Yr Bit, Age Gp, Ethnic, Gender, Hosp) 275 1 1 27 n.s.A Compare Model 31 to 26 (Hospital | Use, Yr Bit, Gender, Ethnic, Age, 275 2 4.833 n.s A Hosp) Compare Model 32 to 26 (Age | Use, Yr Bit, Gender, Ethnic, Hosp) 275 5 2892 n.s A Compare Model 32-t to 26-tt (Age Trend | Use, Yr Bit, Gender, Ethnic, 275 1 2.473 n.s.A Hosp) Interaction 33 Model 26 with Use * Year Built Interaction 275 265 859 16 25.2 00664 34*** Model 26 with Dummy Interaction Variables 275 263.264 18 27.795 0.0652 34-t Model 34 with Year Built Trend 275 268.166 16 22 893 0.1166 34-tt Model 34 with Age Group T rend 275 263.698 14 27.362 0.0173 Compare Model 33 to 26 275 2 0.375 n.s.A Compare Model 34 to 26 275 3 2.97 n.s.A Reduced Model with Taaaina Information Included with Previously Identified Stat. Sianif. Structural Variables 35 Mx-fam, Yr Bit, Red & Yellow Tags 59 57 862 5 4.364 0.4983 36 Model 35 with Demographics 59 38 462 16 23.763 0.0948 Compare Model 36 to Model 35 59 11 194 n s A A n.s.=not statistically significant (p > 0 05) A A LRT=Likelihood Ratio Test; for interaction models 17,33,34, Chi-Square LRT on 1 df=difference in likelihoods from interaction model versus model 12, 26, 26, respectively. * Core Model, Structural Characteristics ** Final Model, Structural Characteristics with Demographics *** Final Model, with Interaction. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-7. Polvtom ous Logistic Regression: Geologic Characteristics Model Development. Chi-Sq. Prop. -2 * Log C hi-Sq. C hi-Sq. LRT O d d s A s s u m p p- M o d e l# M odel V ariables n L ikelihood df LRT p -v alu e v alu e 1 Soil G roups (R ock as reference category) 419 5 0 8 8 1 6 2 3 2 8 6 0.1934 0.0619 2 Liquefiable vs. All Other 419 5 0 8 9 8 4 1 3 118 0 0 7 7 4 0 2 5 4 C om pare Model 2 with Model 1 1 0 168 n.s. 3 MMI (Trend Levels VI - IX) 415 5 1 5 7 1 7 1 0 146 0 7 0 2 8 0.4552 4 MMI (Levels VIII & IX vs. VI & VII) 415 5 1 5 5 3 5 1 0 328 0.5669 0 7 1 3 3 5 PGA (Trend - Quartiles) 420 516.927 1 1.213 0 2 7 0 8 0.1676 6 PGA a s C ontinuous Variable 420 5 1 5 8 3 5 1 2 3 0 4 0.129 0.1581 7 PGA (< 0.62 g a s reference category) 420 5 1 2 1 9 8 3 5.942 0 1 1 4 5 0 4 4 7 2 8* MMI (Dichotomous), Liquefiable Soil, & PGA Quartiles 414 5 0 0 4 5 5 9.399 0 0 9 4 2 0 4 5 4 6 9 MMI (Dichotomous), Liquefiable Soil, & PGA (Trend) 414 505.378 3 4.472 0 2 1 4 8 0.3353 10 MMI 414 509.623 1 0 2 2 7 0 6341 0.6007 11 Liquefaction 414 507.079 1 2.77 0.096 0 2 5 6 2 12 PGA Quartile 414 503 101 3 6 7 4 9 0 0 8 0 4 0 5 1 0 3 13 MMI (Dichotomous), Liquefiable Soil 414 506.865 2 2.985 0 2 2 4 8 0.4514 14 PGA Quartiles & MMI 414 501.011 4 8.838 0.0653 0.6459 15 PGA Quartiles & Liquefiable Soil 414 502.348 4 7.501 0.1117 0 3 5 9 4 C om pare Model 13 with Model 8 (PGA | MMI & Liquefaction) 414 3 6 4 1 4 n.s. C om pare Model 14 with Model 8 (Liquefaction | PG A & MMI) 414 1 0.561 n.s. C om pare Model 15 with Model 8 (MMI | PGA & Liquefaction) 414 1 1 898 n s . NO INTERACTION TERMS (Insufficient observations in polytom ous model) G eoloaic C haracteristics with DemoaraDhics 16 MMI 279 369 181 1 0 2 7 8 0 5982 0.4407 17 Liquefaction 279 369.137 1 0 322 0 5 7 0 2 0.3521 18 PGA Quartiles 279 365 149 3 4.31 0.2299 0.5686 18-t PGA Trend 279 369.452 1 0.007 0 9324 0 1659 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-7. (Continued) ' Chl-Sq. Prop. -2 " Log Chl-Sq. Chi-Sq. LRT Odds A ssum p p- M o d e l# M odel V ariab les n L ikelihood d f LRT p -v alu e v alu e 19 G ender 279 364 666 1 4 7 9 3 0 0 2 8 6 0 4 2 9 3 20 Ethnicity 279 3 6 9 3 7 3 2 0.086 0.9578 0.4501 21 A ge G roup 279 360.161 5 9.298 0.0978 0.0608 21 -t A ge G roup Trend 279 360 313 1 9.146 0.0025 0.0491 22 Hospital 279 3 5 8 2 9 4 2 11 165 0.0038 0.4241 2 3 " MMI, Liquefaction, PGA Quartiles, G ender, Ethnic, Age, Hospital 279 336 56 15 32.899 0 0 0 4 8 0.3803 24-t1 MMI, Liquefaction, PGA Trend, G ender, Ethnic, Age, Hospital 279 340.419 13 29.04 0.0065 0.274 24-t2 MMI, Liquefaction, PGA Quartiles, G ender, Ethnic, Age Trend, Hospital 279 337 318 11 32.142 0.0007 0.519 25 D eleted G ender from Model 23 279 337 804 14 31.655 0.0045 0.2979 26 D eleted Ethnicity from Model 23 279 337.539 13 31.92 0.0025 0.2662 27 Deleted A ge from Model 23 279 343.034 10 26.425 0.0032 0 8064 28 Deleted Soil Type from Model 23 279 336.746 14 32.713 0.0032 0.5411 29 Deleted MMI from Model 23 279 3 3 8 7 6 14 30.699 0.0061 0.2912 30 Deleted PGA from Model 23 279 347.315 12 2 2 1 4 5 0 0 3 5 9 0.2263 31 Deleted Hospital from Model 23 279 352.765 13 16.694 0 2 1 3 7 0.3876 C om pare Model 25 to 23 (G ender | Ethnic, Age, Soil, MMI, PGA, Hosp) 279 1 1 244 n.s C om pare Model 26 to 23 (Ethnic | G ender, Age, Soil, MMI, PGA, Hosp) 279 2 0.979 n.s. C om pare Model 27 to 23 (Age | Ethnic, G ender, Soil, MMI, PGA, Hosp) 279 5 6.474 n.s. C om pare Model 28 to 23 (Soil | Ethnic, Age, G ender, MMI, PGA, Hosp) 279 1 0.186 n.s. C om pare Model 29 to 23 (MMI | Ethnic, Age, Soil, G ender, PGA, Hosp) 279 1 2.2 n.s. C om pare Model 30 to 23 (PGA | Ethnic, Age, Soil, MMI, G ender, Hosp) 279 3 10.754 p < 0.025 C om pare Model 31 to 23 (H osp | Ethnic, Age, Soil, MMI, PGA, G ender) 279 2 16.205 p < 0.0005 *Core Geologic C haracteristics " C o re Geologic C haracteristics with D em ographics •u oo Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-8. Model Development: Dichotomous Logistic Regression. Geologic Characteristics and Demographics with Respect to Injury Severity. -2 * Log Chi-Sq. Chi-Sq. LRT p- Model #__________Model Variables_____________ n______Likelihood df______ LRT_______ value Soil Groups (Rock as reference 1 category) 419 416.195 2 3.696 0.1576 2 Liquefiable vs. A ll Other 419 416.558 1 3.333 0.0679 Compare Model 2 with Model 1 1 0.363 n.s. 3 M M I (Trend Levels V I - IX ) 415 420.722 1 0.097 0.755 4 M M I (Levels V III & IX vs. V I & V II) 415 420.453 1 0.367 0.5449 5 PGA (Trend - Quartiles) 420 422.193 1 0.903 0.3419 6 PGA (< 0.62 g as reference category) 420 417.582 3 5.514 0.1378 7 PGA as Continuous Variable 420 421.219 1 1.877 0.1707 8 * M M I (Dichotomous), Liquefiable Soil, 414 408.745 5 8.894 0.1134 & PGA Quartiles 9 M M I (Dichotomous), Liquefiable Soil, 414 413.266 3 4.373 0.2239 & PGA (Trend) 10 M M I 414 417.365 1 0.274 0.6007 1 1 Liquefaction 414 414.659 1 2.98 0.0843 12 PGA Quaitile 414 411.357 3 6.282 0.0987 13 PGA Trend 414 417.075 1 0.563 0.4529 14 M M I (Dichotomous), Liquefiable Soil 414 414.401 2 3.238 0.1981 15 PGA Quartiles & M M I 414 409.439 4 8.2 0.0845 16 PGA Quartiles & Liquefiable Soil 414 410.461 4 7.177 0.1266 00 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-8. (Continued) -2 * Log Chi-Sq. Chi-Sq. LRT p- Model # Model Variables n Likelihood df LRT value Compare Model 14 with Model 8 (PGA | M M I & Liquefaction) 414 3 5.656 n.s. Compare Model 15 with Model 8 (Liquefaction | PGA & M M I) 414 1 0.694 n.s. Compare Model 16 with Model 8 (M M I | PGA & Liquefaction) 414 1 1.717 n.s. NO INTERACTION TERMS (Insufficient observations in polytomous model) Geologic Characteristics with Demographics 17 M M I 295 310.269 1 0.86 0.3538 18 Liquefaction 295 310.535 1 0.594 0.441 19 PGA Quartiles 295 306.625 3 4.504 0.212 20 PGA Trend 295 311.068 1 0.061 0.8049 21 Gender 295 306.195 1 4.934 0.0263 22 Ethnicity 295 310.007 3 1.121 0.7719 23 Age Group 295 302.407 5 8.722 0.1207 24 Age Group Trend 295 303.238 1 7.891 0.005 25 Hospital 295 298.758 2 12.371 0.0021 26** M M I, Liquefaction, PGA Quartiles, Gender, Ethnic, Age, Hospital 295 280.369 16 30.76 0.0144 27 M M I, Liquefaction, PGA Trend, Gender, Ethnic, Age, Hospital 295 283.537 14 27.591 0.0161 28 M M I, Liquefaction, PGA Quartiles, Gender, Ethnic, Age Trend, Hospital 295 281.569 12 29.56 0.0033 ■u 00 u> Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-8. (Continued) M o d e l # M o d e l V a r ia b le s n -2 • L o g L ik e lih o o d d f C h i-S q . L R T C h i- S q . L R T p - v a lu e 2 9 D e le te d G e n d e r fro m M o d el 2 6 2 9 5 2 8 1 .6 1 2 15 2 9 .5 1 7 0 .0 1 3 8 3 0 D e le te d E th n icity fro m M o d el 2 6 2 9 5 2 8 1 .8 1 3 13 2 9 .3 1 6 0 .0 0 5 9 31 D e le te d A g e fro m M o d el 2 6 2 9 5 2 8 5 .6 5 2 11 2 5 .4 7 6 0 .0 0 7 8 3 2 D e le te d S o il T y p e fro m M o d el 2 6 2 9 5 2 8 0 .4 1 8 15 3 0 .7 1 1 0 .0 0 9 6 3 3 D e le te d MMI fro m M o d el 2 6 2 9 5 2 8 2 .3 6 7 15 2 8 .7 6 2 0 .0 1 7 3 3 4 D e le te d P G A fro m M odel 2 6 2 9 5 2 8 7 .9 0 4 13 2 3 .2 2 5 0 .0 3 9 1 3 5 D e le te d H o sp ita l fro m M o d el 2 6 2 9 5 2 9 3 .6 3 4 14 1 7 .4 9 5 0 .2 3 0 8 C o m p a re M o d el 2 9 to 2 6 (G e n d e r | 2 9 5 1 1 .2 4 3 n .s. E th n ic, A g e, S oil, MMI, P G A , H o sp ) C o m p a re M o d el 3 0 to 2 6 (E th n ic | 2 7 9 3 1 .4 4 4 n .s. G e n d e r, A g e , S oil, MMI, P G A , H o sp ) C o m p a re M o d el 31 to 2 6 (A g e | 2 7 9 5 5 .2 8 4 n .s. E th n ic, G e n d e r, S oil, MM I, P G A , C o m p a re M o d el 3 2 to 2 6 (S o il | 2 7 9 1 0 .0 4 9 n .s . E th n ic, A g e, G e n d e r, MMI, P G A , C o m p a re M o d el 3 3 to 2 6 (MMI | 2 7 9 1 1 .9 9 8 n .s. E th n ic, A g e , S o il, G e n d e r, P G A , C o m p a re M o d el 3 4 to 2 6 (P G A | 2 7 9 3 7 .5 3 5 n .s. E th n ic, A g e , S o il, MMI, G e n d e r, C o m p a re M o d el 3 5 to 2 6 (H o sp | 2 7 9 2 1 3 .2 6 5 p < 0 .0 0 5 E th n ic, A g e , S o il, MMI, P G A , G e n d e r) *C ore G e o lo g ic C h a ra c te ris tic s ••C o re G e o lo g ic C h a ra c te ris tic s w ith D e m o g ra p h ic s 0 0 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-9. Polvtomous Logistic Regression: A ssociated with Injury Severity Model Development. Demographics. Injury. Geologic. Structural Characteristics M o d el # M odel V a ria b le s n -2 * L o g L ik e lih o o d d f C h i-S q . LRT C h i-S q . P ro p . C h i-S q . LRT O d d s A s s u m p p -v a lu e p -v a lu e D em o a ra o h ic s (U nivariate) 1 G e n d e r 196 2 3 2 .4 3 5 1 4.1 5 0 .0 4 1 6 0 .2 4 3 8 2 Ethnicity 196 2 3 5 .7 2 2 2 0 .8 6 4 0 .6 4 9 3 0 .6 7 7 8 3 A ge G roup 196 228.561 5 8 .0 2 5 0 .1 5 4 9 0.2 0 0 4 3-t A ge G roup T rend 196 2 2 9 .7 9 1 6 .7 9 6 0.0091 0.0 8 2 9 Iniurv C h a ra c te ristic s (U nivariate) 4 B ody L ocation 196 2 3 4 .3 4 7 3 2 .2 3 9 0 .5 2 4 3 0 .5 8 6 8 5 M ech an ism of Injury (F alls vs. S tru ck /C u t/C au g h t by) 196 2 1 6 .7 6 4 1 19.821 0.0001 0.4051 S tructural C h a ra c te ristic s (U nivariate) 6 U se (m ulti-fam ily re s id e n c e vs. sin g le/duplex re s id e n c e ) 196 2 3 3 .3 9 6 1 3.19 0.0741 0 .7 1 6 6 7 Y ear Built G ro u p 196 2 3 3 .1 4 5 3 3.441 0 .3 2 8 5 0.6 1 9 2 7-t Y ear Built G roup T rend 196 2 3 4 .3 4 8 1 2 .2 3 8 0 .1 3 4 7 0 .2 3 3 4 G eo lo aic C h a ra c te ristic s (U nivariate) 8 MMI (d ich o to m o u s) 196 2 3 6 .5 3 5 1 0.051 0 .8 2 1 3 0 .2 6 6 9 Soil T ype (L iquefiable vs. all o th er) 196 23 6 .5 5 2 1 0.0 3 4 0 .8 5 3 9 0.9 3 5 4 10 P G A (d ich o to m o u s) 196 2 3 6 .2 9 5 1 0 .2 9 0.5 9 0 .5 1 5 8 10-t PG A (T rend) 196 23 6 .1 4 4 1 0 .4 4 2 0.5 0 6 4 0 .9 7 6 2 H osoital (U nivariate) 11 Level I T rau m a C e n te rs w .r.t. aU o th e rs 196 23 1 .8 4 7 2 4 .7 3 8 0 .0 9 3 6 0 .9 1 5 2 Full M odel 12* G en d er, Ethnicity, A ge, B ody L ocation, M ech an ism of 196 182.178 21 5 4 .4 0 8 0.0001 0.3601 Injury, S tru ctu re U se, Y ear Built, MMI, Soil T ype, PG A, 12-11 G en d er, Ethnicity, A ge G roup T rend, B ody L ocation, 196 188.333 17 4 8 .2 5 3 0.0001 0 .3 8 9 5 oo M ech an ism of Injury, S tru ctu re U se, Y ear Built, MMI, Soil T ype, PGA, H ospital Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-9. (Continued) M odel # M odel V a ria b le s 12-t2 G en d er, Ethnicity, A ge, B ody L ocation, M ech an ism of Injury, S tru ctu re U se, Y ear Built T rend, MMI, Soil T ype, 12-t3 G en d er, Ethnicity, A ge, B ody L ocation, M ech an ism of Injury, S tru ctu re U se, Y ear B u ilt, MMI, Soil T ype, PG A T rend, H ospital C o m p a re M odel 12-t1 to M odel 15 C o m p a re M odel 12-t2 to M odel 19 C o m p a re M odel 12-t3 to M odel 22 P artial M odels 13 D eleted G e n d e r from M odel 12 14 D eleted Ethnicity from M odel 12 15 D elete d A ge from M odel 12 16 D eleted B ody L ocation from M odel 12 17 D eleted M ech an ism from M odel 12 18 D elete d S tru ctu re U se from M odel 12 19 D eleted Y ear Built from M odel 12 2 0 D elete d MMI from M odel 12 21 D eleted Soil T ype from M odel 12 2 2 D eleted PG A from M odel 12 23 D eleted H ospital from M odel 12 00 a C o m p a re M odel 13 to 12 (G e n d e r | All o th er v ariab les) C o m p a re M odel 14 to 12 (Ethnicity j All o th er v ariab les) C o m p a re M odel 15 to 12 (A ge j All o th er v ariab les) C o m p a re M odel 16 to 12 (B ody L ocation | All o th er C o m p a re M odel 17 to 12 (M ech an ism | All o th e r v ariab les) C o m p a re M odel 18 to 12 (S tru ctu re j All o th er v ariab les) C o m p a re M odel 19 to 12 (Y ear Built | All o th er v ariab les) C o m p a re M odel 2 0 to 12 (MMI All o th e r v ariab les) C h i-S q . P ro p . -2 * L o g C h i-S q . C h i-S q . LRT O d d s A s s u m p n L ik e lih o o d d f LRT p -v a lu e p -v a lu e 196 1 82.969 19 5 3 .6 1 7 0.0001 0 .4 4 1 4 196 186.146 21 5 0 .4 4 0.0 0 0 3 0 .3 2 1 2 196 1 4 .7 4 7 p < 0.05 196 1 4 .4 2 5 p< 0.05 196 1 1 2.695 p<0.001 196 182.591 20 5 3 .9 9 5 0.0001 0 .3 3 1 8 196 183.534 19 5 3 .0 5 2 0.0001 0 .6 2 2 2 196 193.08 16 4 3 .5 0 6 0.0 0 0 2 0 .2 3 196 1 85.912 18 5 0 .6 7 3 0.0001 0.3 7 7 7 196 195.83 20 4 0 .7 5 6 0.0 0 4 0 .2 0 7 6 196 190.761 20 4 5 .8 2 5 0 .0 0 0 9 0 .5 7 5 4 196 187.393 18 4 9 .1 9 2 0.0001 0.6 1 3 4 196 185.116 20 5 1 .4 7 0.0001 0 .2 5 2 7 196 182.374 2 0 54.211 0.0001 0 .4 0 4 8 196 198.84 20 3 7 .7 4 5 0 .0 0 9 5 0 .2 6 8 7 196 190.231 19 4 6 .3 5 5 0 .0 0 0 4 0.3161 196 1 0 .4 1 3 n.s. 196 2 1.356 n.s. 196 5 1 0 .9 0 2 n.s. 196 3 3 .7 3 5 n.s. 196 1 13.652 p < 0 .0 0 0 5 1 % 1 8 .5 8 3 p < 0 .0 0 5 196 3 5 .2 1 6 n.s. 196 1 2 .9 3 8 n.s. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-9. (ContinuedI M odel # M odel V a ria b le s C o m p a re M odel 21 to 12 (Soil T ype | All o th er v ariab les) C o m p a re M odel 22 to 12 (PG A | All o th er v ariab les) C o m p a re M odel 23 to 12 (H ospital j All o th e r v ariab les) Interaction 24** G e n d e r, Ethnicity, A ge G roup, B ody L ocation, E xternal C a u se , S tru ctu re U se, Y ear of C onstruction, Interaction b e tw e e n S tru ctu re U se & Y ear o f C onstruction, MMI, L iquefiable Soil, PG A 24-11 G en d er, Ethnicity, A ge G roup T rend, B ody L ocation, E xternal C a u s e , S tru ctu re U se, Y ear o f C onstruction, Interaction b e tw e e n S tru ctu re U se & Y e a r of C onstruction, MMI, L iquefiable Soil, PG A 24-t2 G en d er, Ethnicity, A ge G roup, B ody L ocation, E xternal C a u se , S tru ctu re U se, Y ear Built T rend, Interaction b e tw e e n S tru ctu re U se & Y ear of C onstruction, MMI, L iquefiable Soil, C o m p a re M odel 35 to 12 (Interaction | All o th er v ariab les) P artial M odels with Interactio n s 2 5 D eleted G e n d e r from M odel 24 26 D eleted E thnicity from M odel 24 27 D elete d A ge from M odel 24 28 D eleted B ody L ocation from M odel 24 29 D eleted M ech an ism from M odel 24 30 D eleted S tru ctu re U se from M odel 24 31 D eleted Y ear Built from M odel 24 3 2 D eleted MMI from M odel 24 33 D eleted Soil T ype from M odel 24 34 D elete d PG A from M odel 24 C h i-S q . P ro p . -2 * L o g C h i-S q . C h i-S q . LRT O d d s A s s u m p n L ik e lih o o d d f LRT p -v a lu e p -v a lu e 196 1 0 .1 9 7 n.s. 196 1 16.663 p < 0 .0 0 0 5 196 1 8 .0 5 3 p < 0 .0 0 5 1 % 18 0 .9 5 2 24 5 5 .6 3 4 0 .0 0 0 3 0 .4 4 9 6 196 187.1 20 4 9 .4 8 5 0 .0 0 0 3 0 .5 1 4 8 196 1 8 1 .3 8 9 2 2 5 5 .1 9 7 0.0001 0 .4 4 3 6 196 2 1.226 n.s. 196 1 81.198 23 5 5 .3 8 8 0 .0 0 0 2 0 .3 9 5 5 196 1 8 2 .4 8 22 5 4 .1 0 6 0 .0 0 0 2 0 .7 8 6 4 196 1 92.727 19 4 3 .8 5 8 0.001 0 .3 4 7 8 196 184.88 21 5 1 .7 0 6 0 .0 0 0 2 0 .5 3 1 5 196 1 94.696 23 4 1 .6 8 9 0 .0 0 9 9 0 .3 1 4 5 196 1 8 2 .0 6 8 23 5 4 .5 1 8 0 .0 0 0 2 0 .4 0 8 2 196 181.911 21 5 4 .6 7 5 0.0001 0 .6 2 7 2 196 1 83.634 23 5 2 .9 5 2 0 .0 0 0 4 0 .3 2 9 196 1 81.045 23 5 5 .5 4 0 .0 0 0 2 0 .5 5 9 7 196 1 9 6 .5 1 5 23 4 0.071 0.0151 0 .3 6 8 7 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-9. (Continuedi C h i-S q . P ro p . -2 * L og C h i-S q . C h l-S q . LRT O d d s A s s u m p M odel # M odel V a ria b le s n L ik e lih o o d d f LRT p -v a lu e p -v a lu e 35 D eleted H ospital from M odel 24 196 189.171 2 2 4 7 .4 1 4 0 .0 0 1 3 0 .3 7 1 7 C o m p a re M odel 2 5 to 24 (G e n d e r | All o th e r variab les) 196 1 0 .2 4 6 n.s. C o m p a re M odel 2 6 to 24 (Ethnicity | All o th er v ariables) 196 2 1.528 n.s. C o m p a re M odel 27 to 2 4 (A ge | All o th e r variab les) 196 5 11.776 p < 0.05 C o m p a re M odel 2 6 to 2 4 (B ody L ocation | All o ther 196 3 3 .9 2 8 n.s. C o m p a re M odel 29 to 2 4 (M ech an ism | All o th er variab les) 196 1 13.945 p < 0 .0 0 0 5 C o m p a re M odel 3 0 to 2 4 (S tru ctu re U se | All o ther 196 1 1.116 n.s. C o m p a re M odel 31 to 2 4 (Y ear Built | All o th er variab les) 196 3 0 .9 5 9 n.s. C o m p a re M odel 32 to 24 (MMI j All o th e r v ariables) 196 1 2 .6 8 2 n.s. C o m p a re M odel 33 to 24 (Soil T ype j All o th e r variab les) 196 1 0 .0 9 4 n.s. C o m p a re M odel 34 to 2 4 (PG A | All o th e r v ariables) 196 1 15.563 p < 0 .0 0 0 5 C o m p a re M odel 35 to 2 4 (H ospital | All o th er v ariables) 196 2 8 .2 2 p < 0 .0 2 5 u oo oo Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-10. Model Development: Dichotomous Logistic Regression. Demographics. Injury. Structural A Geologic Characteristics with Respect to Moderate and Severe Inlurv. -2 * Log Chi-Sq. Likelihoo Chi-Sq. LRT p- Model # Model Variables n d df LRT value Demographics (Univariate) 1 Gender 2 Ethnicity 3 Age Group 4 Age Group Trend Iniurv Characteristics (Univariate) 5 Body Location 6 Mechanism o f ; , Structural Characteristics (Univariate) 7 Structure Use 8 Year Built Group 9 Year Built Group Trend Geologic Characteristics (Univariate) 10 M M I 1 1 Soil Type (Liquefiable vs. all other) 12 PGA (quartiles) 12-t PGA Trend Hospital (Univariate) 13 Level I Trauma Centers w.r.t. all others 230 237.387 1 3.463 0.0628 230 240.545 3 0.305 0.9592 230 234.439 5 6.411 0.2683 230 234.736 1 6.113 0.0134 230 235.79 3 5.059 0.1675 230 209.03 6 31.819 0.0001 230 239.642 1 1.208 0.2717 230 234.135 3 6.714 0.0816 230 235.545 1 5.305 0.0213 230 240.1 1 0.750 0.3866 230 240.85 1 0.000 1.0000 230 239.078 3 1.771 0.6212 230 240.389 1 0.461 0.4971 230 231.546 2 9.304 0.0095 -u 00 V O 8 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-10. (Continued) M odel # M odel Variables F ull M o d el, n o In te ra c tio n s 14 G e n d e r, E th n icity , A g e , B o d y L o catio n , M e c h a n is m o f Injury, S tru c tu re U se , Y e a r B uilt, MMI, S oil T y p e , P G A , H o sp ita l 15 G e n d e r, E th n icity , A g e G ro u p T re n d , B o d y L o catio n , M e c h a n is m o f Injury, S tru c tu re U se , Y e a r B uilt, MMI, S o il T y p e , P G A , H o sp ita l 1 6 G e n d e r, E th n icity , A g e, B o d y L o catio n , M e c h a n is m o f Injury, S tru c tu re U se , Y e a r Built T re n d , MMI, S o il T y p e , P G A , H o sp ita l 1 6-t G e n d e r, E th n icity , A g e, B o d y L o catio n , M e c h a n is m o f Injury, S tr u c tu re U se , Y e a r B uilt, MMI, S o il T y p e , P G A T re n d , H o sp ital P a rtia l M o d e ls 1 7 D e le te d G e n d e r fro m M o d el 14 18 D e le te d E th n icity fro m M o d el 14 1 9 D e le te d A g e fro m M o d el 14 2 0 D e le te d B o d y L o catio n fro m M o d el 14 21 D e le te d M e c h a n is m fro m M o d el 14 2 2 D e le te d S tr u c tu re U s e fro m M o d el 14 2 3 D e le te d Y e a r B uilt fro m M o d el 14 2 4 D e le te d MMI fro m M o d el 14 2 5 D e le te d S o il T y p e fro m M o d el 14 2 6 D e le te d P G A fro m M o d el 14 2 7 D e le te d H o sp ita l fro m M o d el 14 •2 * L og Chi-Sq. Likelihoo Chi-Sq. LRT p- n d d f LRT value 2 3 0 1 7 9 .0 7 2 2 9 6 1 .7 7 8 0 .0 0 0 4 2 3 0 1 8 2 .4 5 5 2 5 5 8 .3 9 5 0 .0 0 0 2 2 3 0 1 7 9 .4 1 6 2 7 6 1 .4 3 4 0 .0 0 0 2 2 3 0 1 8 2 .6 7 9 27 58 .1 7 1 0 .0 0 0 5 2 3 0 1 7 9 .1 0 4 2 8 6 1 .7 4 6 0 .0 0 0 2 2 3 0 1 7 9 .9 0 4 2 6 6 0 .9 4 6 0 .0 0 0 1 2 3 0 1 6 4 .6 0 1 24 5 6 .2 4 9 0 .0 0 0 2 2 3 0 1 8 6 .5 0 4 2 6 5 4 .3 4 6 0 .0 0 0 9 2 3 0 2 0 4 .7 8 6 2 3 3 6 .0 6 4 0 .0 4 0 6 2 3 0 1 8 3 .1 1 6 2 8 5 7 .7 3 4 0 .0 0 0 8 2 3 0 1 8 3 .3 9 3 2 6 5 7 .4 5 7 0 .0 0 0 4 2 3 0 1 8 2 .3 0 6 2 8 5 8 .5 4 4 0 .0 0 0 6 2 3 0 1 7 9 .8 9 3 2 8 6 0 .9 5 6 0 .0 0 0 3 2 3 0 1 8 9 .9 0 5 2 6 5 0 .9 4 5 0 .0 0 2 4 2 3 0 1 8 6 .5 0 8 2 7 5 4 .3 4 1 0 .0 0 1 4 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-10. (Continued) Model # Model Variables C o m p a re M o d el 17 to 14 ( G e n d e r | All o th e r v a ria b le s ) C o m p a r e M o d el 18 to 14 (E th n icity | All o th e r v a ria b le s ) C o m p a re M o d el 19 to 14 (A ge | All o th e r v a ria b le s ) C o m p a re M o d el 2 0 to 14 (B o d y L o c a tio n | All o th e r v a ria b le s ) C o m p a re M o d el 21 to 14 (M e c h a n ism | All o th e r v a ria b le s ) C o m p a re M o d el 2 2 to 14 (S tru c tu re | All o th e r v a ria b le s ) C o m p a re M o d el 15 to 19 (A g e G ro u p T r e n d ) C o m p a re M o d el 1 6 to 2 3 (Y e a r B uilt T re n d ) C o m p a re M o d el 17 to 2 6 (P G A T re n d ) C o m p a re M o d el 2 3 to 14 (Y e a r Built | All o th e r v a ria b le s ) C o m p a re M o d el 2 4 to 14 (MMI | All o th e r v a ria b le s ) C o m p a re M o d el 2 5 to 14 (S oil T y p e | All o th e r v a ria b le s ) C o m p a re M o d el 2 6 to 14 (P G A | All o th e r v a ria b le s ) C o m p a re M o d el 2 7 to 14 (H o sp ital | All o th e r * v a ria b le s ) -2 * Log Chi-Sq. Likelihoo Chi-Sq. LRT p- n d df LRT value 2 3 0 1 0 .0 3 2 n .s . 2 3 0 3 0 .8 3 2 n .s . 2 3 0 5 5 .5 2 9 n .s. 2 3 0 3 7 .4 3 2 n .s . 2 3 0 6 2 5 .7 1 4 p < 0 .0 0 0 5 2 3 0 1 4 .0 4 4 p < 0 .0 5 2 3 0 1 2 .1 4 6 n .s . 2 3 0 1 3 .9 7 7 p < 0 .0 5 2 3 0 1 7 .2 2 6 p < 0.01 2 3 0 3 4 .3 2 1 n .s. 2 3 0 1 3 .2 3 4 n .s . 2 3 0 1 0 .8 2 2 n .s . 2 3 0 3 1 0 .8 3 3 p < 0 .0 2 5 2 3 0 2 7 .4 3 7 n .s . Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-10. (Continued) M o d e l # M o d e l V a r ia b le s In te ra c tio n 28** G e n d e r, E th n icity , A g e G ro u p , B o d y L o catio n , E x te rn a l C a u s e , S tr u c tu re U s e , Y e a r o f C o n stru c tio n , In te ra c tio n b e tw e e n S tru c tu re U s e & Y e a r o f C o n stru c tio n , MMI, L iq u e fia b le S oil, P G A 2 9 G e n d e r, E thnicity, A g e G ro u p T re n d , B o d y L o c a tio n , E x te rn a l C a u s e , S tr u c tu re U se , Y e a r of C o n stru c tio n , In te ra c tio n b e tw e e n S tru c tu re U s e & Y e a r o f C o n stru c tio n , MM I, L iq u e fia b le S o il, P G A 3 0 G e n d e r, E th n icity , A g e G ro u p , B o d y L o catio n , E x te rn a l C a u s e , S tru c tu re U s e , Y e a r B uilt T re n d , In te ra c tio n b e tw e e n S tr u c tu re U s e & Y e a r of C o n stru c tio n , MMI, L iq u e fia b le S oil, P G A C o m p a re M o d el 2 8 to 14 (In te ra c tio n | All o th e r v a ria b le s ) P a rtia l M o d e ls w ith In te ra c tio n s 31 D e le te d G e n d e r fro m M o d el 2 8 3 2 D e le te d E th n icity fro m M o d el 2 8 3 3 D e le te d A g e fro m M o d el 2 8 3 4 D e le te d B o d y L o c a tio n fro m M o d el 2 8 3 5 D e le te d M e c h a n is m fro m M o d el 2 8 3 6 D e le te d S tru c tu re U s e fro m M o d el 2 8 3 7 D e le te d Y e a r Built fro m M o d el 2 8 3 8 D e le te d MMI fro m M o d el 2 8 -2 * L o g C h i- S q . L ik e lih o o C h i-S q . L R T p - n d d f L R T v a lu e 2 3 0 1 7 8 .2 5 5 2 3 0 1 8 1 .6 4 2 2 3 0 1 7 9 .2 2 4 2 3 0 2 3 0 1 7 8 .2 7 4 2 3 0 1 7 9 .1 4 2 2 3 0 1 8 3 .8 3 9 2 3 0 1 8 5 .4 2 4 2 3 0 2 0 3 .3 0 7 2 3 0 1 7 9 .5 3 6 2 3 0 1 8 0 .4 6 9 2 3 0 1 8 0 .8 0 6 6 2 .5 9 5 0 .0 0 1 0 5 9 .2 0 7 0 .0 0 0 5 6 1 .6 2 6 0 .0 0 0 6 0 .8 1 7 n .s. 6 2 .5 7 6 0 .0 0 0 7 6 1 .7 0 8 0 .0 0 0 4 57 .0 1 1 0 .0 0 0 6 5 5 .4 2 6 0 .0 0 2 2 3 7 .5 4 2 0 .0 6 6 7 6 1 .3 1 4 0 .0 0 0 9 6 0 .3 8 1 0 .0 0 0 6 6 0 .0 4 4 0 .0 0 1 3 3 2 2 8 3 0 3 31 2 9 2 7 2 9 2 6 31 2 9 31 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table A7-10. (Continued) M o d e l # M o d e l V a r ia b le s n -2 * L o g L ik e lih o o d d f C h i-S q . L R T C h i-S q . L R T p - v a lu e 3 9 D e le te d S o il T y p e fro m M o d el 2 8 2 3 0 1 7 9 .1 1 5 31 6 1 .7 3 5 0 .0 0 0 8 4 0 D e le te d P G A fro m M o d el 2 8 2 3 0 1 8 8 .0 7 7 2 9 5 2 .7 7 2 0 .0 0 4 5 41 D e le te d H o sp ita l fro m M o d el 2 8 2 3 0 1 8 5 .9 3 0 5 4 .9 5 0 0 .0 0 3 6 C o m p a re M o d el 31 to 2 8 (G e n d e r | All o th e r v a ria b le s ) 2 3 0 1 0 .0 1 9 n .s. C o m p a re M o d el 3 2 to 2 8 (E th n icity | All o th e r v a ria b le s ) 2 3 0 3 0 .8 8 7 n .s. C o m p a re M o d el 3 3 to 2 8 (A g e | All o th e r v a ria b le s ) 2 3 0 5 5 .5 8 4 n .s . C o m p a re M o d el 3 4 to 2 8 (B o d y L o c a tio n | All o th e r v a ria b le s ) 2 3 0 3 7 .1 6 9 n .s. C o m p a re M o d el 3 5 to 2 8 (M e c h a n ism | All o th e r v a ria b le s ) 2 3 0 6 2 5 .0 5 3 p < 0 .0 0 0 5 C o m p a re M o d el 3 6 to 2 8 (S tru c tu re U s e | All o th e r v a ria b le s ) 2 3 0 1 1.281 n .s. C o m p a re M o d el 3 7 to 2 8 (Y e a r B uilt | All o th e r v a ria b le s ) 2 3 0 3 2 .2 1 4 n .s . C o m p a re M o d el 3 8 to 2 6 (MMI | All o th e r v a ria b le s ) 2 3 0 1 2 .5 5 1 n .s. C o m p a re M o d el 3 9 to 2 8 (S oil T y p e | All o th e r v a ria b le s ) 2 3 0 1 0 .6 6 0 n .s. C o m p a re M o d el 4 0 to 2 8 (P G A | All o th e r v a ria b le s ) 2 3 0 3 9 .8 2 3 p < 0 .0 2 5 C o m p a re M o d el 41 to 2 8 (H o sp ita l | All o th e r v a ria b le s ) 2 3 0 2 7 .6 4 5 p < 0 .0 2 5
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar