University of Southern California Dissertations and Theses

An analysis of postacute treatment and outcome differences between Medicare fee-for-service and managed care

(USC Thesis Other)

An analysis of postacute treatment and outcome differences between Medicare fee-for-service and managed care

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS This manuscript has been reproduced from the microfihn master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter 6ce, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adva^ely affect reproduction. In the unlikely event that the author did not send UMI 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, b%inning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced fr>rm at the back of the book. 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. UMI A Bell & Howell Infonnation Conqaiy 300 North Zeeb Road, Ann A rbor MI 48106-1346 USA 313/761-4700 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. AN ANALYSIS OF POSTACUTE TREATMENT AND OUTCOME DIFFERENCES BETWEEN MEDICARE FEE-FOR-SERVICE AND MANAGED CARE By Joseph James Angelelli A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (Gerontology and Public Policy) August 1998 Copyright 1998 Joseph James Angelelli Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ÜMl Number: 9919008 Copyright 1998 by Angelelli, Joseph James A U rights reserved. UMI Microform 9919008 Copyright 1999, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeb Road Ann Arbor, MI 48103 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, vmtten by Joseph James Angelelli under the direction of h. 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 Or PHILOSOPHY lean of Graduate Studies / April 28, 1998 D ate.................................. DISSERTATION COMMITTEE . A c i . . é = . . . . . Chairperson ^ //l • ^ ** ##* »#* # • • # ***### # Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. u DEDICATION This dissertation is dedicated to my grandparents — to my grandfathers for lending me their good names and to my grandmothers for blessing me with their spirits. Guisseppie Angelelli Rose Angelelli Jim George Alberta “Birdie” George Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. tu ACKNOWLEDGMENTS This dissertation would not have been possible without the interest and support of many, many colleagues. Mends and family members. My thanks to Kate for her tireless devotion to the ideas and for knowing when to push and pull me in the right direction. The impetus for many of the dissertation’s research questions resulted from the combined wisdom of the other members of the health care policy research team (Fred DeJong, Bob Myrtle, Judy Yip, Lauren Craig, David Grazman, and Freddie Segal-Gidan) to whom I feel privileged to have worked with. Special thanks also to Merril for his wise mentoring during ray time at the Andrus Center and for setting a first-class example of how to conduct oneself as a researcher. Many o f the other graduate students shared their growing expertise with me and listened to me expound on dead-end ideas. Kaskie and Hyde deserve special recognition for their guidance and generosity. Finally, I would like to thank Cara for her love and patience with me during this ordeal and to thank my parents, the two most influential persons in my life, without whom none of this would be possible. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV TABLE OF CONTENTS Dedication ii Acknowledgements iii List of Tables viii List of Figures xii Abstract xiii Chapter I: Introduction 1 Conceptual Model 2 Characteristics at Admission 2 Sociodemogr^hics 3 Health status 3 Pre-treatment health care utilization 5 Characteristics of Treatment 5 Characteristics of Discharge 6 Chapter II: Background and Literature Review 7 Managed Care Financing and Postacute Care 7 Growth of Medicare and Managed Care 8 Public policy and managed care 11 Research on managed care financing and health care quality 12 Skilled Nursing Facilities in the Health Care System 15 Utilization of Skilled Nursing Facility Care 15 Risk factors 16 Heterogeneity of nursing home population 18 Organizational Role of Skilled Nursing Facilities 19 Public policy and skilled nursing facilities 20 Demand factors 20 Subacute care 21 Prospective payment system 23 Skilled nursing facility information Infirastructure 24 Applied Health Outcomes Research 26 Hip Fracture and Stroke 30 Hip Fracture 30 Mortality, morbidity, and risk factors for hip fiactures 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Treatment protocol and rehabilitation for hip fractures 32 Costs of hip fractures 33 Stroke 34 Mortality, morbidity, and risk factors for strokes 34 Treatment protocol and rehabilitation for strokes 36 Costs of strokes 37 Summary 37 Chapter HI: Research Design and Methods 39 Sample and Data 39 Managed Care Contracts 40 Procedures 41 Measurement of Variables 42 Characteristics at Admission 42 Sociodemographics 42 Health status 42 Pre-treatment health care utilization 47 Characteristics of Treatment 48 Characteristics at Discharge 48 Primary Research Hypotheses 49 Bivariate Relationships 49 Characteristics at admission 49 Hypothesis 1 50 Hypothesis 2 51 Hypothesis 3 51 Hypothesis 4 52 Hypothesis 5 52 Hypothesis 6 53 Hypothesis 7 54 Hypothesis 8 54 Characteristics of treatment 54 Hypothesis 9 55 Hypothesis 10 55 Hypothesis 11 55 Hypothesis 12 55 Hypothesis 13 55 Characteristics at discharge 55 Hypothesis 14 56 Hypothesis 15 56 Hypothesis 16 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V I Hypothesis 17 57 Hypothesis 18 57 Hypothesis 19 57 Multivariate Models 58 Equation 1 58 Ordinal dependent variable 58 Interaction effects 60 Equation 2 60 Hypothesis 20 63 Hypothesis 21 63 Sununary 64 Chapter IV: Results: Characteristics at Admission 65 Sociodemographics 65 Health Status 66 Primary diagnosis 66 Comoibidity 71 Functional Usability 72 Pre-treatment health care utilization 79 Admission source 80 Days between date of onset and admission into therapies 80 Admitted with orders for therapy 81 Therapy prognosis selection effect 83 Summary 87 Chapter V: Results: Characteristics of Treatment 88 Evaluation Units 88 Length of Stay 89 Therapy Units Per Day 91 Total Therapy Units 93 Summary 94 Chapter VI: Results: Characteristics at Discharge 96 Discharge Destination 96 Mortality 101 Re-Admittance into Therapy 101 Functional Disability 102 Bivariate Analysis 102 Multivariate Analysis 105 Summary 119 Chapter VU: Discussion and Conclusions 121 Summary of Study Results 121 Characteristics at Admission 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. vu Characteristics o f Treatment 124 Characteristics at Discharge 125 Implications for Public Policy and Applied Health Outcomes Research 128 References 130 Appendices Appendix A 139 Appendix B 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UST OF TABLES V lll Table 1. Description of Variables Table 2. Rehabilitation Outcome Measure Deficit Areas by Discipline Table 3. Charlson Comorbidity Index Table 4. Sociodemographic Characteristics of Medicare FFS and MCO Rehabilitation Patients Table 5. Primary Diagnoses Assigned by the SNF, by Payment Source Table 6. Primary Diagnoses Assigned by Physical Therapy Program, by Payment Source Table 7. Primary Diagnoses Assigned by Occupational Therapy Program, by Payment Source Table 8. Primary Diagnoses Assigned by Speech Therapy Program, by Payment Source Table 9. Comorbidity Differences by Payment Source Table 10. Frequency of Assessments Within Each ROM Deficit Area Table 11. ROM Scores for All Diagnosis Classification Groups at Admission, by Payment Source Table 12. ROM Scores for SNF-Assigned Stroke Diagnosis Classification Group, by Payment Source Table 13. ROM Scores for SNF-Assigned Fractures of the Hip and Pelvis Diagnosis Classification Group, by Payment Source Table 14. ROM Scores for the Therapy-Assigned Signs and Symptoms Diagnosis Classification Group, by Payment Source 44 45 47 65 67 68 69 70 73 73 76 77 78 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IX Table 15. Average Number of Days Between Admission into the Hospital and Admission into Physical, Occupational, and Speech Therapy Programs Table 16. Percentage of Therapy Patients Admitted into the SNF with a Physician’s Orders for Physical, Occupational, and/or Speech Therapy, by Payment Source Table 17. Differences Between Rehabilitation Patients and Non-Rehabilitation Patients Sociodemographic Characteristics, Total and by Payment Source Group Table 18. Differences Between Rehabilitation Patients and Non-Rehabihtation Patients in the Average Number of Secondary Diagnoses, Total and by Payment Source Group Table 19. SNF-Assigned Primary Diagnoses for FFS RehabiUtation Patients and FFS Non-Rehabilitation Patients 81 82 83 84 85 Table 20. SNF-Assigned Primary Diagnoses for MCO Rehabilitation Patients and MCO Non-Rehabilitation Patients 86 Table 21. Average Number of Evaluation Therapy Units in Physical, Occupational, and Speech Therapy, by Payment Source Table 22. Five Percent Trimmed Mean Lengths of Stay in the SNF and Physical, Occupational, and Speech Therapy Programs Table 23. SNF-Assigned Stroke Diagnosis: Five Percent Trimmed Mean Lengths of Stay in the SNF and Physical, Occupational, and Speech Therapy Programs Table 24. SNF-Assigned Fractures of the Hip and Pelvis Diagnosis: Five Percent Trimmed Mean Lengths of Stay in the SNF and Physical, Occupational, and Speech Therapy Programs 89 90 91 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 25. Average Number of Therapy Units Per Day in Physical, Occupational, and Speech Therapy, by Payment Source Table 26. SNF-Assigned Stroke Diagnosis: Average Number of Therapy Units Per Day in Physical, Occupational, and Speech Therapy, by Payment Source Table 27. SNF-Assigned Fractures of the Hip and Pelvis Diagnosis: Average Number of Therapy Units Per Day in Physical, Occupational, and Speech Therapy, by Payment Source Table 28. Average Total Therapy Units in Physical, Occupational, and Speech Therapy, by Payment Source Table 29. Discharge Destination from SNF of Surviving Patients, by Payment Source Table 30. Discharge Destination from SNF of Surviving Stroke and Fractures of the Hip and Pelvis Patients, by Payment Source Table 31. Discharge Destination from Physical Therapy, by Payment Source Table 32. Discharge Destination from Occupational Therapy, by Payment Source Table 33. Discharge Destination from Speech Therapy, by Payment Source Table 34. Percentages of All Patients, Stroke, and Fractured Hip or Femur Patients Who Died in the SNF, by Payment Source Table 35. Percentage Re-Admitted into Physical, Occupational, and Speech Therapy Programs, by Payment Source Table 36 ROM Scores for All Diagnosis Classification Groups at Discharge, by Payment Source 92 93 93 94 97 98 99 100 100 101 102 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XI Table 37. ROM Scores for Stroke Diagnosis Classification Group at Discharge, by Payment Source Table 38. ROM Scores for Fractures of the Hip and Pelvis Diagnosis Classification Group at Discharge, by Payment Source Table 39. Main Effects Ordered Logit Model; Predictors of Bed Mobility at Discharge fix>m Physical Therapy (N=325) Table 40. Interaction Effects Ordered Logit Model: Predictors of Bed Mobility at Discharge from Physical Therapy (N=325) Table 41. Predicted Probabilities of Outcomes Within the Sample for the Ordered Logit Model (N=325) Table 42. Predicted Probabilities by Payment Source and Initial Health Status for the Bed Mobility Ordered Logit Model (N=325) Table 43. Predicted Probabilities by Payment Source and Initial Health Status for the Bed Mobility Ordered Logit Model, Total Therapy Units =10 (N=325) Table 44. Main Effects Ordered Logit Model: Predictors of Transfers at Discharge from Physical Therapy (N=429) Table 45. Predicted Probabilities by Payment Source and Initial Health Status for the Transfers Ordered Logit Model (N=429) Table 46. Ordered Logit Model: Predictors of Gait-Level Surfaces at Discharge from Physical Therapy (N=407) Table 47. Predicted Probabilities by Payment Source and Initial Health Status for the Gait-Level Surfaces Ordered Logit Model (N=407) 104 105 108 110 111 113 114 115 116 118 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. XU LIST OF nOURES Figure I. Model of Treatment Effects on Health Outcomes in SNFs 4 Figure 2. Description of Operational Sample 40 Figure 3. Distribution of Deyo-Cbarlson Comborbidity Index 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X lll ABSTRACT The Southern California region is at the forefront of Medicare managed care penetration. The present study uses skilled nursing facility (SNF) therapy services data frrom a Southern California provider to compare fee-for-service (FFS) beneficiaries to managed care organization (MCO) emrollees in characteristics at admission, treatment characteristics and characteristics at discharge. Fee-for-service (N=240) and MCO (N=276) postacute patients had similar fimctional disability levels at admission as measured by the Rehabilitation Outcome Measure (ROM). Fee-for-service beneficiaries scored higher (worse) on the Deyo-Charlson Comorbidity Index. Managed care enrollees received significantly fewer units of physical, occupational and speech therapy and experienced significantly shorter lengths of stay in the SNF and in each rehabilitation program. Managed care enrollees were more likely to be discharged home than FFS beneficiaries who were more likely to be re-hospitalized from the SNF. The two payment source groups did not differ significantly in fimctional disability outcomes at discharge. Predicted probability analysis illustrated the selection effect caused by MCO utilization review. Those MCO patients who received therapy units were doing so because they continued to show improvement. The results have policy implications for understanding the growth of Medicare managed care. The dissertation also contributes to the literature as an example of how applied health outcomes research can play a role in studying the effects of SNF prospective payment implementation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I. INTRODUCTION Policies to increase the number of Medicare managed care enrollees and implement a prospective payment system (FFS) for skilled nursing facilities (SNFs) may affect the quahty of postacute health care for Medicare beneficiaries. While it is difficult to predict the eventual impact of such broad payment structure changes, there are methods by which the poUcies can be “forward-mapped” to better understand elements of their implementation (Weimer and Vining, 1992). The primary goal of this dissertation is to conduct a case study analysis of payment source effects on treatment processes and rehabilitation outcomes in a large Southern California SNF. The exploration o f important methodological and statistical issues will inform future efforts to determine the effects of Medicare managed care organizations (MCOs) and SNF prospective payment on the health outcomes of older adults. The objectives of the study are: 1) to determine the extent to which Medicare FFS beneficiaries and Medicare MCO enrollees receiving care in a SNF are comparable in terms of functional health status at admission and health care utilization history dtuing a specific episode of care. 2) to obtain empirical estimates of the relationship between payment source (Medicare FFS vs. Medicare MCO) and rehabilitation treatment in a SNF; 3) to obtain empirical estimates of the relationship between payment source and functional health outcomes following rehabilitation treatment in a SNF; 4) to determine the viability of predicted probability analysis as an empirical tool within a postacute performance measurement system designed to track SNF prospective payment. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. To accomplish these aims, this study utilizes physical, occupational, and speech therapy provider data. The use of postacute provider data is consistent with the mission of the Agency for Health Care Policy Research to promote relationships between health services researchers and policymakers, health care systems and individual clinicians. The results will add to the growing literature on the basic question of differences in postacute processes and quality of care. Ultimately, predicted probability analysis of postacute patient data could inform the development of risk-adjusted algorithms to be used in the negotiating process between payers and providers. A. Conceptual Model Modeling the factors influencing functional health outcomes within SNFs is an important first step in any attempt to establish meaningful quality indicators. Figure 1 portrays the conceptual model of treatment effects on health outcomes used for this study. The figure is a variation of a model developed by Kane and his colleagues (1997). Treatments are conceptualized as intervening in the relationship between initial and terminal health status. Health outcomes are also influenced by patient sociodemographic characteristics. Characteristics at Admission The modeling of characteristics at admission is multidimensional. The conceptual model includes the following elements: sociodemographic characteristics, health status characteristics including the patient’s primary diagnosis, comorbid conditions, the baseline functional disability score of the patient, and the patient’s health care utilization characteristics for the episode of care under review. Each area is discussed below. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Together the dimensions represent an attempt at risk adjustment - modeling factors beyond treatment that can affect outcomes. SociodemoCTaohics Sociodemographic characteristics in the model include patient age, sex, and marital status. Including these factors in the conceptual model is important because they define patient subgroups for which the treatment may be more or less effective. Increasing age has been associated with declines in functional status and recuperative skills (Pol and Thomas, 1992). Women have higher levels of morbidity for chronic conditions (Verbrugge, 1989). Finally, marital status is included as a proxy for the presence of a patient advocate. Each of these sociodemographic characteristics help to identify selection bias, can be treated as independent effects, and are useful when conducting subgroup analysis. Health status Comorbidities are conditions that are unrelated in etiology or causality to the principal diagnosis (Nitz, 1997). A patient’s comorbid conditions may increase or decrease the likelihood of a positive or negative outcome of treatment for the primary illness. The reasons to account for comorbid diseases when modeling health outcomes are that they help to identify selection bias, they can be treated as independent effects, and they are useful when conducting subgroup analysis. Comorbidity measures are also important because they help to establish the patient’s usual health status before treatment (Nitz, 1997). Accounting for comorbid conditions is especially important when studying the health outcomes of older adults, who are more likely as a group to present with secondary chronic conditions (Pol and Thomas, 1992). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure I. Model of Treatment Effects on Health Outcomes in SNFs Characteristics at Discharge Functional disability Destination from SNF Destination from TX Mortality Re-Admission in TX within 18 mos. Characteristics of Treatment Evaluation units Total units of therapy Length of stay in TX Units of therapy/day Length of SNF stay Financial Setting of Treatment Characteristics at Admission Sociodemographics Age Sex Marital status Health Status Primary diagnosis Comorbidity index Functional disability Pre-treatment health care utilization Admission source Days since onset Admitted w/ orders Therapy selection effect Functional disability scores also are modeled as characteristics at admission and can be considered as an operationalization of severity. Ability to function in multiple deficit areas (e.g., dressing, bathing, bed mobility) will form the basis of the functional health outcome measure described more fully in the description of variables section. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. modeling o f functional disability at admission is one aspect o f controlling for initial health status. Pre-treatment health care utilization. A patient’s health care utilization characteristics during the episode o f care are important factors to consider when modeling health outcomes. Admission source identifies where a patient was admitted from (i.e., hospital vs. home). The number of days since the onset of the condition can provide information about the duration o f the hospital stay for those patients admitted from a hospital and/or the length of time spent in the SNF before receiving therapy. Knowing whether a patient was admitted into the SNF with a physician’s orders for physical, occupational, and/or speech therapy provides information about patient prognosis at admission. Finally, comparing patients who receive therapy to those who do not is a method of modeling the therapy selection effect that may influence treatment and eventual health outcomes. Characteristics of Treatment The conceptual model accounts for the influence of treatment processes in several ways. These include the number of therapy evaluation units provided, the number of total therapy units provided, the length of stay in the rehabilitation program, the units of therapy provided per day, and the length of stay in the SNF. The treatment variables in the study are “procedures,” as opposed to medication or counseling treatments (Hebert, 1997). At issue is whether more units of therapy (evaluation, total and per day) and/or more time spent in rehabilitation or the facility are related to likelihood of recovery. Consistent with Hebert’s approach. Figure 1 models payment source as the financial setting o f the treatment. The financial setting of the treatment influences the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. treatment processes. Payment source is conceptualized as a patient-level variable. Reimbursement for an individual’s care in this study is either Medicare FFS or Medicare MCO. Characteristics at Discharge Patient performance on a clinician-administered measure o f functional disability is the primary health outcome variable in the study. Discharge destination is included as a characteristic to examine how payment source influences discharge from the therapy program as well as discharge from the SNF. Mortality is also an outcome of interest when comparing the two payment source groups. Finally, the two groups may differ in the rates at which they are re-admitted into therapy within 18 months. This study uses bivariate statistics to examine many of the relationships described in Figure 1. Multivariate analyses also are conducted to examine potential independent effects of payment source on functional health outcomes. Lastly, analyses are performed to examine possible interaction effects between initial health status and payment source. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. n. BACKGROUND AND LITERATURE REVIEW This chapter develops the rationale for the dissertation research. The dissertation contributes to an emerging body of literature on the topic of quality and postacute care. The published research covers a broad spectrum of disciplines, &om clinical trials appearing in the medical literature, to organizational behavior studies completed by academic health services researchers, to policy analysis reports done by private sector research firms. Viewed together, the published findings raise many compelling questions about how payment source is associated with postacute processes of care and health outcomes. The importance of the dissertation is best understood in the context of several substantive domains: a) the growth of managed care financing and its relationship to postacute care; b) the role of SNFs in the health care system; c) the emergence of applied health outcomes research and the functional assessment of performance following rehabilitation in SNFs; and d) the significance of stroke and hip-fracture as SNF admitting diagnoses; A review of the literature in each area establishes a foundation for the dissertation research questions and offers justification for the study. A. Managed Care Financing and Postacute Care Southern California is an ideal setting in which to analyze payment source effects on processes of care and functional disability outcomes using postacute provider data. Because the region stands at the forefiront of enrollment in Medicare managed care organizations, the experience of nursing homes in the region is likely to be repeated around the country as Medicare managed care enrollment increases. This study demonstrates how the growth of Medicare managed care will have implications for the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 financing of postacute care and how the health plan/provider dynamic of managed care financing will draw renewed attention to quality of care issues in postacute settings. Growth o f Medicare Managed Care The majority o f Americans under age 65 are enrolled in health plans that include some form of managed care (Halverson, Kaluzny, McLaughlin, and Mays, 1998). Managed care attempts to control or coordinate its members’ use of health services in order to contain costs, improve quality, or both. By centralizing decision-making within a primary care physician and using that provider as the conduit for specialty care, managed care holds the potential to better coordinate an older person’s overall health care delivery. However, a capitated fee system creates incentives that can be unfiiendly to individuals who require substantial resources firom the health care system (Stone and Katz, 1997). Older persons are more likely to have complex health conditions that require care firom many specialist physicians. Patients with major chronic illnesses are most at risk if managed care results in reduced access or quality of care, yet the chronically ill may benefit the most if managed care is able to correct current structural deficiencies in health care (Wagner, 1997). Medicare beneficiary enrollment in risk-contract health plans exceeded 10 percent of all beneficiaries for first time in 1996, with 4.1 million total risk plan enrollees (HCFA, 1996). Enrollment increased at an annual rate of over 40 percent between 1994 and 1996 and overall enrollment is projected to reach nearly 20 percent by the year 2000. In 1996, 241 risk plans participated in Medicare (HCFA, 1996). Seventy-two percent of Medicare beneficiaries had a Medicare risk plan in their geographical area. Fifteen Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. percent of Medicare beneficiaries who could enroll in risk plans did so in 1996 (HCFA, 1996). Medicare managed care enrollment in California as a proportion o f enrollment nationwide was 32 percent in 1996 (Lamphere, Neuman, Langwell, and Sherman, 1997). The six states with the largest numbers of enroUees (California, Florida, Pennsylvania, New Yoric, Texas and Arizona) accounted for 70 percent of total Medicare risk plan enrollment in 1996. Thirty-four percent of California’s Medicare beneficiaries are enrolled in Medicare risk plans (Lamphere et al., 1997). The Medicare risk program allows MCOs to assume responsibility for providing all Medicare-covered services to beneficiaries in return for a capitated payment. The payment per beneficiary is currently set at 95 percent of the adjusted average per capita cost (AAPCC). The county-specific rate is risk-adjusted for the age, disability, and institutional and welfare status of the beneficiary. A revision in the law now allows risk- contract HMOs to retain profits up to the level earned on their non-Medicare business. Profits that exceed this amount must be returned to enrollees either in out-of-pocket reductions or enhanced benefits (Meyer, Silow-Carroll, and Regenstein, 1996). By design, the risk program should lower costs to HCFA by five percent, relative to what HCFA would have paid in FFS reimbursements. However, a four-year evaluation of the risk program conducted by Mathematica Policy Research concluded that HCFA expenditures for the risk program were approximately 5.7 percent more than they would have been for FFS (Brown, Bergeron, Clement, Hill, and Retchin, 1993). More recent data indicate continued overpayment to risk plans. In 1994, the estimated average Medicare payment to risk plans was $3,963. Medicare would have spent an Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 estimated $3,656 per year, or about 7.7 percent less, if the same beneficiaries had remained in Medicare’s traditional FFS program (Riley, Tudor, Chiang, and Ingber, 1996). This increase in costs is primarily due to favorable selection of enrollees into Medicare risk plans, whereby beneficiaries with chronic problems are less likely than those in good health to change doctors or give up the freedom to use their primary care physicians, specialists, and hospitals of their choice. The 95 percent reimbursement rate has been shown to be financially lucrative to health plans in certain regions o f the country if they engage in the “creaming” of healthier beneficiaries (Morgan, Vimig, Devito, and Persily, 1997). In addition to favorable selection, managed care plans are able to maximize their reimbursement by reducing enrollee hospital use, primarily by substituting less expensive types of care, including ambulatory care, nursing home care, and home health visits. In the Mathematica study. Medicare managed care plans shortened the average hospital length of stay (LOS) by 1.5 days (16.8 percent of all days) relative to FFS, but the number of hospital admissions was not different between MCC and FFS (Brown et al., 1993). Medicare risk plans increased the likelihood that beneficiaries visited a physician at least once during the year from 84 percent to 89 percent, but shghtly reduced the likelihood of having one or more visits per month (on average) from 14 percent to 12.5 percent. Medicare managed care plans also increased the likelihood that beneficiaries had a physical exam by 6 percent, consistent with MCOs emphasis on and coverage of preventive care. Medicare risk plans had no effect on the proportion of individuals with Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 some home health care utilization, but they reduced the number of home health visits by 50 percent (Brown et al., 1993). Public policy and managed care. The Balance Budget Act of 1997 (Public Law 105-33) makes provisions to enroll more Medicare beneficiaries in managed care plans. The budget plan makes modest downward adjustments to the reimbursement mechanism that pays health plans 95 percent of the AAPCC. However, most of the $97 billion in budget savings over ten years that have been attributed to Medicare managed care are an indirect effect of savings from cuts to providers in the traditional FFS side of the program. Cuts to FFS providers will serve as a basis for establishing private plan payments (Moon and Gage, 1997). A system of Medicare reimbursement based on competitive bidding among health plans within health care markets was proposed as a demonstration project but failed to gamer political support. The Balanced Budget Act of 1997 also creates incentives for MCOs to enroll Medicare beneficiaries in rural areas where FFS Medicare remains the nearly universal form of reimbursement. In areas where the payments were very low (as low as $221 a month in 1997), private plans simply have not developed (Moon and Gage, 1997). The continued spread of Medicare managed care is taking place in a period of increased public questioning about the quality of health care that is “managed” by health plans. The extent of public concern is evident in the volume of state legislative activity aimed at regulating health plan behavior. In 1996, over 100 managed care bills were passed in 40 state legislatures (Gapay, 1997). The bills focused on issues such as mandatory coverage for emergency services and the prohibiting of provider “gag mles” (i.e., physicians withholding information about viable treatments). Other state-level Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 2 legislation has been passed to address managed care issues specific to older adults. Several states (e.g., Arizona, Minnesota, Florida) are implementing policies to restructure the way acute and long term care is delivered to older adults who qualify for both Medicare and Medicaid (dual eligibles). Capitated payments to managed care plans are a central feature o f dual eligible policies. Research on managed care financing and health care quality. State legislative activities have been based on provocative yet largely anecdotal information about the quality of care provided by managed care plans. Policies aimed at regulating managed care plan behavior need to be informed by sound empirical research. A synthesis o f 37 recently published peer-reviewed studies on managed care plan performance (for enrollees of all ages) describes quality-of-care evidence as split between an equal number of significantly better and worse MCO results, compared with FFS plans (Miller and Lufl, 1997). Evidence comparing hospital and resource use showed no clear pattern, whereas evidence on enrollee satisfaction varied by measure and enrollee type. Miller and Luft note that fears of managed care financing uniformly leading to worse quality of care are not supported by the research evidence. However, the reviewers caution that their review is based on research published using data from the period prior to 1992, a year in which a new round ofMCO-fueled cost cutting began in part because of the debate surrounding President Clinton’s health care reform proposals. The research on quality of care delivered to Medicare managed care enrollees with chronic conditions appears somewhat more conclusive (Miller and Luft, 1997). In general, most Medicare beneficiaries enrolled in a Medicare risk plan are satisfied with the availability and quality of care they receive (Physician Payment Review Commission, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 3 1996). However, beneficiaries who describe their health as fair or poor are less satisfied than those beneficiaries in good or excellent health. Recent research on differences in health care by payment source focused on global health status changes among firail older adults over a four-year period (Ware, Bayliss, Rogers, Kosinski, and Tarlov, 1996). Elderly and low income managed care enrollees were more than twice as likely to suffer declines in health status than those in FFS Medicare. Other research found that FSS home health care patients demonstrated better outcomes, implying that MCOs may be providing an insufScient number of visits (Schlenker, Shaughnessy, and Kittle, 1995). There remain significant gaps in the literature on quality of care in postacute settings like nursing homes and the role payment source may play in determining how such care is delivered (Weiner and Scaggs, 1995; VonKorff, Gruman, Schaefer, et al., 1997). Research demonstrating that MCOs pay for fewer services in postacute settings has not translated into observable dififerences due to payment source in mortality or hospital re-admissions (Retchin and Brown, 1991). Other research focusing on differences in postacute care due to payment source employs hospital discharge destination as an indicator of quality (Retchin, Brown, Yeh, Chu, and Moreno, 1997). Medicare managed care enrollees were found to be more likely than FFS patients to be discharged to nursing homes and less likely to be discharged to more care-intensive rehabilitation hospitals. Discharge destination fails to capture the wide variation of care provided in skilled nursing facilities and rehabilitation hospitals around the country and thus is a rough measure o f quality. Setting differences in quality of care appear to exist for certain Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 diagnoses but not others. In a recent study, hip fracture patients admitted to rehabilitation hospitals did not differ from patients admitted to nursing homes in such quality indicators as returning to the community and the number o f activities of daily living (ADLs) recovered to premorbid levels, but stroke patients in rehabilitation hospitals did achieve better outcomes than patients admitted to traditional nursing homes (Kramer, Steiner, Schlenker, Eilertsen, Hrincevich, Tropea, Ahmad, and Eckhoff, 1997). As Ware and his colleagues conclude, there is a need to examine whether differences due to payment source exist when more subtle measures of health outcomes are analyzed within diagnoses. Managed care financing will play an important role in how the continuum of care for older adults is transformed to meet the needs of the nation’s growing elderly population. Provisions to introduce more managed care financing into the Medicare program have focused primarily on the cost issue: Does managed care save the government money? The consensus among researchers and policy analysts is that managed care does not currently save the federal government money, but perhaps it will in the future as the risk of high cost beneficiaries becomes evenly distributed among health plans. As the cost issue dominates the debate among policymakers, the issue of quaUty is largely secondary in the federal govermnent’s efforts to restructure the way nursing home care is delivered and reimbursed. However, officials at HCFA charged with implementing the reimbursement changes recognize the opportunity to introduce quality indicators into a prospective payment system for SNFs and to make the system of postacute services “beneficiary-centered” (Vladeck, 1996). The reimbursement and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 quality improvement mechanisms of the PPS will further alter the fast-changing operating environments of SNFs. B. Skilled Nursing Facilities in the Health Care Svstem Skilled nursing facilities function as providers of postacute and chronic care. Prior to the restructuring of payment mechanisms to hospitals in the mid-1980s, the typical nursing home did not provide a high level of skilled nursing care nor did it provide high levels of ancillary services such as physical, occupational or speech therapy. Nursing homes continue to attend to the chronic care needs of a population that is mostly white, widowed, female, over the age of 80 and suffering from three or four chronic ailments. However, changes in the acute health care system as well as public policy initiatives such as the Omnibus Budget Reconciliation Act (OBRA) have altered the operating environment of today’s nursing home. Too many nursing homes continue to operate under the circumstances that led Bruce Vladeck in 1980 to author his seminal book. Unloving Care, which described the industry and the history of public policy failures that have shaped its development. But many nursing homes now provide quality care in an organized setting. To understand why this ongoing transformation is taking place, it is necessary to review the role of SNFs from an individual utilization perspective as well as from an organizational perspective. I Jtilization of Skilled Nursing Facilitv Care The use of skilled nursing facilities in the United States is costly. Research using data from the National Mortality FoUowback Survey estimates the expected discounted cost of nursing home care for persons turning 65 in 1990 to be $27,600 (Kemper, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 6 Spillman, et al., 1991). Variation around the average is high though, with the nine percent of persons expected to use at least five years of nursing home care accounting for 64 percent of the aggregate cost for the cohort and the 68 percent using less than three months of care accounting for only one percent of the aggregate cost. The variation in cost per person is better understood in the context o f risk factors for nursing home admission and discharge and the highly heterogeneous nature of the nursing home population. Risk factors. Over a lifetime, the risk of entering a nursing home and spending a long time there is substantial. To model the amount of time of his or her lifetime the average person spends in nursing homes, Kemper and Murtaugh (1991) analyzed the National Mortality FoUowback Survey and found that of those who died in 1986 at 25 years of age or older, 29 percent had at some time been residents of a nursing home. Almost half those who entered a nursing home spent a cumulative total of at least one year there. The probabiUty of nursing home use increases sharply with age at death: 17 percent for age 65 to 74, 36 percent for age 75 to 84, and 60 percent for those age 85 to 94. For persons who turned 65 in 1990, the authors project that 43 percent will enter a nursing home at some time before they die. In other work by the same researchers, data firom the 1982-1984 National Long- Term Care Survey is used to model risk of nursing home use in a similar way (Murtaugh, Kemper, et al, 1990). Thirty-seven percent o f persons in the nationaUy representative sample who died between 1982 and 1984 used a nursing home sometime after turning 65. The proportion was higher among females and whites in the North Central and Western Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 regions of the country. The authors conclude that over half the women and almost one- third of the men turning 65 in 1990 can be expected to use a nursing home sometime before they die. Research by Wolinsky, Callahan, Fitzgerald, and Johnson (1992;1993) uses data fix>m the Longitudinal Study on Aging to develop a more behavioral model of nursing home utilization. In their research, the risk of nursing home placement is greater for older adults, whites, those who lived alone, persons with telephones, those with fewer nonkin social supports, those who did not feel that they had much control over their future health, those with more household ADL or lower body limitations, and those who had been in a hospital prior to baseline. Of the 549 out o f 5,151 respondents placed in nursing homes over the two year period, the risk of dying there was greater for older adults, men, those who had not Uved in multigenerational households, persons who did not worry about their health, individuals with more upper body limitations and respondents having a history of valvular heart disease or cancer. The National Long Term Care Channeling Demonstration offered researchers an opportunity to quantify the predictors of nursing home admission and discharge among a population of 3,332 community-dwelling frail elderly persons determined to be at high risk for nursing home use. Using discrete-time hazard functions, Greene and Ondrich (1990) determined that the significant risk factors of admission to a nursing home from the community were ethnicity (Black and Hispanics were at lower risk), homeownership, advancing age, living alone, exhibiting higher cognitive and functional impairment levels, physician use, and living in an area with a larger nursing home supply. The researchers also modeled the probability of being discharged alive and found that Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 8 significant predictors were again ethnicity (Blacks were less likely), homeownership, being of younger age, and exhibiting better fimctional and cognitive capacities. The effect of family structure on nursing home utilization was specifically modeled in recent research by Freedman (1996). The researcher defined family structure in terms of available spouse, sons, daughters and siblings. Married older persons have about half the risk of nursing home admission of unmarried persons. Having at least one daughter or sibling reduces an older adult’s chance of admission to a nursing home by about one-fourth. Heterogeneitv of nursing home population. Simply modeling the risk-factors for nursing home admission and discharge may mask the heterogeneity of the nursing home population in terms of individual patient characteristics and service use (Manton, Liu, and Cornelius, 1985). Residents of nursing homes have multiple comorbidities and disabilities. Research using a multivariate analytic procedure called Grade of Measurement (GoM) analyzed 111 resident measurements in areas such as functional status, psychological well-being and cognitive performance to identify 11 profiles of health and functioning characteristics in a sample o f4,525 residents of 177 nursing homes (Manton, Cornelius, and Woodbury, 1995). The 11 profiles, representing clinically distinct groups, empirically demonstrate the heterogeneity of the nursing home population. Residents of nursing homes also vary greatly in terms of their functional and residential status transitions. Recent research on a sample of 9,541 long-stay (at least 90 days) nursing home residents shows that rather than simply declining or remaining stable in a uniform pattern, there exists a highly volatile transition process in nursing homes Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 (Gillen, Spore, Mor, and Freiberger, 1996). The researchers used a fully observed continuous-time Madcov chain model to estimate probability intensities of transition processes. Gillen and colleagues determined that the probability of change was highest for modest (one level) functional change and for such change to represent improvement rather than decline. They also found that over 25 percent o f long-stay residents exited in the second quarter and that 37 percent of them returned home. Variation of utilization and resident characteristics within facilities is mirrored by variation in utilization across facilities. Use of SNF care varies markedly according to geographical region. In 1987-88, patients in Minnesota used 32.5 times more SNF care than those in Mississippi, while nationally, rates of use o f SNF care varied by a ratio of more than three to one (Steiner and Neu, 1993). The data on utilization need to be seen in the context of the changing organizational role of nursing homes. The projections of future nursing home use are based on static models. The scenarios assume that the nursing home will continue to serve the same basic population of frail elderly persons. Evidence suggests the basic structure of niursing homes is changing to serve a higher acuity population and that this change has implications for the quality of care provided in nursing homes. Organizational Role of Skilled Nursing Facilities The role of nursing homes in the health care system continues to undergo a transformation in response to several factors. Nursing home regulations at the federal and state level have altered the operating environments of SNFs since the mid-1980s. In addition, demand factors related to changes in hospital reimbursement mechanisms and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 the growth of the managed care model of health care financing have caused SNFs to increase the range and levels o f services they provide. Public policv and skilled nursing facilities. The conditions governing Medicare coverage of SNF services have changed in a number of ways since the mid-1980s. In 1987, OBRA required several nursing home reforms. Facilities were required to implement stricter training regimens for nurse aides. OBRA also abolished the distinction between skilled nursing and intermediate care facilities. The implementing language for the new law took effect in October 1990. In April 1988, HCFA clarified its coverage guidelines regarding the Medicare Part A benefit so that fiscal intermediaries would vary less in their interpretations of what would be covered. By September 1988, SNF denial rates had decreased by some 50 percent (ProPAC, 1992). Another public policy initiative, the Medicare Catastrophic Coverage Act (MCCA) of 1988, resulted in dramatic increases in SNF use during 1989. Admissions rose by 57 percent and Medicare-covered SNF days increased by a factor of four (ProPAC, 1992). Repeal of MCCA in 1989 brought use rates down again, but the clarification of guidelines remained in effect and the operating environments of skilled nursing facilities were substantially altered by the public policy actions. Demand factors. The implementation of the Medicare Part A hospital prospective payment system, enacted as part of the Tax Equity and Fiscal Responsibilities Act (TEFRA) in 1982, created incentives for hospitals to discharge elderly patients earlier (Kane, Finch, Blewett, Chen, Bums, and Moskowitz, 1996). Before hospital PPS, hospitals were paid on an allowable cost basis. The hospital PPS sets a fiat fee according to patient diagnosis Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 and hospitals have an incentive to discharge patients as soon as clinically feasible. As a result, many patients who previously would have spent their entire episode of care in a hospital are now often discharged home or to SNFs, where in some cases integrated delivery systems owning both the hospital and SNF can collect the hospital DRG and the Medicare Part A nursing home benefit (GAO, 1996). The fraction of Medicare patients using both SNF and home health care increased in the years following hospital PPS (Neu and Harrison, 1988). The penetration of Medicare managed care also has altered the average SNF patient population. Health plans have sought less costly alternatives to long hospital stays for their enrollees (Fishman, Von Korff, Lozano, and Hecht, 1997). In areas where Medicare managed care penetration is high, many plans are witnessing the “aging in place” of their enrollees as well as an increasingly frail population of new enrollees. As a result, many health plans and risk-bearing medical groups are contracting with physician- led case management companies to monitor the acute and postacute utilization of their enrollees. SNFs have responded to these demand factors by creating “subacute” units or programs of care designed for higher acuity, more therapy-intensive patients (Lewin- Vm , 1995). Subacute care. Subacute care has gained recognition as a distinct level of care provided to patients who are sufficiently stable to no longer require acute care but whose condition is too complex for treatment in a conventional nursing facility. Subacute care is rendered immediately after, or instead of, acute hospitalization. Common subacute services include physical rehabilitation, ventilator care, intravenous therapy, wound management. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 cardiac treatment, and dialysis (Lewin-VHI, 1995). Most subacute units have increased nursing staff with acute care experience, a medical director, and sophisticated medical technology to monitor patient status. Subacute care is provided for a limited (several days to several months) time, or until a condition is stabilized or a predetermined treatment course is completed. In addition to increased demand caused by hospital DRGs and managed care cost- containment, current government reimbursement policies contain incentives for providers to develop a subacute program of care. HCFA reimburses SNFs on the basis of their reasonable costs up to a specified cost limit. The limit is imposed on the allowable reimbursement of routine costs for general nursing, room and board, and administrative overhead. The limit creates incentives to carefully monitor routine costs per day and to be less concerned with length of stay. Ancillary services such as physical and occupational therapy are subject only to medical necessity criteria. As a result, between 1992 and 1995 reported ancillary costs per day in SNFs increased 67 percent, from $75 per day to $125 per day, while reported routine costs per day increased only 20 percent, from $123 to $148 (GAO, 1996). Providers who offer more ancillary services can cite high ancillary service use to justify an exception or exemption from routine service payment limits. Medicare expenditures in 1994 were $5.8 billion for 1.25 million SNF cases (HCFA, 1996). The average Medicare payment for SNF care in 1994 was $4,672 per stay (HCFA, 1996). The incentive structure of the current reimbursement system will change as the SNF prospective payment system is implemented. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 Prospective payment system. The SNF prospective payment system will pay per diem rates covering all facility cost types and payments will be adjusted for differences in patient case mix. HCFA’s ongoing SNF PPS demonstration project has tested the use of per diem rates adjusted for resource need using the Resource Utilization Group (RUG-EII) patient classification system. The RUG system originated with nine groups created with four patient-specific variables (dress, ambulate, feed, intake/output of fluids monitored) to differentiate levels of resource consumption as measured by staff time (Fries and Cooney, 1985). The latest version, RUG-HI, is a more refined method for classifying SNF residents according to health characteristics and the amount and type of resources needed (Fries, Schneider, Foley, Gavazzi, Burke, and Cornelius, 1994). The RUG-HI classification system for nursing home residents was developed based on a sample of 7,658 residents in seven states. The system has 44 distinct groups and achieves 55.5 percent variance explanation of total (nursing and therapy) per diem cost. RUG-III can and will be used in nursing home management, staffing level determination, and quality assurance. The impact of PPS will depend on the degree to which payment rates recognize characteristics of a given facility as well as on the extent to which providers are prepared to implement the system. An unexamined issue of SNF prospective payment implementation is whether SNFs have the information technology and empirical expertise necessary to adapt to the new financing system. Quality of care could improve if providers have the proper tools to track per diem case-mix reimbursement. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 Skilled nxirsing facilitv infonnation infrastructure. Providers of postacute health care are continuing to improve their information systems to track alternative treatments, resource use, and clinical outcomes (Enthoven and Vorhaus, 1997). However, the progress is slow in comparison to information technology development in other industries because of the diffused nature o f the health care sector and the historical unwillingness of government policymakers to regulate data collection and information systems (Starr, 1997). Using provider-collected data to track prospective payment and monitor functional health outcomes will soon be possible as implementing regulations for the electronic transmission of the national Minimum Data Set (MBS) for nursing homes are finalized under the Nursing Home Reform Act as part of OBRA 1987. Some states already require providers to submit patient-level data to a centralized location where the data are used to determine how Medicaid payments should vary according to resident needs and nursing resource use (i.e., case-mix). Linking utilization data with resident assessment data will become easier as more nursing homes modernize their information systems to comply with the final OBRA regulations. This practice will become universal with the implementation of the SNF PPS. Changes in skilled nursing facility reimbursement mechanisms have been a significant impetus for information infirastructure building among providers. The dominant methods of reimbursement for nursing homes have been FFS Medicare, Medicaid, and private pay. These reimbursement mechanisms provide few financial incentives to invest in capital improvements such as information systems and, as such, there is little need to track costs associated with individual-level functional improvement. With the emergence of managed care as a significant payer in the postacute market. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 providers require the ability to link costs to services and establish with some degree of confidence their ability to provide services within a level of care per diem reimbursement system. In addition to operational reasons for improving their information systems, providers are acting also to establish their credibility with payers and independent accreditation agencies. As of March 2, 1998 the Joint Commission on Accreditation of Healthcare Organizations (JCAHO) mandated that providers implement performance measurement systems to maintain accreditation (JCAHO, 1997). Established in 1951, the Joint Commission evaluates and accredits more than 16,000 health care organizations in the United States. The accreditation entity has developed the ORYX system to integrate the use of outcomes and other performance measures into the accreditation process. ORYX will require health care organizations to begin collecting and using performance data and to transmit the data to the Joint Commission. The system is designed to allow organizations to compare their performance with that of peer organizations using the same measures within the same performance measurement system. Provider compliance requirements are modest in the first year of participation, but the Joint Commission intends to gradually increase the breadth and depth of quality indicators required of accredited skilled nursing facilities. Prospective payment reimbursement and JCAHO accreditation are forcing SNF providers to develop information systems capable of tracking health outcomes. However, the information systems alone will not improve performance. Determining the impact of PPS and managed care financing on quality of care will require empirical tools used as part of an applied health outcomes research agenda. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 6 C. Applied Health Outcomes Research Attempts among postacute providers to establish outcomes research agendas originate from their current and expected relationships with payers and accreditation entities. Yet as demand grows for more rigorous outcomes research in SNFs, the potential of empirical models to measure the processes and outcomes of care in the health care industry is only beginning to be realized (Starr, 1997; Kane, 1997). To understand why this lag exists, it is first necessary to define what is meant by health outcomes research. The Institute of Medicine (loM) defines health outcomes research as “research that studies the end results of the structure and processes of health care on the health and well-being o f patients and populations” (1996, p.l). The loM suggests two types of outcome measures deserve special attention: (1) health-related quality of life, including functional health status, and (2) satisfaction with care. Others argue that health outcomes research should attempt to measure three separate components of quality: efficacy of treatment, appropriateness of treatment, and the interpersonal, supportive and psychological aspects of health care (Caper, 1988). Measuring efficacy requires one to answer the question: Does the therapy accomplish its’ goal? Randomized clinical trials and epidemiological research are useful approaches for measuring this aspect of quality. Appropriateness refers to the measurement of the costs and benefits of a particular therapeutic course. Cost-benefit analyses encompass both human (clinical risk and benefits) and economic considerations. The third component concerns measuring satisfaction with care and empowering the patient with aggregated information about provider and health plan performance. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 Outcomes analysis is increasingly important because it has the potential to provide valuable insight into all three aspects o f health care quality. Consideration o f efficacy, appropriateness and patient satisfaction are becoming central features of the way care is deUvered in America’s market-based health care system. Kane (1997) cites several reasons why applied health outcomes research has become so prominent an issue: Outcomes analysis can inform market decisions made by individual patients or employer benefits managers acting on behalf o f patients; it can also provide accountability to payers and regulatory entities; and it can improve the knowledge base of medicine which ultimately serves the interests of the entire health care sector. Unfortunately, the absence of a conunon “tool-box” of vahdated measurement instruments to track health care quality continues to be the major obstacle to implementing outcome measures in postacute settings (loM, 1996, Wagner, 1997). There are many ways to measure efficacy, appropriateness and patient satisfaction and the lack o f standardization makes it difficult for providers to create a uniform applied outcomes research agenda. The lack of a conunon validated measurement system is especially apparent in the postacute setting. A typical postacute provider has contracts to provide services to enrollees o f many different health plans (Lewin-VHI, 1995). Each health plan may request a unique set of quality indicators as part o f their contract with the provider. In addition, a SNF may generate different quality indicators finom its multiple relationships with outside rehabilitation companies providing physical, occupational, and/or speech therapy to its’ residents. Finally, the scope of quality indicators are likely to differ depending on the audience. Therapists may be concerned with individual health status Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 8 within functional domains (efficacy), health plans may be concerned with aggregate utilization patterns (appropriateness), and consumers may be concerned with overall provider and health plan performance (satisfaction). Together these factors make it difficult for postacute providers to generate outcomes research that can be used to improve processes and outcomes of care. The MDS is the core functional assessment instrument in the larger Resident Assessment Instrument (RAI), which OBRA 1987 mandates that nursing homes use as a standardized, comprehensive assessment system. Recent evaluation research on the impact of the RAJ shows that its implementation has significantly improved process quality in areas such as the accuracy of information in residents’ medical records and the comprehensiveness of care plans (Hawes, Mor, Phillips, Fries, Morris, Steele-Friedlob, Greene, and Nennstiel, 1997). The MDS also can be used to investigate improvements in outcome quality. Recent research documents substantial reductions in hospitalizations of nursing home residents with cognitive impairment and those with stable ADL following the implementation of the RAI (Mor, Intrator, et al., 1997). Other research on the RAI’s impact shows that the system’s implementation may have improved the quality of care of nursing home residents by reducing overall rates of decline in important areas of resident function, such as ADLs, social engagement and cognitive function (Phillips, Morris, et al., 1997). The use of MDS data for research is the subject o f an ongoing debate in the literature (Hawes, Phillips, Mor, Fries, and Morris, 1992; Teresi and Holmes, 1992). The debate over using MDS data for research revolves around several issues. The utility of MDS data collected for clinical or financial considerations as opposed to data collection Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 driven by the goals of the primary research questions is a concern among those skeptical of using MDS data for outcomes analysis (Teresi and Holmes, 1992). Another concern is the lack of standardization of data collection across and within facilities. Data fields may be completed by staff with varying levels of training and assessment expertise. Accurate descriptions of patient status may also be compromised if reimbursement is tied to patient functional improvement. In addition, facilities are required to conduct full assessments within fourteen days of admission into the SNF. The brief stays of many MCO rehabilitation patients means that few MDS assessments exist for that patient population. These concerns notwithstanding, MDS data remain an untapped resource for improving quality of care in nursing homes (Kane, 1995) and continue to be used to inform public policy debates (Arling, Rhyther, Collins, and Zimmerman, 1991; Philips, Hawes, and Fries, 1993; Riter and Fries, 1992). The use of MDS data to improve quality of care can take place within a larger restructuring of SNF job roles, work group and organizational design levels (Zinn, Braimon, and Mor, 1995). Applied outcomes research can be used within the organizational context of a nursing home as a key feature of continuous quality improvement efforts. The MDS provides a means for caregivers to identify patterns and associations of symptoms, behaviors, treatment and outcomes (Zinn et al., 1995). The electronic submission of the MDS will be a critical step in providers’ efforts to initiate applied outcomes research agendas. The MDS was designed to provide functional assessment data on nursing home residents with a wide range of chronic health problems. In addition to other focus areas (e.g., psychosocial wellbeing, falls, pressure ulcers, etc.) the MDS is designed to measure Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 the amount o f assistance a nursing home resident receives with basic activities o f daily living, such as bathing, dressing, feeding, and transfers (Hawes et al., 1992). However, the MDS was not designed to assess short-term postacute rehabilitation admissions. The growth of SNF-based subacute units has created a need for functional assessment measures designed specifically for patients with rehabilitation needs. Accurate measurement of rehabilitation outcomes requires attention to the unique characteristics of the specific diagnoses treated in nursing homes. While the range of admitting diagnoses is varied, hip fiacture and stroke are the two most common diagnoses treated in SNFs (Steiner and Neu, 1993). A discussion of the diagnoses informs the diagnoses-specific analysis of treatment and outcome differences by payment source and highlights the significance o f the dissertation’s focus. D. Hin Fracture and Stroke This section reviews the mortality and morbidity statistics for hip fi-acture and stroke diagnoses as well as the risk factors associated with each condition. The acute treatment protocol and typical course of rehabilitation therapy also are described for each diagnosis. Finally, the costs of hip fracture and stroke in terms of health care and lost productivity are reviewed. Hip Fracture Mortalitv. morbiditv. and risk factors for hin fi-actures. The hip is a ball-and-socket joint; the ball is the spherical head of the thighbone, or femur, and the socket is a region on the side of the hipbone known as the acetabulum. In older people, even relatively minor injuries or falls may cause a firacture of the neck of the femur, the small portion that lies just below the head o f the femur. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 The incidence of hip fractures in the United States is approxiamately 80 per 100,000, increasing with age, doubling every five to seven years over the age of 60 (Zuckerman and Schlon, 1990). Ninety-five percent of hip firactures occur in people 50 years and older and it is estimated that 250,000 hip fi-actures occur each year among the 65 or older age group (Riggs and Melton, 1986). The highest risk of mortality after hip fractures occurs in the first 4 to 6 months. The mortality rate among elderly patients one year after a hip firacture ranges firom 14 to 36 percent (White, Fisher, and Laurin, 1987). After the first year, the mortality rate is similar to that among age- and sex-matched persons without hip firactures (Zuckerman, 1996). In a prospective study assessing the independent effect of hip firacture on mortality and hospitalization, Wolinsky and his colleagues used the Longitudinal Study on Aging (LSOA) to analyze the 368 (out of 7,527) persons who suffered a hip firacture between 1984 and 1991. The researchers found that hip fi-acture was significantly related to mortality and that the effect was concentrated in the first 6 months postfiracture (Wolinsky, Fitzgerald, and Stump, 1997). Hip fracture is a major cause of chronic disabihty among older adults. At the age of 80 years, every fifth woman has suffered a hip firacture (Kannus, Parkkari, Sievanen, Heinonen, Vuori, and Jarvinen, 1996). The factors associated with the recovery of walking ability include young age, male sex, absence of dementia, absence of postoperative confused state, and use of a walking aid before the fracture (Lyons, 1997). Many determinants of outcome are independent of the level of care given and are dependent on pre-firacture status. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 Risk factors for hip fracture include increased age and increased incidence of falls, which account for nearly 90 percent of hip fractures in older adults (Means, 1990), increased osteoporosis and being female and/or Caucasian (Farmer, White, and Brody, 1984), dementia (Buchner and Larson, 1987), moderately impaired or poor vision (Felson, Anderson, Hannan, and Milton, 1989), and a low calcium diet (Holbrook, Barret-Connor, and Wingard, 1988). Women over age 65 have a l-in-5 chance of having a hip fracture during their lifetime (Farmer et al., 1984). Treatment protocol and rehabilitation for hip fractures. Hip joints severely damaged by falls or arthritis are now often removed and replaced with an artificial hip joint that allows nearly normal activity in most persons. The type of reconstructive surgery after a hip firacture depends on the characteristics of the fi-acture and the patient. Patients with symptomatic rheumatoid arthritis or osteoarthritis of the hip generally require total hip replacement after a fracture. Postoperative care includes early mobilization to guard against thromboembolic complications. Most patients receive anticoagulation medications such as aspirin or warfarin. Complications from surgical treatment of hip fi-actures are rare. Infection occurs in less than five percent of patients (Zuckerman, 1996). In a study concerning the timing of surgical and rehabilitative care for hip firacture patients, researchers analyzed the records of 1880 elderly Medicare recipients admitted to 284 acute care hospitals in five states during 1981 and 1982 or 1985 and 1986. Earlier surgical repair (within the first two days of hospitalization) and more than five physical therapy or occupational therapy sessions were associated with earlier ambulation, shorter lengths of stay, higher liklihood of returning to the community, and better 6-month Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 survival (Hoenig, Rubenstein, Sloane, Homer, and Kahn, 1997). However, others argue that early surgery should only be performed in premorbidly fit patients, whereas surgery should be delayed if correctable comorbidities are present (Lyons, 1997). All patients who survive a hip fracture will benefit from some type of rehabilitative services. The primary goal of a rehabilitation program after a hip fi-acture is to reduce disability and maximize fimction to allow the person to return to their prior activity level. It is imperative that the services begin early (typically by the first post operative day), be interdisciplinary in nature and continue until the person reaches her maximal fimctional level. The majority of patients after hip fracture will return to their premorbid level of basic fimctions within 4-6 weeks after the firacture (Cifu, 1993). Studies of hip firacture outcomes highlight several factors associated with successful rehabilitation. Research examining the determinants of potential outcomes in 2,624 elderly newly admitted to SNFs following hip fractures found those residents with social support more likely to return home (Kiel, Eichom, Intrator, Silliman, and Mor, 1994). Physically and cognitively impaired residents and those taking narcotics, cardiac medications, or antidepressants were most likely to die. Younger men, those with social support, those with functional dependency, and those who were free of disorientation were more likely to be re-hospitalized. Costs of hip fractures. Hip fractures account for about S O percent of all inpatient hospital days for fracture care in the United States. The annual costs for all medical and rehabilitation of hip firactures have been estimated at 7 billion dollars (Holbrook, Grazier, Kelsey, and Stauffer, 1984). With the aging of U.S. population, utilization of acute and rehabilitation Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 services associated with hip fractures will increase (Friedman and Elixhauser, 1993). Forty percent o f hip fracture patients 65 and older are discharged from hospitals to long term care facilities (Friedman and Elixhauser, 1995). In a prospective study of the economic cost of hip fractures in a sample of 759 community-dwelling older adults, researchers measured the resource use for direct medical care, formal nonmedical care, and informal care in the year preceding and year after a hip fracture. Incremental costs in the year after the fracture, compared to the costs in the year before the fracture, ranged between $16,322 and $18,727 (Brainsky, Glick, Lydick, Epstein, Fox, Hawkes, Kashner, Zimmerman, and Magaziner, 1997). In the study cited above using the LSOA data to analyze hip fiacture mortality, hip fracture significantly increased the liklihood of rehospitalization and increased the number of hospital days by 21.3 percent, increasing total charges by 16.3 percent (Woiinsky, Fitzgerald, et al., 1997). Stroke Mortality, morbidity and risk factors for strokes. Stroke is a cerebroyascular injury that occurs when blood flow to the brain is interrupted by a clogged or burst artery. The interruption deprives the brain of blood and oxygen, and causes brain cells to die. Some of the signs of major stroke are facial weakness, inability to talk, loss of bladder control, difficulty in breathing and swallowing, and paralysis or weakness, particularly on one side of the body. Stroke is also called cerebral apoplexy and cerebrovascular accident (CVA). Stroke is the third leading cause of death in the United States, following coronary heart disease and cancer. Stroke kills roughly 150,000 o f the 500,000 Americans who Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 suffer a new or recurrent stroke each year (American Heart Association, 1992). In 1950 the death rate from stroke was 88.8 percent; in 1993, it was 26.9 percent (National Center for Health Statistics, 1996). The reasons for decreasing stroke mortality could be related to either decreasing incidence and/or improved prognosis, though researchers examining observed decreases in stroke mortality for the period 1968-1988 conclude that improved survival explains more of the decrease in mortality than does a decline in incidence (Modan and Wagener, 1992). Stroke is the number one cause o f adult disability. Approxiamately two-thirds o f all strokes occur in people age 65 and over (National Center for Health Statistics, 1996). Over three million Americans are currently living with some degree of impairment due to a stroke (Greshman, Duncan, Stason, et al., 1995). Stroke is a disease that is frequently associated with many other chronic diseases. Three-quarters of all recent deaths where stroke was recorded as the underlying cause of death had at least one associated condition and almost 120,000 deaths per year occur with stroke when another condition is the underlying cause o f death (Baum and Manton, 1987). Risk factors for stroke are based on heredity, natural processes, and/or as a result of a person’s lifestyle. Modifiable risk factors include high blood pressure, heart disease, cigarette smoking, transient ischemic attacks (brief episode of stroke symptoms), high cholesterol, sedentary lifestyle or physical inactivity, obesity and excessive alcohol intake. The chance of having a stroke more than doubles for each decade of life after age 55. Men have about a 19 percent greater chance o f stroke than women, but women die more often as a result of stroke (American Heart Association, 1992). Afiican-Americans have a much higher risk of death and disability fix)m a stroke than whites (94 percent Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 higher for males and 75.8 percent higher for females), in part because the Afiican- American population has a greater incidence of high blood pressure (American Heart Association, 1992). Treatment protocol and rehabilitation for strokes. RehabiUtation from stroke requires specialized help from neurologists, physical therapists, speech therapists, and other medical persons. Most progress in rehabilitation from stroke is made in the first six months. Passive stretching exercises and thermal applications are used to regain motor control of limbs, which become rigidly flexed after stroke has occurred. A patient may recover enough to do pulley and bicycle exercises for the arms and legs and, through speech therapy, may regain the language abilities often lost following a stroke. Rehabilitation after a stroke varies greatly firam patient to patient. Some people do not need rehabilitation after a stroke because the stroke was mild or they have fully recovered. Others may be too disabled to participate. People who are too disabled at first may recover enough to enter into a rehabilitation program. In hospital or SNF-based rehabilitation programs, the patient may spend several hours a day in physical, occupational, or speech therapy. There are several factors associated with improvement or decline in quality of life following a nonfatal stroke. A recent study used the LSOA to analyze the factors predicting stroke survival and changes in disability and activity Umitations over a two- year period. Individuals who were less than 80 years old were found to be likely to show fewer activity limitations and less disabilities (Dighe, Aparasu, and Rappaport, 1997). The study also found that increased use of health services as represented by physician. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 hospital and nursing home visits resulted in lower disability and activity limitations among the elderly sample. However, increased use of health services is not without costs. Cost of strokes. Estimates are that stroke costs the United States $30 billion annually in health care and lost productivity (American Heart Association, 1992). Stroke is the main reason for hospitalization in people aged 65 years and above (National Center for Health Statistics, 1992). Estimates are that stroke accounts for half of all patients hospitalized for acute neurological disease (American Heart Association, 1992). In addition to societal health care costs, the costs related to informal caregiving are substantial. The role of a person’s family and support network in rehabilitation is significant. Summary. The spread of managed care and the changing operating environment of nursing homes are creating a postacute model of care that requires sound empirical research. The focus on building information networks, collecting data fields on individual patients and reporting aggregate quality indicators to payers and accreditation agencies is necessary but not sufficient to produce meaningful indicators of health care quality. The National Committee for Quality Assurance (NCQA), a leading accreditation entity, has released guidelines that set parameters for establishing meaningful health outcome improvement (NCQA, 1997). The process relies on the presence of (1) an appropriate study design, (2) an analysis of the study results, (3) an appropriate intervention, and (4) a remeasurement that demonstrates the improvement. The guidelines represent a concerted attempt to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 impose structure on applied outcomes research efforts that currently vary widely in scope and usefulness. Kane and his colleagues at the University o f Minnesota suggest that sound conceptual models and the use of appropriate statistical techniques to test such models remain the exception rather than the norm. The viability o f outcomes research depends on how well health services researchers are able to translate often complex findings resulting fi-om the use of sophisticated statistics and research designs into useful information that can be acted upon by providers. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 ffl. RESEARCH DESIGN AND METHODS This section first describes the sample and data used in the research. The measurement of variables section then clarifies how the concepts in the model are operationalized. Finally, the research hypotheses section elaborates on how specific research questions are tested in statistical models. B. Sample and Data The primary data set consists o f 523 rehabilitation patients age 65 and over who were admitted, treated and discharged between May 1, 1996 and December 30,1997 in a 180 bed JCAHO-accredited SNF owned and operated by Country Villa Health Services. Country Villa is a mid-size (12 facility, 1,0004- bed) Southern Califomia-based postacute provider. The contract to provide rehabilitation services to the Country Villa SNF patients was held by South Coast Rehabilitation Services (SCRS) during this period. SCRS and its parent company. Regency Health Services, were acquired by Sun Healthcare Group in December 1997. Thus, SCRS now operates as Sundance Rehabilitation Services. The 523 first-time admissions represent an exhaustive sample o f rehabilitation patients age 65 and over treated in the Country Villa facility during the 20 month period. Figure 2 illustrates how the sample of 523 patients was derived firom a larger sample of 1,367 total MCO and FFS admissions during the period. Of the 1,367 total admission between May 1996 and December 1997, 1,029 were first-time admissions. Seven hundred and thirty-eight of the 1,029 first-time admissions were age 65 or older and were no longer residing in the SNF. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 Three hundred and ninety-five persons age 65 and over were enroUees in a Medicare managed care plan. Of those, 276 received therapy services. Three hundred and forty-three patients received care within the traditional Medicare FFS system. Of those, 247 received therapy services. Each patient within the two payment source categories received some combination of physical, occupational, or speech therapy. Figure 2. Description of Operational Sample Age 65 still in SNF N=20 Under Age 65 N=271 NoTX N=119 NoTX N=96 Speech Tx N=79 Speech Tx N=69 Physical Tx N-240 Occup. Tx N=101 Occup. Tx N=163 Physical Tx N =2G G Age G 5 and over N=738 Managed Care N=395 Fee-for-Service N=343 MCO Recieved TX N=27G FFS Received TX N=247 First-time Admissions N=1029 Total MCO & FFS Admissions May 199G to December 1997 N=13G7 Managed Care Contracts. The 180 bed SNF had managed care contracts with between six and nine managed care organizations over the 20 month study period. However, for the first nine months of the study, nearly 95 percent of the managed care patients over age 65 treated in the SNF Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 were enrolled in just two of the health plans. Then on February 1, 1997, the larger o f the two major health plans acquired the second plan. Thus, for the last 11 months of the study period, 95 percent of the managed care patients were members of the same risk- assuming health plan. The contract between the large managed care plan and the SNF is in Appendix A. The contract details the services included in the per diem level of care arrangement. The level of care required by the patient was determined by the managed care physicians and nurse practitioners, but was not explicitly listed in the patient’s computerized medical record. All per diem rates included basic services such as prescription and non prescription medications not to exceed $40 per day, nursing care, oxygen services and supplies, nutrition services, laboratory services. X-ray services, and patient and/or family education. The agreement established three levels of care that ranged from no therapy services provided (level 1) to 60 to 180 minutes per day of physical, occupational, and/or speech therapy. A custodial care per diem was also in the contract, defined as non-acute institutional care which does not meet guidelines defined by Medicare for skilled nursing care. Procedures. The treatment and outcome data were stored on a separate information system apart from most of the sociodemographic data. The treatment and outcome data were created from computer-generated discharge reports. The reports contained information about each patient’s rehabilitation history at the facility. Patient-level information was then merged using a unique patient ID shared by both information systems. A dataset containing sociodemographic, primary diagnosis Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 and comorbidity data for those individuals not receiving therapy services was created for subsequent analysis of potential selection effects. B. Measmement of Variables The variables representing the constructs of the study are operationalized in ways that have subsequent importance to the direction and scope of the analysis. Table I lists the variables within each domain and describes how each is operationalized. Characteristics at Admission Sociodemogranhics. Age, sex, and marital status are all sociodemographic variables included in Country Villa’s information system. Age is measured as continuous variable in years computed from the date of birth freld. Sex is coded as female =1 and male =0. Marital status is coded as married (1) versus other (0). Health status. The measure of functional disability in the proposed study is the Rehabilitation Outcome Measure (ROM). The ROM was designed by South Coast Rehabilitation Services (SCRS) to be used by physical, occupational, and speech-language therapists to rate postacute patients regarding their level of function for discipline-specific deficit areas on a seven point rating scale (SCDRS, 1994). The ROM resembles the Functional Independence Measure (FIM), the most widely used assessment tool in acute rehabilitation settings. The FIM is used by a variety of health care providers ranging from nursing to rehabilitation clinicians as an indicator of a patient’s overall functional performance. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 The ROM differs from the FIM in several ways. ROM scores are assessed in a discipline-specifrc manner whereas the FIM is designed to be an interdisciplinary assessment tool. The discipline-specifrc versions of the ROM cover defrcit areas similar to those in the FIM (e.g., bathing, toileting, dressing) but there are also unique ROM defrcit areas (e.g., gait-uneven terrain, joint mobihty, sitting balance). Empirical research has demonstrated that the FIM should be treated as an ordinal rather than interval measure (Granger, Hamilton, Linacre, Heineman, and Wright, 1993; Linacre, Heineman, Wright, Granger, and Hamilton, 1994). The ROM employs a scaling technique nearly identical to the FIM (both are based on seven point scales ranging from dependent to independent). The importance of recognizing the potential ordinal nature of the ROM is discussed later in the data analysis section. Table 2 details the defrcit areas o f the ROM. Licensed therapists undergo a four-hour Certifrcation Training process before administering the ROM. Part of the certifrcation process includes successful completion of a written Skills Assessment designed to test the clinician’s knowledge of the ROM and its correct administration. However, the data collected regarding individual profrciency on the Skills Assessment is not useful in terms of demonstrating inter-rater and test-retest reliability of the ROM instrument. Sundance is currently participating in a collaborative study with the Uniform Data Set (UDS) Data Management Service, the organization responsible of the creation and psychometric verifrcation of the FIM. The goal of the collaborative relationship is to establish the comparability of the discipline-specifrc ROM to the FIM. The forthcoming Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 results of the Sundance/UDS study will provide reliability data to be included in manuscripts submitted for peer-reviewed publication. Table 1. Description of Variables. Variable Name Description CHARACTERISTICS AT ADMISSION Health Status Rehabilitation Outcome Measure (ROM) 7 p L scale (0=dep.; 6=indep.) Primary Diagnosis Diagnosis ClassiHcatioa G roup Comoibidity Deyo-Charlson Comoibidity Index Sum of secondary diagnoses Sociodemographic Age Measured in years Sex (l=female; 0=maie) Marital status (l=married; O = other) Pre-Treatment Health Care Utilization Admission source (l=hospital; O =other) Days since onset of condition Measured in days Admitted into SNF with therapy orders (l=yes; 0=no) CHARACTERISTICS OF TREATMENT Evaluation units Number of therapy evaluation units Total rehabilitation units Total num ber of therapy units Length of rehabilitation episode Measured in days Rehabilitation units per day Number of therapy units per day Length of SNF stay Measured in days Payment source (1=MC0; 0=FFS) CHARACTERISTICS AT DISCHARGE Rehabilitation Outcome Measure (ROM) 7 point scale Discharge destination from facility (l=hospital; 2=home; 3=other SNF; 4=Board and care/assisted living) Discharge destination from therapy (l=hospitaU 2=home; 3=SNF improved function; 4=SNF same level; 5=restorative 6=Board and care/assisted living). Mortality ( l=deceased; O = not deceased) Re-Admission into therapy w/in 20 mos. (l=yes; 0=no) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 ROM assessments are completed by therapists when the patient is admitted into and discharged from a program o f therapy. Therapists are given discretion in determ ining which defrcit areas to assess at admission. The number of defrcit areas assessed per patient ranges from one to three in each therapy program. Thus a patient receiving physical, occupational, and speech therapy could have assessments made in as few as three and as many as nine defrcit areas. Table 2. Rehabilitation Outcome Measure Defrcit Areas by Discipline Speech Therapy Occupational Therapy Physical Therapy Auditory Comprehension Bathing Bed Mobility Auditory Discrimination Bed Mobility Gait-Level Surfaces Cognitive-Linguistics Community Activities Gait-Uneven Terrain Expressive Language Dressing Joint Mobility Reading Comprehension Emergency Response Physical Restraints Speech Production Feeding Sitting Balance Swallowing Funct. Communication Skin Integrity Voice Grooming/Hygiene Stair Climbing Written Expression Home Management Standing Balance Joint Mobility Transfers Meal Preparation/Cleanup Medication Routine Physical Restraints Toilet Hygiene Wheelchair Mobility Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 SNF patients receiving rehabilitation are classified according to primary diagnosis using International Classification of Diagnoses - Version 9 (ICD-9) codes. Patients are assigned an ICD-9 code by the facility which may or may not precisely correspond to the ICD-9 code used by the rehabilitation therapist to describe the patient. Both ICD-9 codes are used in the analysis to describe the diagnosis distribution of each payment source patient population. The distribution of ICD-9 codes are summarized using a Diagnosis Classification Group (DCG) system developed by SCRS for postacute care to mirror the Diagnosis Related Group (DRG) system used by acute hospitals. Appendix B lists the ICD-9/DCG conversion tables. The Charslon Comorbidity Index (CCI) is a method of weighting secondary conditions and summing them to create an index reflecting comorbidity (Charlson, Pompei, Ales, and MacKenszie 1987). The CCI was originally created for use with medical chart abstracts to predict mortality for hospitalized patients. The measure incorporates information about the severity of 19 common conditions. Table 3 contains the weights for the diseases calculated from the adjusted relative risk of mortality associated each disease. The weights are summed and the resulting number represents the CCI score. The clinical disease definitions have been “mapped” to specific ICD-9 codes for use with specific diseases o f interest (Deyo, Cherkin, and Ciol, 1992). The resulting Deyo-Charlson Comorbidity Index (DCCI) is used as a measure of comorbidity in this study. From zero to five comorbid conditions can be listed in Country Villa’s information system as ICD-9 codes. Research has showed that a simple of sum of secondary diagnoses Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 (without weights) performs well in a model predicting length o f stay and mortality and does not suffer &om problems associated with miscoding on claims data (Melfi, Holleman, Arthur, and Katz, 1995). Thus, the sum of secondary diagnoses also is used a measure of comorbidity. Table 3. Charlson Comorbidity Index. Assigned Weights for Diseases Conditions 1 Myocardial infarct Congestive heart failxire Peripheral vascular disease Cerebrovascular disease Dementia Chronic pulmonary disease Connective tissue disease Ulcer disease Mild liver disease Diabetes 2 Hemiplegia Moderate or severe renal disease Diabetes with end organ damage Any tumor Leukemia Lymphoma 3 Moderate or severe liver damage 6 Metastatic solid tumor AIDS Pre-treatment health care utilization. Patients undergo a variety of treatments for their conditions prior to admission into the SNF. FFS patients must have had a prior hospitalization within 3 days of admission into the SNF to have their stay reimbursed under Medicare. There is no prior Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 hospitalization requirement for MCO patients. Thus, the two payment source groups may differ in their source of admission (hospital vs. home). A “date of onset” data field in the therapy services database represents the date of the patient’s most recent hospital admission. For those individuals admitted into the SNF fiom a hospital, this date was used to calculate the number of days that had passed between onset of the condition and admission into a SNF program of therapy. Patients receiving rehabilitation therapy in a SNF can either be admitted into the SNF with a physician’s orders for therapy or they may be considered for therapy at a later point in their episode of care. FFS beneficiaries and MCO enroUees may differ on this characteristic. Characteristics of Treatment Rehabilitation treatment information is contained in the therapy services database. Included are length of stay in the SNF, length of rehabilitation service stay, average evaluation units, average units of service per day, total units of therapy received, and payment source. Therapists provide one to eight units of therapy per day depending on patient fimctional status. One unit represents 15 minutes of therapy. The length of rehabilitation stay is measured in days and is often but not always different firom the length of a patient’s skilled nursing facility stay. Payment source is a dichotomous variable representing either Medicare FFS or Medicare MCO. Other payment source categories (e.g., Medicaid, private pay) are not considered in this study. Characteristics at Discharge A patient’s ROM scores at discharge in each of the deficit areas are used as outcome variables. Discharge destination is measured in two ways. Patients can be Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 discharged horn the SNF to a hospital, to home, to a board and care facility or to another skilled nursing facility. The therapy services database also contains information on the patient’s discharge destination from the physical, occupational, and/or speech therapy program. Re-admittance into the therapy program within 20 months is measured as a yes/no variable. Finally, mortality is measured by recoding the “expired” category in the discharge destination data field to its own variable (l=yes, 0=no). D. Primarv Research Hvpotheses This section describes the specific research hypotheses and the statistical analyses used to test them. The section is divided into two subsections; 1) bivariate relationships; and 2) multivariate models. Bivariate Relationships Characteristics at admission. The issue of selection is a persistent problem when researching potential differences in treatments and outcomes between FFS and MCO patients in a SNF. Any observed differences in treatments and outcomes may be due to unobserved “inputs” elsewhere in the course of a patient’s health care episode (e.g., the decision to enroll in a Medicare MCO, utilization review decisions during the acute care stay, etc.). Comparing the treatment and outcomes of FFS and MCO patients in a SNF may be equivalent to comparing apples and oranges. To determine if this is the case, FFS and MCO patients can be compared across several sociodemographic and pre-treatment health care utilization characteristics. Bivariate statistics are used to examine the relationship between payment source and sociodemographic characteristics, health status at admission, and pre-treatment Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 50 health care utilization characteristics. Cross tabulations with chi-square tests are used to describe the relationship between categorical variables such as sex and marital status. When a variable is distributed normally and measured using an interval scale (e.g., age), the Student’s T-test is used to test for a significant difference in the means of the two payment source groups. When the variable is measured ordinally (e.g., ROM score) or the distribution is highly skewed (e.g., the Deyo-Charlson Comorbidity Index), the non- parametric Mann-Whitney U rank test is used to compare ± e distribution of the variable across the two payment source groups. The two payment source groups may differ in their sociodemographic profiles because of the selection bias that has been observed in Medicare MCO enrollment (Brown et al., 1993). Medicare MCO enroUees are on average younger and thus more likely to be married and male. However, the two payment source groups are not being observed as members of the general population of older adults — they are patients receiving rehabilitation therapy in a SNF. Nevertheless, the MCO-enrollment selection effect may manifest itself in this study’s patient populations. Hypothesis 1 Medicare FFS beneficiaries receiving rehabilitation therapy in a SNF are more likely than Medicare MCO enroUees to be older, female, and single. Payment source may influence health status at admission in ways that make predicting the direction of baseline difference problematic. Some research suggests that managed care plans serve healthier people (Counte and Glandon, 1996). Thus one might expect managed care patients admitted into a SNF to have fewer comorbidities and perform better on tests of functional disability than Medicare FFS patients. However, the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 discharge destination research cited in the literature review chapter suggests that FFS patients are more likely to be admitted into rehabilitation hospitals (Retchin et al., 1997). Since rehabilitation hospitals are on average designed to serve a population with more rehabilitation needs, it follows that a selection effect may be occurring such that FFS patients admitted to nursing homes will perform better on tests of fimctional disability than managed care patients. Also, if the utilization review mechanism of managed care functions to reduce the length of costly acute care stays. Medicare MCO patients may enter the SNF earlier in their episode of care and thus in poorer health than Medicare FFS patients. With these seemingly contradictory factors to consider, the direction of baseline differences in health status by payment source is an empirical question amenable to bivariate statistical analysis. Hypothesis 2 There is a statistically significant difference in health status at admission to a SNF between Medicare FFS beneficiaries and Medicare MCO enroUees. In addition to clarifying baseline differences in health status, bivariate statistical analysis can be used to examine pre-admission treatment differences between FFS and MCO patients. The rules governing Medicare FFS reimbursement for SNF care are such that FFS beneficiaries must have had a hospital stay within 3 days prior to a SNF admission. No such rules apply to Medicare MCO enroUees — patients can be admitted into a SNF directly from home. Hypothesis 3 Medicare FFS beneficiaries are significantly more likely than Medicare MCO enroUees to be admitted into a SNF fi-om a hospital than fiom home. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 The mechanism of managed care utilization review may have an effect on the length of hospitalization and/or the time spent in a SNF prior to receiving rehabilitation services. Medicare MCO enroUees may be discharged from the hospitals at an earlier point in their episode of care when compared to Medicare FFS beneficiaries. MCO enroUees may also receive SNF-based therapy services at an earUer point in their SNF episode of care when compared to FFS beneficiaries. Hypothesis 4 Medicare MCO enroUees admitted from a hospital spend significantly less time in the period between onset o f the episode and admission into a SNF rehabilitation program than do Medicare FFS beneficiaries. Medicare MCO patients who receive therapy services in a SNF may be more likely than Medicare FFS patients to be admitted into the SNF with a physician’s orders for physical, occupational, and/or speech therapy. SNF reimbursement for Medicare FFS patients encourages the deUvery of therapy services, whereas the utilization review mechanism of Medicare managed care plans operates to limit the expansion of new services once patient prognosis is determined. Thus, it is likely that MCO patients who have been admitted into a therapy program are there because their course of treatment was agreed upon by the provider and the MCO at admission into the SNF. Hypothesis 5 Medicare MCO rehabilitation patients are more likely than Medicare FFS rehabilitation patients to be admitted into the SNF with a physician’ s orders for therapy. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 Utilization review and the use of established clinical pathways by managed care organizations influence the decision to enter a patient into a program of rehabilitation therapy. All patients who enter therapy are selected for one o f two reasons; a) their prognosis for recovery is observably better than patients receiving custodial or non-rehabilitation skilled care; or b) their functional status is significantly worse than other SNF patients, thus the need for rehabilitation therapy. Thus patients who receive therapy services may be different than those who do not, regardless of payment source. Hypothesis 6 Patients receiving rehabilitation therapy have a significantly different level o f comorbidity on average than SNF patients who do not receive rehabilitation therapy. The extent that one reason for admitting patients outweighs another in a given population can be tested by comparing the within-group comorbidity levels of patients who receive therapy to those who do not. If reason (a) prevails, then one might expect the differences in comorbidity level to be more pronounced in the Medicare MCO group compared to the Medicare FFS group because of the closer oversight of the at-risk health plan. Alternatively, if reason (b) prevails, then one might expect the difference in comorbidity levels between rehabilitation and non-rehabilitation groups to be most pronounced in the FFS patient population because of the financial incentives of the SNF to provide more rehabilitation care to FFS patients. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 Hypothesis 7 Medicare MCO enroUees receiving therapy services have significantly fewer comorbidities than SNF Medicare MCO enroUees who do not receive therapy services. Hypothesis 8 Medicare FFS beneficiaries receiving therapy services have significantly more comorbidities than SNF Medicare FFS beneficiaries who do not receive therapy services. Characteristics o f treatment. Bivariate statistics can also be generated to test for differences in rehabilitation therapy treatment by payment source. Interpreting differences in treatment by payment source requires an understanding of several important elements of per diem contractual agreements between providers and health plans. The first is the utilization review mechanism that MCOs employ to monitor the course and effectiveness of rehabilitation services provided in SNFs. There may be differences in average length of stay between managed care and FFS Medicare patient groups because MCOs are less likely to continue reimbursement for services if improvement is not demonstrated within the parameters of the utilization review mechanism. Also, depending on how financially advantageous the terms of the managed care contract are to the SNF provider, the level of care criteria may create incentives for providers to offer more or less units of rehabilitation services per day. However, predicting the direction of differences in treatment intensity is confounded by the potential for differences in health status at admission. If Medicare FFS and Medicare MCO patients differ in health status at admission (Hypothesis 2), their treatment regimens may also differ. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 Hypothesis 9 There is a statistically significant difference between Medicare FFS beneficiaries and Medicare MCO enroUees in the average number of evaluation units received during a rehabilitation stay in a SNF. Hypothesis 10 There is a statistically significant difference between Medicare FFS beneficiaries and Medicare MCO enroUees in the average length of a stay in a rehabilitation program. Hypothesis 11 There is a statistically significant difference between Medicare FFS beneficiaries and Medicare MCO enroUees in the average length of stay in a SNF. There is a statistically significant difference between Medicare FFS beneficiaries and Medicare MCO enroUees in the average number of units of therapy received per day in a SNF. There is a statistically significant difference between Medicare FFS beneficiaries and Medicare MCO enroUees in the average total number of units of therapy received in a SNF. Characteristics at discharge. The two payment source groups may also differ in their discharge destination characteristics. Medicare managed care plans have a financial incentive to avoid costly re-hospitalizations of their enroUees. Managed care plans also are more likely to arrange for less costly home health care when the patient’s health status permits it. Hypothesis 12 Hypothesis 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 Hypothesis 14 Medicare MCO enroUees are less likely than Medicare FFS beneficiaries to be re-hospitalized directly from a SNF. Hypothesis 15 Medicare MCO enroUees are more likely than Medicare FFS beneficiaries to be discharged home from a SNF. Rehabilitation patients can undergo two discharges — one from the rehabilitation program and another from the SNF. The two discharges may also be concurrent The options for discharge from the rehabilitation program are more extensive than the discharge from the SNF. Patients in a rehabilitation program can be discharged back from the program into skilled nursing care with either the same level of functioning or improved functioning. Patients can also be discharged from the rehabilitation program to receive less intense “restorative” care in the facility. These non-therapy levels o f care are similar to services that could be rendered by a home health care provider. Thus, Medicare MCO enroUees may be less likely to be discharged to these levels of care because the managed care plan may opt for less expensive home health care. Hypothesis 16 Medicare MCO enroUees are less likely than Medicare FFS beneficiaries to be dischargedfrom the therapy program into skilled nursing or restorative care in the facility. Finally, the payment source groups may differ in the proportion of patients who die during the SNF stay and/or rehabilitation episode. The direction of differences in mortality is difficult to predict because it is closely related to differences in health status at admission and treatment characteristics. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 Hypothesis 17 There is a statistically significant difference in level o f mortality between Medicare FFS beneficiaries and Medicare MCO enroUees receiving rehabilitation care in a SNF. FFS beneficiaries and MCO enroUees may also be re-admitted into the therapy programs at different rates. The direction of difference depends on the severity o f their primary diagnoses, their general health status including adverse effects of comorbid conditions, and health care utilization factors such as greater use of home health care and other alternatives to SNF-based rehabilitation programs. Hypothesis 18 There is a statistically significant difference in percentage of therapy patients re-admitted into a therapy program between Medicare FFS beneficiaries and Medicare MCO enroUees receiving rehabilitation care in a SNF. The above hypotheses presuppose aggregate differences between payment source groups in health status at rehabilitation admission and differences in treatment intensity during rehabilitation episodes. Contradictory factors may cancel each other out such that no significant differences in initial health status and/or treatment intensity wiU be observed between FFS and MCO patient groups. However, payment source may still influence individual-level rehabilitation outcomes in SNFs. The two groups of patients may differ in functional status at discharge. Hypothesis 19 There is a statistically significant difference in health status at discharge between Medicare FFS beneficiaries and Medicare MCO enroUees. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 Bivariate tests comparing means (or distributions in the case of non-normally distributed variables) do not produce independent effects of payment source on functional health outcomes. It is necessary to employ empirical tools such as multivariate regression to model independent and interaction effects of payment source on health outcomes. Multivariate Models According to Hebert (1997), the conventional health outcome multivariate regression equation takes the following form: Equation 1 Outcome = y f ib + Demographic, + fiz Comorbidityj + Pi Severityj + y S t Treatment, + C j This basic equation is a departure point for multivariate models, but there are several issues unique to functional health outcomes research that can complicate the analysis. The first concerns the often ordinal nature of the dependent variable. The second is the need to consider how treatment effects interact with other independent variables like baseline health status and sociodemographic characteristics. Ordinal dependent variable. Modeling the dependent variable is complicated in instances where the intent is to measure ordinally-scaled functional disability scores. When a variable is ordinal, its categories can be ranked from low to high, but the distances between adjacent categories are unknown (Long, 1997). Interval variables are ones in which the distances between adjacent categories are equal. Recognition of the ordinal nature of functional disability scales is important for postacute providers. In instances where reimbursement is tied to health outcomes via Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 increases in patient functional performance, the treatment of an ordinal scale as an interval level variable is problematic. A “gain score” of 2 on scale can have different meanings depending on the point on the scale where the “gain” is measured. For example, one person could receive a score of 0 (dependent) at admission and 2 (moderate assistance) at discharge, while another could receive a score 4 (standby assistance) at admission and a 6 (independent) at discharge. Both would have a gain score o f 2, but the true distances between each pair of points may not be equal. The distances may not be equal due to the floor effect in functional disability measurement. For example, the FIM does not indicate the full extent of dependence of patients whose functioning is judged to be in the most dependent category (Heineman, Linacre, Wright, Hamilton, and Granger, 1993). Hence, the resources necessary to help a person improve from dependent to moderate assistance may be much greater than those needed to help a person improve from standby assistance to independent. The proposed study employs predicted probability analysis to address the issue of the ordinal nature of the ROM. Predicted probability analysis permits the estimation of the effect of treatment processes on the probability of reaching a specific functional level along the ordinal ROM scale at discharge, holding all other variables constant at their means or medians. The statistical analysis software known as Stata is used to conduct ordered logit analysis. Patient ROM score at discharge is modeled as the ordered dependent variable, with most of the individual-level variables in Table 1 included in the equation as independent variables. Interpretation of predicted probability results is such that the effect of one independent variable can only be modeled while holding other variables constant. This Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 0 involves setting the values of the other interval-level independent variables at their means or medians and computing separate analyses for both values of dichotomous variables o f interest (though means can be computed for dichotomously-scaled variables that are secondary to each analysis). Thus, to interpret the effect of payment source on the predicted probability of a patient scoring minimal assistance (3) at discharge, one can hold all other interval-level variables constant at their means or medians and produce the predicted probabilities of performing at various functional levels at discharge for several combinations of the dichotomously scaled independent variables. Interaction effects. The effects of treatments on health outcomes are often mediated by initial health status and sociodemographic characteristics. Hebert (1997) summarizes the difficulty in modeling the treatment as an independent effect (as in Equation 1); The model suggests that initial health status as measured by comorbidity and severity variables does affect the outcome variable but that initial health status has no mitigating effect on the effect of treatment. That is, the amount of improvement in the outcome variable due to the treatment is the same, regardless of the initial health status of the patient. This is clearly at odds with the accepted clinical conceptualization of the effect of treatment, yet it is standard practice in outcomes research, (p. 117) This dissertation responds to Hebert’s call to develop predictive models that are more “conceptually isomorphic” with the clinical conceptualization of the effect of treatment. The following equation illustrates how an interaction effect can be modeled: Equation 2 Outcome = > 5 b + A Demographic, + Comorbidity, + Pi Severity] + > 3 » Treatment, + Ps Treatment, x Severity) + e, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 61 Multivariate regression models with interaction terms are sufficient to c^ture the interplay between the effects of independent variables on a dependent variable. Analyses will be conducted to determine if the effect of payment source on health outcomes is mediated by the severity of initial health status (as measured by ROM score at admission). Introducing interaction effects into a predicted probability analysis is straightforward when at least one of the independent variables of interest is measured on an interval scale or in cases where both independent variables are measured dichotomously. The process can be cumbersome when one of the independent variables o f an interaction is ordinal, as is the case in the proposed study where ROM score at admission represents the initial health status construct. One alternative is to create a “dummy” variable created for each level of the ordinal scale (e.g., l=dependent at admission, 0=all other levels; l=maximum assistance at admission, 0=all other levels, etc.). Each dummy variable could then be subsequently interacted with payment source to test whether the influence of payment source on functional disability outcomes is consistent across all levels of initial health status. This method is used in the results chapter to test for interaction effects between initial health status and payment source. Another method of modeling the relationship between initial health status and payment source is based on this study’s substantive hypothesis about the effect of payment source on health outcomes. The method involves conducting predicted probability analysis to model how the effect of payment source on functional status at discharge varies depending on two different levels or profiles of initial health status. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 The reason for testing two profiles of initial health status is based on the potential “floor effect” in fimctional disability scoring. As stated above, it is often more difficult to help a patient improve firom a dependent state to a state of moderate assistance than it is to help a patient improve firom a state of moderate assistance toward independence. This dynamic of rehabilitation measurement may play an important role in how initial health status interacts with the reimbursement mechanisms of postacute care and may be observable in an analysis of level effects in predicted probability models. Payment source may affect functional disability outcomes differently depending on whether patients enter into rehabilitation with dependent ROM scores versus entering into rehabiUtation with relatively moderate levels of functional disabiUty. The financial incentive structures of provider/MCO contracts and Medicare FFS reimbursement are very different. The utilization review mechanisms of managed care plans are structured to monitor the course of rehabiUtation services. If improvement is not demonstrated, the provider’s services may no longer quaUfy for reimbursement. In Medicare FFS reimbursement, the patient quaUfies for a specific number of days of therapy and the provision of services is not subject to strict utiUzation review. Indeed, the final incentives of Medicare FFS are such that the SNF is rewarded for providing more therapy units to patients. MCO patients with dependent scores may not improve quickly due to the nature of dependency and so the utilization management process of MCOs may not permit the patients to continue to receive rehabiUtation services. Thus MCO patients with very poor initial health status may be less likely as group to improve enough to demonstrate modestly better functional disabiUty scores at discharge. FFS patients entering into a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 facility with dependent functional ability scores may be more likely to reach more independent functional disability levels at discharge given provider incentives to maximize reimbursement and the lack of utilization review mechanisms. The hypothesized role that initial health status plays in the relationship between payment source and functional health outcomes can be stated formally as: Hypothesis 20 FFS Medicare status (compared to MCO status) increases the probability of scoring minimal assistance (3) at discharge when functional disability score at admission is total assistance (0). holding all other sociodemographic and treatment protocol variables at their means or medians. Payment source is less likely to have an effect on functional disability outcomes for those FFS and MCO patients entering into rehabiUtation with relatively moderate functional disabiUty scores. The ordinal nature of fimctional disabiUty is hypothesized to be such that as relative independence increases, so too does the likelihood of achieving greater independence. MCO patients entering into rehabiUtation with a ROM score of minimal assistance (3) are more likely to demonstrate improvement and the financial incentives to discontinue rehabiUtation will not be as strong. Hypothesis 21 Medicare FFS status (compared to MCO status) does not increase the probability of scoring modified independence (5) at discharge when functional disability score at admission is minimal assistance (3). holding all other sociodemographic and treatment protocol variables at their means or medians. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 Summary. The hypotheses outlined above are specific statements about the relationship between payment source and characteristics of the postacute treatment experience for older adults. Each hypothesis is explored in the context of the substantive areas of the conceptual model in the following three chapters describing the results of the analyses. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 IV. RESULTS: CHARACTERISTICS AT ADMISSION Sociodemographics. Medicare FFS and Medicare MCO patients who received rehabilitation therapy services were mostly similar in their sociodemographic composition (see Table 4). There was no significant difference in the mean age of the two payment source groups. There was a modest difference in the percentage of Medicare FFS beneficiaries who were female (65.5 percent) compared to MCO enrollees (58.1 percent) (% ^ value p < .07). FFS beneficiaries were also slightly less likely to be married (25.8 percent) compared to MCO enrollees (33 percent) (% ^ value p<.07). The results do not conclusively support Hypothesis 1. Table 4. Sociodemographic Characteristics of Medicare FFS and MCO Rehabilitation Patients Sociodemographic Characteristic FFS MCO N=247 N=276 Average Age 81.5 (8.1) 80.4 (8.3) Female* 65.5% 58.1% Married* 25.8% 33.0% (X ^ value p < .07). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 Health Status. Primary diagnosis. The wide range of primary diagnoses makes it difBcult to test for statistically significant differences in the distribution of diagnoses within each payment source group. Recall that patients may receive several primary diagnoses, one assigned by the facility and others assigned by the rehabilitation therapist upon admission into a program of care. The diagnoses may or may not be identical. Table 5 lists the distribution of the primary diagnoses assigned by the SNF within each payment source group. Tables 6,7, and 8 list the distributions of diagnoses assigned by physical, occupational, and speech therapy, respectively. Specific cerebrovascular disorders (strokes and transient ischemic attacks) and fractures of the hip or pelvis were the two most frequent SNF-assigned primary diagnoses. A slightly larger percentage of MCO patients (14.9 percent) had a specific cerebrovascular disorder as their primary diagnosis of than did FFS patients (12.5 percent). The MCO and FFS groups had similar percentages of hip or pelvis firactures (10.1 percent and 11.2 percent, respectively). The most frequent primary diagnosis assigned by the physical and occupational therapy programs was “signs and symptoms with chief complaint.” Physical and occupational therapists used this category, which encompasses the ICD-9 code of “debility” to classify patients who would otherwise be classified in one of the more specific DCGs. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 Table 5. Primary Diagnoses Assigned by the SNF, by Payment Source. Diagnosis Classification Group FFS(%) N=247 MCO(%) N=276 Degenerative nervous system disorders 8.5 9.8 Specific cerebrovascular disorders 12.5 14.9 Ear, nose, mouth, & throat malignancy 4.5 3.6 Chronic obstructive pulmonary disease 2.4 1.8 Simple pneumonia with chief complaint (CC) 3.2 9.4 Circulatory diseases with acute myocardial 4.0 5.1 Peripheral vascular disorders with CC 4.9 2.5 Esophagilitis, gastroenteritis, & misc. digestive disorders 6.9 4.3 Fractures of the hip or pelvis 10.1 11.2 Bone diseases & specific arthropathies 3.2 6.5 Fractures, sprains, strain or dislocation of upper arm or lower leg with CC 2.8 1.1 Cellulitis age >17 with CC 1.2 4.3 Nutritional and misc. metabolic disorders with CC 3.2 2.2 Kidney and urinary tract infections with CC 2.4 2.9 Signs and symptoms with CC 2.0 0.7 Medical back problems 2.0 2.9 Skin ulcers 2.0 1.8 Other 24.2 15.0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 8 Table 6. Primary Diagnoses Assigned by Physical Therapy Program, by Payment Source. Diagnosis Classification Group FFS (%) N=240 MCO (%) N=266 Degenerative nervous system disorders 2.0 0.8 Specific cerebrovascular disorders 17.6 23.8 Ear, nose, mouth, & throat malignancy 0.4 0.0 Chronic obstructive pulmonary disease 1.2 0.8 Simple pneumonia with chief complaint (CC) 2.0 2.7 Circulatory diseases with acute myocardial 2.4 3.0 Peripheral vascular disorders with CC 0.8 0.8 Fractures o f the hip or pelvis 4.0 4.2 Bone diseases & specific arthropathies 2.0 2.3 Fractures, sprains, strain or dislocation of upper arm or lower leg with CC 0.0 1.5 Cellulitis age > 17 with CC 1.2 3.0 Signs and symptoms with CC 34.8 25.0 Medical back problems 2.4 1.1 Skin ulcers 3.2 1.5 Other multiple significant trauma 4.8 6.1 Fractures of femur 6.8 5.7 Other 14.4 17.7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 Table 7. Primary Diagnoses Assigned by Occupational Therapy Program, by Payment Source. Diagnosis Classification Group FFS (%) N=163 MCO (%) N=101 Degenerative nervous system disorders 5.3 5.0 Specific cerebrovascular disorders 15.9 30.0 Ear, nose, mouth, & throat malignancy 0.6 0.0 Chronic obstructive pulmonary disease 2.4 3.0 Simple pneumonia with chief complaint (CC) 0.6 0.0 Circulatory diseases with acute myocardial 1.2 0.0 Fractures of the hip or pelvis 7.6 10.0 Bone diseases & specific arthropathies 2.4 1.0 Signs and symptoms with CC 45.3 36.0 Medical back problems 0.6 0.0 Skin ulcers 0.6 0.0 Cranial & peripheral nerve disorders with CC 1.2 0.0 Nonspecific arthropathies 1.8 1.0 Traumatic injury 0.6 3.0 Other multiple significant trauma 4.7 1.0 Other 9.2 11.0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 The two most frequent primary diagnoses assigned by the speech therapy program were specific cerebrovascular disorders (MCO=52.3 percent and FFS=32.5 percent) and esophagilitis, gastroenteritis, and miscellaneous digestive disorders (MCO=28.4 percent and FFS=37.5 percent). The speech therapy program rarely assigned the signs and symptoms with chief complaint diagnosis to patients. Table 8. Primary Diagnoses Assigned by Speech Therapy Program, by Payment Source. Diagnosis Classification Group FFS (%) N=79 MCO (%) N=69 Degenerative nervous system disorders 2.5 0.0 Specific cerebrovascular disorders 32.5 52.3 Esophagilitis, gastroenteritis, & misc. digestive disorders 37.5 28.4 Signs and symptoms with CC 1.3 1.5 Other mental disorders 12.5 9.0 Other 13.7 8.8 The relatively high percentage of stroke, hip fracture, and signs and symptoms diagnoses allows for exploratory diagnosis-specific analyses o f treatment and outcome differences between payment source groups. The results of diagnosis-specific comparisons are described later in this chapter and in the treatment and outcomes chapters. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 Comorbidity. Previous research has noted the skewed distribution of the Deyo-Charison Comorbidity Index (Deyo, Cherkin, and Ciol, 1992). The distribution o f the DCCI in this study is similarly skewed (see Figure 3). Figure 3. Distribution of Deyo-Charlson Comorbidity Index )0 6.00 Deyo-Charison Comorbidity Index Score The skewed distribution of the DCCI necessitates use of non-parametric tests to determine whether the two sampled populations (payment source groups) are equivalent in location. The Mann-Whitney U test is the most popular of the non-parametric two- independent sample tests. It is equivalent to the Wilcoxan rank sum test and the Kruskal- Wallis test for two groups. The individual DCCI scores from both groups are combined and ranked, with the average rank assigned in the case of ties. If the populations are identical in location, the ranks should be randomly mixed between the two samples. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 number of times a score from the MCO group precedes a score from the FFS group and the number of times a score from the FFS group precedes a score from the MCO group are calculated. The Mann-Whimey U statistic is the smaller of these two numbers. Table 9 lists the results of the Mann-Whitney test for differences in the distribution of the DCCI scores between the payment source groups. There was a modest difference in the location of the distributions o f the two groups. FFS patients had slightly higher (worse) DCCI scores on average (Mann-Whitney = 33817, p < .08). The mean DCCI scores for the groups are included to indicate the direction of the difference. A sum score of secondary diagnoses was also computed to create a description of patient comorbidity less susceptible to errors in diagnosis coding. A t-test for equality of means was conducted to examine differences in the average number of secondary diagnoses. FFS patients had on average significantly more secondary diagnoses (3.89) than MCO patients (3.36) (p < .001). The results of both comorbidity analyses indicate that the FFS patients had higher levels of comorbidity than did their MCO counterparts. Other measures of health status must be considered before it can be concluded that FFS patients had significantly worse health status than MCO patients upon admission into the SNF. Functional Disabilitv Each therapy program appeared to limit its assessments to a core set of deficit areas. Physical therapy concentrated on bed mobility, transfers, and gait-level surfaces. Occupational therapy concentrated on bathing, dressing, and grooming/hygiene. Speech therapy was less concentrated, though swallowing, cognitive-linguistics, and auditory Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 comprehension were the most frequent deficit areas assessed. Table 10 lists the fr’ equency of assessments made within each deficit area. Table 9. Comorbidity Differences by Payment Source Comorbidity Variable FFS MCO (N=247) (N=276) Deyo-Charlson Comorbidity Index Mean Rank 283.40 261.20 Sum of Ranks 74819.00 72877.00 Maim-Whitney U Statistic 33817.00* Mean Score 1.21 1.01 Number of Secondary Diagnoses Mean 3.89 3.36*** ♦ Mann-Whitney U test p < .08 *** T-test p<. 001 Table 10. Frequency of Assessments Within Each ROM Deficit Area Deficit Area Number of Valid Assessments Physical Bed mobility 349 Gait-level surfaces 433 Gait-uneven terrain 0 Joint Mobility 13 Physical restraints 0 Sitting balance 23 Skin integrity 13 Stair climbing 6 Standing balance 129 Transfers 458 Wheelchair mobility 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 Table 10 (continued) Frequency o f Assessments Within Each ROM Deficit Area Deficit Area Number of Valid Assessments Occupational Bathing 253 Bed mobility 3 Community activities 6 Dressing 249 Emergency response/safety 18 Feeding 18 Functional communication 6 Grooming/hygiene 181 Home management 21 Joint mobility 45 Meal preparation/cleanup 16 Medication routine 0 Physical restraints 1 Toilet hygiene 0 Speech Auditory comprehension 57 Auditory discrimination 0 Cognitive-linguistics 69 Expressive language 53 Reading comprehension 0 Speech production 36 Swallowing 104 Voice 19 Written Expression 2 The ordinal qualities of the ROM deficit areas prohibit the generation of additive summary scores for each therapy program. The ordinal properties of the ROM also make a comparison of means scores within each deficit area problematic. Thus non-parametric tests are employed to examine differences in the distribution of deficit area ROM scores between FFS and MCO patients. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 Table 11 displays the three most frequently assessed ROM deficit areas in each therapy program along with the median score of all patients assessed and the mean scores for MCO patients and FFS patients. The range of scores in each deficit area was 0 to 6, with 6 representing the highest level of independence in each therapy program. Median scores for each deficit are reported as a measure of central tendency. The mean scores, though difficult to interpret due to the original nature of each ROM scale, are provided to illustrate the direction of differences between payment source groups. The Mann-Whitney U statistic was calculated for each deficit area to test for differences in the location of the score distribution between the two payment source groups. Gait-level surfaces was the only ROM deficit area in which there was a clearly significant difference between FFS and MCO patients at admission. MCO patients had significantly higher (more independent) gait-level surface scores at admission (p < .001). There was a modest trend in the same direction for grooming/hygiene (p < .10). Diagnosis-specific tests for differences in the distribution of ROM deficit area scores were conducted for the SNF-assigned diagnoses of stroke and firactures o f the hip and pelvis and for the therapy program-assigned diagnosis of signs and symptoms with chief complaint. Table 12 lists the results of the analysis for the SNF-assigned stroke classification group. Table 13 lists the results of the analysis of the SNF-assigned firactures of the hip and pelvis diagnosis classification group. Only deficit areas assessed by the physical and occupational th er^ y program are included in Table 13 because speech therapy is not part of the treatment regimen for firactures of the hip and femur. Table 14 lists the results of the analysis for the therapy program-assigned diagnosis of signs and symptoms with chief complaint. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 Table 11. ROM Scores for All Diagnosis Classification Groups at Admission, by Payment Source ROM Deficit Area Median FFS N MeanfSD) MCO Mann- Whitney N Mean(SD) U Statistic Physical Bed Mobility 2 164 1.83(1.08) 173 1.90(1.09) 14353.0 Transfers 2 207 1.83(1.08) 237 1.79(1.12) 26069.0 Gait-Level Surfaces*** 2 194 1.79(1.16) 227 2.04(1.14) 20012.0 Occupational Bathing 1 154 1.47(0.83) 89 1.57(0.94) 6861.0 Dressing 1 147 1.51(0.92) 91 1.58(0.92) 6753.0 Grooming/ Hygiene* 2 109 2.18(1.07) 67 2.46(0.98) 3261.5 Speech Swallowing 2 55 2.45(1.57) 45 2.28(1.57) 1276.0 Cognitive Linguistics 2 39 1.79(1.17) 28 2.03(1.29) 563.5 Auditory Comprehension 2 28 1.96(1.07) 27 1.88(1.01) 352.0 * Mann-Whitney U p < . 10 *** Mann-Whitney U p < .001 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 Table 12. ROM Scores for SNF-Assigned Stroke Diagnosis Classification Group, by Payment Source. ROM Deficit Area N FFS Mean(SD) N MCO Mean(SD) Mann- Whitney U Statistic Physical Bed Mobility 18 1.72(1.31) 20 1.45(1.14) 171.0 Transfers 21 1.52(1.29) 29 1.27(1.09) 295.5 Gait-Level Surfaces 19 1.36(1.21) 26 1.61(1.26) 221.0 Occupational Bathing 15 1.13(0.91) 16 1.81(0.94) 123.0 Dressing 13 1.00(0.81) 17 1.47(1.00) 93.0 Grooming/Hygiene 12 1.50(1.24) 15 1.93(1.38) 74.0 Speech Swallowing 7 2.00(1.15) 17 2.11(1.57) 69.0 Cognitive Linguistics* 9 0.89(1.16) 13 2.07(1.32) 40.0 Auditory Comprehension 12 1.58(1.08) 12 1.50(0.79) 69.0 * Mann-Whitney U p < .10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 Table 13. ROM Scores for SNF-Assigned Fractures of the Hip and Pelvis Diagnosis Classification Group, by Payment Source. ROM Deficit Area N FFS Mean(SD) MCO N Mean(SD) Mann- Whitney U Statistic Physical Bed Mobility 20 1.95(0.99) 24 1.87(0.89) 228.5 Transfers 25 1.80(1.04) 27 1.70(0.86) 361.0 Gait-Level Surfaces 23 1.86(1.32) 27 2.03(0.93) 301.5 Occupational Bathing 17 1.17(0.63) 13 1.30(1.03) 116.5 Dressing 17 1.23(0.75) 14 1.14(0.94) 126.5 Grooming/Hygiene 14 2.21(1.31) 10 3.10(0.73) 43.5* p < .10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 Table 14. ROM Scores for the Therapy-Assigned Signs and Symptoms Diagnosis Classification Group, by Payment Source. ROM Deficit Area FFS N MeanfSD) MCO N Mean(SD) Mann- Whitney U Statistic Physical Bed Mobility 61 1.85(1.10) 40 1.85(1.02) 1211.0 Transfers 71 1.71(1.16) 62 1.75(1.15) 2188.5 Gait-Level Surfaces 67 1.71(1.09) 58 1.77(1.27) 1857.5 Occupational Bathing 58 1.53(0.82) 16 1.56(0.89) 462.0 Dressing 51 1.60(0.89) 16 1.62(0.82) 401.0 Grooming/Hygiene 37 1.97(0.98) 13 2.38(0.65) 189.5 In general, the diagnosis-specific analyses fail to capture significant differences in the functional disability scores of the two payment source groups. The results parallel the ROM score analysis o f all diagnosis classification groups and thus do not support Hypothesis 2 concerning differences in health status between the two groups at admission. Pre-Treatment Health Care Utilization For purposes of this dissertation, "pre-treatment" refers to patient characteristics and health care utilization patterns o f patients prior to or at the time of admission into the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 0 program of therapy services. This qiialifîer allows for a comparison of the two payment source groups to test for selection effects related to the decision to offer therapy services. Admission source. Medicare FFS and Medicare MCO patients did not significantly differ in their source of admission. O f the 246 FFS patients admitted into the SNF for therapy, 93.2 percent were admitted from a hospital. Ninety-one percent of MCO patients were admitted into the SNF firom a hospital. A chi-square tests for a significant difference between the two percentages was not significant. Hypothesis 3 is not supported by the data. Days between date of onset and admission into therapies. MCO patients admitted from a hospital into a SNF rehabilitation program averaged significantly fewer days in the period between admission into the hospital and admission into one of the SNF therapy programs than did FFS patients. The dates used to compute this span of days were the date of admission into the hospital and the date of admission into each therapy program, thus the difference between the two dates does not represent a clean estimate of the length of hospitalization. Rather, the period may also include days spent in the SNF prior to admission into one of the therapy programs. Table 15 lists each payment source group’s five percent trimmed mean number of days between the onset of the episode and admission into each of the therapy programs. The five percent trimmed mean is used to eliminate the extreme cases from the computation to create a better estimate of central tendency. Several patients with exceptionally long hospital stays in the FFS group skewed the distribution. The largest five percent and the smallest five percent of the cases for both payment source groups Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 were eliminated and comparisons of the means were conducted using the Student’s T- test. Table 15. Average Number of Days Between Admission into the Hospital and Admission into Physical, Occupational, and Speech Therapy Programs. * Rehabilitation Area FFS N M ean\SD ) MCO N M ean\SD) Physical*** 229 15.0(31.2) 241 5.8 (7.0) Occupational*** 156 18.9 (18.3) 87 9.5 (10.6) Speech*** 67 31.4(52.1) 55 9.5(18.2) ^ Five percent trimmed mean estimates *** T-test p < .001 The average number of days between onset of the episode and admission into the physical therapy program for the FFS group was 15 days, compared to only 5.8 days for the MCO group (p < .001). The span of days between onset and admission into the occupational therapy program for the FFS group was 18.9 days, compared to just 9.5 days for MCO patients (p < .001). FFS patients receiving speech therapy averaged 31.4 days between admission into the acute hospital and admission into the speech therapy program, compared to just 9.5 days for MCO patients (p < .001). Hypothesis 4 is supported by the data. Admitted with orders for theranv. The payment source differences in the number of days between onset and admission into a SNF program may be explained in part by differences in the percentage of patients within each patient group who were admitted into the SNF with physician’s Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 2 orders for therapy. If a patient is not admitted with orders, admission into the therapy program must take place a later point in the episode o f care after subsequent physician visits. Table 16 lists the percentage of patients within each payment source group who were admitted into physical, occupational, and speech therapy. Pearson’s chi-square tests were conducted to determine if the observed count within each 2 x 2 cell matched the expected count. Speech therapy was the only rehabilitation program in which there was a significant difference between payment source groups. Only 34.9 percent of FFS speech therapy patients were admitted into the SNF with a physician’s orders for speech therapy, while 62.9 percent of MCO speech therapy patients were admitted into the SNF with orders for speech therapy value p < .001). This finding partially supports Hypothesis 5 and may explain why there is such a large payment source difference in days between onset and admission into speech therapy. If MCO patients receive speech therapy, they are more likely to receive it earlier in their SNF stay (because they were admitted with orders for it) than are FFS patients. Table 16. Percentage of Therapy Patients Admitted into the SNF with a Physician’s Orders for Physical, Occupational, and/or Speech Therapy, by Payment Source. Rehabilitation Program FFS MCO Physical 96.9% 98.5% Occupational 82.7% 78.4% Speech*** 34.9% 62.9% *** (x value p < .001) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 Therapy prognosis selection effect. Patients who received rehabilitation therapy were compared to SNF patients who did not receive therapy to determine if the two patient populations differed in their sociodemographic characteristics, in their number of secondary diagnoses and in the distribution of their primary diagnoses. Statistical analyses were also conducted to examine whether therapy selection effect differences existed within payment groups. Table 17 lists the sociodemographic characteristics of SNF patients admitted into therapy compared to those not receiving therapy in the SNF. Fee-for-service patients receiving rehabilitation therapy were slightly younger on average than FFS patients who did not receive therapy. Table 17. Differences Between Rehabilitation Patients and Non-Rehabilitation Patients Sociodemographic Characteristics, Total and by Payment Source Group. Sociodemographic Characteristic N Rehabilitation Mean or % N Non-Rehabilitation Mean or % Age All Patients 523 80.8 215 80.7 FFS** 247 80.9 96 82.2 MCO 276 80.6 119 79.4 Female All Patients 323 61.8% 131 60.9% FFS 160 64.8% 62 64.6% MCO 163 59.1% 69 58.0% Married All Patients 155 29.6% 63 29.3% FFS 63 25.5% 22 22.9% MCO 92 33.3% 41 34.5% *♦ p<.05 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 Table 18 presents the results o f the therapy selection effect analysis concerning number of secondary diagnoses. The group of patients over age 65 who received no therapy (N=215) did not significantly differ in the average number of secondary diagnoses (Mean=3.44) compared to the group of patients over age 65 who received rehabilitation therapy (N=523, Mean=3.61). There were no significant differences in comorbidity within payment source groups either. These findings do not support Hypotheses 6, 7 or 8. Table 18. Differences Between Rehabilitation Patients and Non-Rehabilitation Patients in the Average Number of Secondary Diagnoses, Total and by Payment Source Group. Group Rehabilitation Non-Rehabilitation N Mean (SD) N Mean(SD) All 523 3.44(1.43) 215 3.60(1.43) FFS 247 3.85(1.31) 96 3.67(1.38) MCO 276 3.39(1.50) 119 3.26(1.46) The distribution of SNF-assigned primary diagnoses for FFS rehabilitation and FFS non-rehabilitation patients are presented in Table 19. The rehabilitation and non rehabilitation FFS patients had similar distributions of primary diagnoses. A larger percentage of non-rehabilitation patients (7.3) than rehabilitation patients (3.2) had a primary diagnosis of simple pneumonia. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 85 Table 19. SNF-Assigned Primary Diagnoses for FFS Rehabilitation Patients and FFS Non-Rehabilitation Patients. Diagnosis Classification Group Rehab. N=247 % Non-Rehab. N=96 % Degenerative nervous system disorders 8.5 7.3 Specific cerebrovascular disorders 12.5 9.3 Ear, nose, mouth, & throat malignancy 4.5 6.3 Chronic obstructive pulmonary disease 2.4 3.1 Simple pneumonia with chief complaint (CC) 3.2 7.3 Circulatory diseases with acute myocardia 4.0 3.1 Peripheral vascular disorders with CC 4.9 6.3 Esophagilitis, gastroenteritis, & misc. digestive dis. 6.9 9.4 Fractures of the hip and pelvis 10.1 9.4 Bone diseases & specific arthropathies 3.2 5.2 Fractures, sprains, strain or dislocation of upper arm or lower leg with CC 2.8 0.0 Cellulitis age >17 with CC 1.2 2.1 Nutritional and misc. metabolic disorders with CC 3.2 1.0 Kidney and urinary tract infections with CC 2.4 2.1 Signs and symptoms with CC 2.0 1.0 Medical back problems 2.0 1.0 Skin ulcers 2.0 2.1 Other 24.2 24.0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 Table 20. SNF-Assigned Primary Diagnoses for MCO Rehabilitation Patients and MCO Non-Rehabilitation Patients. Diagnosis Classification Group Rehab. N=276 % Non-Rehab. N=119 % Degenerative nervous system disorders 9.8 6.7 Specific cerebrovascular disorders 14.9 11.7 Ear, nose, mouth, & throat malignancy 3.6 13.4 Chronic obstructive pulmonary disease 1.8 2.5 Simple pneumonia with chief complaint (CC) 9.4 10.9 Circulatory diseases with acute myocardial 5.1 5.0 Peripheral vascular disorders with CC 2.5 4.2 Esophagilitis, gastroenteritis, & misc. digestive dis. 4.3 1.7 Fractures of the hip and pelvis 11.2 5.0 Bone diseases & specific arthropathies 6.5 5.0 Fractures, sprains, strain or dislocation of upper arm or lower leg with CC 1.1 0.0 Cellulitis age >17 with CC 4.3 1.7 Nutritional and misc. metabolic disorders with CC 2.2 2.5 Kidney and urinary tract infections with CC 2.9 2.5 Signs and symptoms with CC 0.7 0.8 Medical back problems 2.9 1.7 Skin ulcers 1.8 0.8 Other 15.0 23.9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 Table 20 presents the distribution of SNF-assigned primary diagnoses for MCO rehabilitation and MCO non-rehabilitation patients. The two patient populations had similar distributions of primary diagnoses. The only noteworthy difference was in the fractures o f the hip or pelvis category. Five percent of non-rehabilitation patients had a frractures o f the hip and pelvis diagnosis, compared to 11.2 percent of MCO rehabilitation patients. Summarv. The two payment source groups appear similar in their sociodemographic characteristics at admission but may vary somewhat in their health status characteristics. The MCO patient group had a lower average number of comorbidities as measured by the count of secondary diagnoses. However, FFS beneficiaries and MCO emollees did not differ significantly in their average fimctional status at admission as measured by the ROM. The two groups had different pre-therapy experiences, with FFS patients spending significantly longer in the period between onset of the condition and admission into therapy. The reasons for these differences are explored more fully in the discussion chapter. The results of the therapy experience are presented next in the characteristics o f treatment chapter. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 V. RESULTS: CHARACTERISTICS OF TREATMENT The payment source groups received significantly different levels of treatment during their stay in the respective therapy programs. Medicare MCO patients received fewer evaluation units, experienced shorter lengths of the stay in the rehabilitation program and SNF, received fewer therapy units per day, and received fewer total units of therapy than Medicare FFS patients. The rehabilitation treatment differences persisted in diagnosis-specific analyses of patients with stroke and fractures of the hip and pelvis. Evaluation Units. Medicare MCO enrollees received significantly fewer evaluation therapy units than Medicare FFS beneficiaries. The differences were evident in each of the three rehabilitation therapy programs. Table 21 lists the average evaluation units for the payment source groups in each rehabilitation program. T-tests to measure the differences between the means were significant at the p < .001 level. FFS patients received an average of 6.6 occupational therapy evaluation units, 7.3 speech therapy evaluation units and 4.2 physical therapy units. MCO enrollees received an average of 3.8 occupational therapy evaluation units, 4.6 speech therapy evaluation units and 3.9 physical therapy evaluation units. The standard deviations for the FFS group were notably larger than the MCO group, reflecting the uniformity imposed on the evaluation process by the terms of managed care contract The difference in average number of evaluation units is also partially attributable to FFS patients’ longer average length of stay in the therapy programs - the longer one is in a therapy program, the greater the likelihood of evaluation. Hypothesis 9 was supported by the data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 Table 21. Average Number of Evaluation Therapy Units in Physical, Occupational, and Speech Therapy, by Payment Source. Rehabilitation Program Evaluation Units FFS N Mean(SD) N MCO Mean(SD) Physical*** 240 4.2 (1.9) 266 3.9 (1.2) Occupational*** 163 6.6 (2.2) 101 3.8 (0.9) Speech*** 79 7.3 (3.2) 69 4.6 (1.8) *** p < .001 Length ofStav. Medicare MCO enrollees had significantly shorter lengths of stay in the rehabilitation programs and in the SNF (see Table 22). A five percent trimmed mean was computed for the rehabilitation lengths of stay and SNF length of stay to remove outliers firom the comparison of the means analysis. The adjusted mean length of stay for Medicare MCO patients in physical therapy was 9.4 days (SD=8.1), compared to 16.4 days (SD=15.1) for Medicare FFS beneficiaries. The adjusted mean length of stay for MCO enrollees in occupational therapy was 7.1 days (SD=4.9) while FFS patients had an average length of stay of 15 days (SD=11.1). MCO enrollees averaged 7.7 days (SD=8.2) in the speech therapy program while FFS patients stayed an average of 16.7 days (SD=13.9) in the program. Hypothesis 10 was supported by the data. The difference between the two payment source groups in average length of stay in the SNF was also marked. MCO enrollees averaged 13.6 days in the SNF (SD=22.4). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 FFS beneficiaries averaged 29.1 days (SD=43.1). Hypothesis 11 was supported by the data. The large differences in the standard deviations of the two means suggests that FFS patients experience much more variation in their rehabilitation and SNF lengths of stay than do MCO patients. The implications o f this finding are explored in the discussion chapter. Table 22. Five Percent Trimmed Mean Lengths of Stay in the SNF and Physical, Occupational, and Speech Therapy Programs. Program o f Care N Length of Stay FFS Mean (SD) N MCO Mean (SD) Physical*** 226 16.4(15.1) 242 9.4 (8.1) Occupational*** 154 15.0(11.1) 97 7.1 (4.9) Speech*** 75 16.7 (13.9) 66 7.7 (8.2) SNF*** 235 29.1 (43.1) 262 13.6 (22.4) *** p < .001 The differences in lengths of stay between the two payment source groups persist in diagnosis-specific analyses. Tables 22 and 23 list the five percent trimmed mean lengths of stay for patients with SNF-assigned stroke and fractures of the hip and pelvis primary diagnoses. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 Table 23. SNF-Assigned Stroke Diagnosis: Five Percent Trimmed Mean Lengths of Stay in the SNF and Physical, Occupational, and Speech Therapy Programs. Program of Care N Length of Stay FFS Mean (SD) N MCO Mean (SD) Physical*** 28 18.3 (20.1) 38 12.8(13.1) Occupational*** 18 14.9 (8.4) 21 7.5 (6.9) Speech*** 14 26.2 (16.9) 25 8.7 (7.2) SNF*** 29 63.5 (73.1) 45 21.9(20.4) *** p < .001 Table 24. SNF-Assigned Fractures of the Hip and Pelvis Diagnosis: Five Percent Trimmed Mean Lengths of Stay in the SNF and Physical, Occupational, and Speech Therapy Programs. Program of Care N Length of Stay FFS Mean (SD) N MCO Mean (SD) Physical*** 25 17.4(11.6) 29 12.5 (7.1) Occupational*** 17 16.0 (6.8) 14 9.1 (6.0) SNF*** 25 27.0(18.1) 29 14.8 (10.4) *** p < .001 Therapy Units Per Day. Medicare MCO enrollees received significantly fewer therapy units per day on average than Medicare FFS beneficiaries. Table 25 lists the average th er^ y units per day in each rehabilitation program for the two payment source groups. The smaller Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 number of patients within each group (compared to the evaluation analysis) reflects the fact that only patients who received non-evaluation therapy units were included in the analysis. All t-tests were significant at the p < .001 level. MCO patients received an average of 4.2 physical therapy units per day (SD=1.2), 2.1 occupational therapy units per day (SD=0.6) and 2.3 speech therapy units per day (SD=0.8). FFS patients received an average of 5.7 physical therapy units per day (SD=5.0), 5.7 occupational therapy units per day (SD=1.5), and 4.0 speech therapy units per day (SD= 1.4). Table 25. Average Number of Therapy Units Per Day in Physical, Occupational, and Speech Therapy, by Payment Source. Therapy Units Per Day FFS MCO Program of Care N Mean (SD) N Mean (SD) Physical*** 233 5.7 (5.0) 255 4.2 (1.2) Occupational*** 155 5.7(1.5) 89 2.1 (0.6) Speech*** 75 4.0 (1.4) 57 2.3 (0.8) *** p < .001 Tables 26 and 27 contain similar results of diagnosis-specific analyses for patients with stroke or fractures of the hip and pelvis as their primary diagnosis. Hypothesis 12 was supported by the data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 Table 26. SNF-Assigned Stroke Diagnosis; Average Number of Therapy Units Per Day in Physical, Occupational, and Speech Therapy, by Payment Source. Therapy Units Per Day FFS MCO Program of Care N Mean (SD) N Mean (SD) Physical*** 23 5.5 (1.9) 34 4.2 (1.4) Occupational*** 17 5.5 (1.3) 15 2.3 (0.6) Speech*** 14 4.9 (1.5) 22 2.3 (0.9) p < .001 Table 27. SNF-Assigned Fractures o f the Hip and Pelvis Diagnosis: Average Number of Therapy Units Per Day in Physical, Occupational, and Speech Therapy, by Payment Source. Therapy Units Per Day FFS MCO Program of Care N Mean (SD) N Mean (SD) Physical*** 26 7.1 (1.1) 29 4.9 (0.8) Occupational* * * 18 6.1 (0.9) 14 2.4 (0.6) *** p < .001 Total Therapy Units. Given the large differences between FFS and MCO patients in lengths o f stay and therapy units per day, it follows that the groups would significantly differ in total number o f therapy units received. Table 28 lists the average total therapy units received for FFS Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 and MCO patients within the three rehabilitation programs. Patients who received evaluation units but no therapy units were included in the analysis. T-tests were all significant at the p < .001 level. MCO enrollees received significantly fewer total therapy units than FFS beneficiaries. MCO patients received an average of 47.3 physical therapy units (SD=43.2), 15.8 occupational therapy units (SEN16.5), and 21.2 speech therapy units (SD=26.1). FFS patients received an average of 99.8 physical therapy units (SD=89.2), 96.8 occupational therapy units (SD=81.3) and 80.7 speech therapy units (80.1). Hypothesis 13 was supported by the data. Table 28. Average Total Therapy Units in Physical, Occupational, and Speech Therapy, by Payment Source. Program of Care Total Therapy Units FFS N Mean (SD) N MCO Mean (SD) Physical*** 240 99.8(89.2) 266 47.3 (43.2) Occupational*** 163 96.8(81.3) 101 15.8(16.5) Speech*** 79 80.7(80.1) 69 21.2(26.1) p < .001 Summarv. FFS beneficiaries and MCO enrollees had very différent treatment experiences. Patients given care under the Medicare FFS arrangement received fewer evaluation units, experienced shorter lengths of the stay in the rehabilitation program and SNF, received fewer therapy units per day, and received fewer total units of therapy than Medicare FFS patients. The reasons for these differences and their implications for quality of care are Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 explored more fully in the discussion chapter. Results concerning discharge destination and functional health status at discharge are discussed next in the characteristics at discharge chapter. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 VI. RESULTS: CHARACTERISTICS AT DISCHARGE Medicare FFS beneficiaries and Medicare MCO enrollees had significantly different characteristics at discharge firom the SNF and the individual rehabilitation programs. The two payment source groups were compared using bivariate statistics to examine differences in discharge destination (firom the SNF as well as firom each rehabilitation program), mortality, re-admittance into therapy in subsequent months, and fimctional disability levels at discharge. Multivariate analyses also were conducted to measure the independent effect of payment source on fimctional disability at discharge as well test for interaction effects. Discharge Destination. Medicare MCO enrollees were less likely than Medicare FFS patients to be hospitalized upon discharge firom the SNF. Cross-tabulation analyses with chi-square tests were conducted to compare the percentage of patients in each payment source group who were hospitalized firom the SNF. Nearly one quarter of all FFS patients were hospitalized upon discharge fi’ om the SNF, compared to 11.2 percent of all MCO enrollees (% ^ value p < .001). Conversely, Medicare MCO enrollees were more likely than FFS beneficiaries to be discharged home firom the SNF. Nearly three quarters of all MCO enrollees were discharged home, compared to 49 percent of all FFS patients (% ^ value p < .001). The percentage of all FFS patients discharged fi’ om the SNF to another nursing home (9.3) was significantly larger than the percentage o f MCO patients discharged to another nursing home (3.6) (% ^ value p < .001). Table 29 lists the distribution of the discharge Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 destinations of patients who did not die during their SNF episode of care. Hypotheses 14 and 15 were supported by the data. Table 29. Discharge Destination from SNF of Surviving Patients, by Payment Source. Destination % of All FFS Patients (N=247) % of All MCO Patients (N=276) Hospital*** 24.7 11.2 Home *** 49.0 74.6 Other Skilled Nursing*** 9.3 3.6 Board & Care/Assisted Living 6.1 5.1 The discharge destination distributions were consistent in a diagnosis-specific analysis of patients with a primary diagnosis of stroke. Forty-four percent of FFS stroke patients were discharged from the SNF to a hospital, compared to 21.4 percent of MCO patients value p<.05). Twenty-four percent of FFS stroke patients were discharged home, compared to 64.3 percent of MCO emollees value p<.05). However, differences between payment source groups were not evident in an analysis of patients with fractures o f the hip and pelvis. Table 30 lists the distributions of the discharge destinations for stroke and fractures of the hip and pelvis diagnoses analyzed for differences between payment groups. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 Table 30. Discharge Destination from SNF of Surviving Stroke and Fractures of the Hip and Pelvis Patients, by Payment Source. Stroke Fractures o f the Hip or Femure Destination % of All FFS (N=25) % of All MCO (N=42) % of All FFS (N=26) % of All MCO (N=30) Hospital 44.0** 21.4 11.5 16.7 Home 24.0** 64.3 69.2 70.0 Other Skilled Nursing 12.0** 4.8 3.8 6.7 Board & Care/Assisted Living 4.0 4.8 7.7 3.3 The therapy services database contained information on the discharge destination from physical, occupational, and/or speech therapy. Patients can be discharged from a rehabilitation program to one of the following: hospital, home, board and care or assisted living facility, skilled nursing care at the same level of frmctioning, skilled nursing care with improved functioning, or restorative (non-skilled) care. Similar discharge destination patterns as those above were evident in the rehabilitation program analysis. Compared to MCO enrollees, a larger percentage of FFS patients in all three rehabilitation programs were discharged out of therapy to skilled nursing care with the same level of or improved functioning. This finding along with the larger percentage of MCO patients discharged from the therapy programs to home may reflect the managed care plans greater reliance on home health care as a substitute for offering facility-based Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 skilled nursing care. Tables 31,32 and 33 detail the discharge destination distributions for physical, occupational, and speech therapies, respectively. Table 31. Discharge Destination from Physical Therapy, by Payment Source. Destination % of All FFS Patients (N=240) % of All MCO Patients (N=266) Hospital*** 20.1 9.8 Home *** 49.8 75.6 SNF - Same Level of Functioning*** 2.9 0.8 SNF - Improved Level of Functioning*** 4.2 0.8 Restorative*** 13.4 6.8 Board & Care/Assisted Living 4.2 4.1 The relatively higher percentages of discharges to skilled nursing from the occupational and speech therapy programs compared to physical therapy (for both payment source groups) is noteworthy. The corresponding higher percentages of physical therapy patients discharged home suggests that patients may be more likely to remain in physical therapy afrer discharge from the other rehabilitation programs. Hypothesis 16 was supported by the data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 0 Table 32. Discharge Destination from Occupational Therapy, by Payment Source. Destination % of All FFS Patients (N=163) % of All MCO Patients (N=101) Hospital** 12.9 9.8 Home ** 41.4 59.8 SNF — Same Level of Functioning 9.8 7.8 SNF — Improved Level of Functioning** 27.6 13.7 Restorative 1.8 3.9 Board & Care/Assisted Living 4.6 3.9 ** value p < .05 Table 33. Discharge Destination from Speech Therapy, by Payment Source. Destination % of All FFS Patients (N=79) % of All MCO Patients (N=69) Hospital** 13.9 10.1 Home ** 20.3 47.8 SNF — Same Level of Functioning 10.1 14.5 SNF — Improved Level of Functioning** 43.0 18.8 Restorative 2.5 0.0 Board & Care/Assisted Living 5.1 5.8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 Mortality. Medicare MCO enrollees were less likely to die in the SNF than were Medicare FFS beneficiaries. Four percent of MCO enrollees died in the SNF, compared to 10.5 percent of FFS patients value p < .001). Diagnosis-specific analyses were conducted for stroke and fractures of the hip and pelvis patients. The payment source differences in mortality were not statistically significant in the diagnosis-specific chi-square analyses, but the patterns of difference were similar. Table 34 lists the percentages of all patients who expired as well as the results of the diagnosis-specific analyses. Hypothesis 17 was supported by the data. Table 34. Percentages of All Patients, Stroke, and Fractured Hip or Femur Patients Who Died in the SNF, by Payment Source. Group FFS MCO All Patients*** 10.5 4.0 Stroke 16.0 4.8 Fractures o f the hip and pelvis 2__ i.:_ , ■ ■ ■ ■ 7.7 0.0 Re-Admittance Into Therapy. The therapy services database contained information on multiple rehabilitation episodes for each patient. While only the first episode of rehabilitation was used to address the primary research questions, analysis was conducted to determine if FFS and MCO patients differed in rates of re-admission into a therapy program. The time firame Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 0 2 for re-entry was restricted to the 20 month study period. Table 35 contains the results of the re-admission analysis. A larger proportion of FFS occupational therapy recipients (9.3 percent) were re admitted into the occupational ther^y program at some point later in the study period, compared to MCO occupational therapy recipients (3.3 percent) (% ^ value p < .05). There were no differences in re-admission between payment source groups in the areas of physical or speech therapy, though the percentage of physical therapy recipients who were re-admitted was comparatively high for both payment source groups (FFS=21.5 percent, MCO=19.2 percent). Hypothesis 18 was partially supported by the data. Table 35. Percentage Re-Admitted into Physical, Occupational, and Speech Therapy Programs, by Payment Source. Rehabilitation Program FFS MCO Physical 21.5 19.2 Occupational** 9.3 3.3 Speech 2.0 1.4 Functional Disabilitv. Bivariate analysis. An analysis of the two groups’ distributions of functional disability scores at discharge revealed few significant differences. The ordinal quality of the ROM scales again necessitated the use of non-parametric statistics to analyze differences in the locations of functional disability scores. Consistent with differences observed at Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 admission into therapy, the two groups only differed in the gait-level surfaces deficit area. MCO patients scored significantly higher than FFS patients. Table 36 lists the ROM deficit area scores at discharge for the two payment source groups. Table 36. ROM Scores for All Diagnosis Classification Groups at Discharge, by Payment Source ROM Deficit Area Median FFS N Mean(SD) MCO N Mean(SD) Mann- Whimey U Statistic Physical Bed Mobility 3 164 2.33(1.83) 173 2.52(1.56) 13340.5 Transfers 3 207 2.34(1.69) 235 2.49(1.50) 23522.5 Gait-Level Surfaces** 3 194 2.30(1.60) 227 2.61(1.43) 19728.5 Occupational Bathing 3 153 2.78(1.29) 89 2.69(1.15) 6578.5 Dressing 3 148 3.14(1.57) 91 2.97(1.45) 6305.0 Grooming/ Hygiene 4 110 3.79(1.81) 67 3.80(1.91) 3632.0 Speech Swallowing 4 55 3.51(1.92) 44 3.09(1.98) 1061.5 Cognitive Linguistics 3 39 3.10(1.63) 28 2.60(1.77) 456.5 Auditory Comprehension 3 28 3.43(1.17) 27 2.62(1.77) 283.5 * * p < .05 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 Diagnosis-specific analysis of ROM score differences at discharge were conducted (see Tables 37 and 38). The absence of significant differences between the two patient groups mirrored the analysis of the total population of patients. Hypothesis 19 was not supported. Table 37. ROM Scores for Stroke Diagnosis Classification Group at Discharge, by Payment Source ROM Deficit Area Median FFS N Mean(SD) MCO N Mean(SD) Mann- Whitney U Statistic Physical Bed Mobility 2 18 2.22(1.83) 20 2.10(1.55) 179.0 Transfers 2.5 21 1.71(1.67) 29 2.37(1.42) 241.5 Gait-Level Surfaces Occupational 3 19 1.78(1.75) 26 2.46(1.27) 201.0 Bathing 2 15 2.13(0.64) 16 1.93(1.48) 103.0 Dressing 2 14 1.93(0.82) 17 2.23(1.56) 111.0 Grooming/ Hygiene Speech 3 12 2.83(1.46) 15 2.33(2.10) 76.5 Swallowing 4 7 3.85(2.03) 17 2.94(1.91) 39.5 Cognitive Linguistics 3 9 3.00(1.32) 13 2.46(1.98) 49.5 Auditory Comprehension 3 12 3.17(1.11) 12 2.25(1.91) 51.0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105 Table 38. ROM Scores for Fractures o f the Hip and Pelvis Diagnosis Classification Group at Discharge, by Payment Source ROM Deficit Area Median FFS N Mean(SD) MCO Mann- Whitney N Mean(SD) U Statistic Physical Bed Mobility 3 20 2.75(1.91) 24 3.08(1.66) 225.0 Transfers 3 25 2.88(3.07) 27 3.07(1.35) 336.5 Gait-Level Surfaces 3 23 3.04(1.82) 27 3.11(1.36) 308.0 Occupational Bathing 3 17 3.05(1.56) 13 2.92(1.18) 102.0 Dressing 3 17 3.52(1.50) 14 3.42(1.55) 116.0 Grooming/ Hygiene 4 14 4.00(1.56) 10 5.00(1.33) 46.0 Multivariate analysis. Aggregating ROM scores and comparing the distributions of the two payment source groups is not sufficient for measuring potential independent or interaction effects of payment source on functional disability at discharge. Predicted probability models are needed to test the hypotheses about whether the influence o f payment source varies depending on initial health status. Due to the need for a sizable sample size in predicted probability analysis, it was only possible to generate predicted probability analysis for the three most common deficit areas used in the physical therapy program. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 The results of the predicted probability analysis illustrate the effect that utilization review has on therapy service use and patient outcomes. Hypothesis 20 predicted that FFS patients would be more likely than MCO patients to improve from the most functionally disabled ROM score (0). It was thought that the absence of a utilization review mechanism would encourage continued provision of therapy services even though the patient failed to demonstrate early improvement. Thus, the FFS patients would be more likely to show eventual improvement at discharge. Similarly, Hypothesis 21 predicted no differences in the likelihood of improvement between the FFS and MCO patients in cases where the patient had relatively modest dependency levels at admission into therapy. It was hypothesized that MCO patients with modest functional disability levels would have a better prognosis than MCO patients with more dependent functional disability levels and thus not be differentiated from FFS patients who received therapy units in the absence of a strict utilization review mechanism. The results indicate that MCO patients were more likely than FFS patients to demonstrate improvement while enrolled in the therapy program, regardless of initial health status. Also, MCO patients were less likely to deteriorate in function while enrolled in the therapy program. The critical phrase common to both of the above summary statements is while enrolled in the therapy program. There appears to have been a selection effect operating such that MCO patients who received therapy units did so for two reasons: a) there was an explicit determination prior to admission into the therapy program concerning the patient’s prognosis; and b) those MCO patients receiving therapy units were doing so because they were continuing to show improvement. Once a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 MCO enrollee failed to demonstrate improvement, he or she was discharged fix>m the therapy program. Conversely, a FFS patient who did not show continuous improvement could remain enrolled in the therapy program and continue to receive services. In some instances, a patient might get better and in others a patient might deteriorate. In terms of data analyses to determine the effect of payment source on functional disability outcomes, the end result of the FFS operating enviromnent is a FFS therapy patient population with more varied episode of care experiences. The end result of the MCO operating environment is a MCO therapy patient population with more predictable episode of care experiences. Table 39 lists the results of an ordered logit model designed to examine the independent effects of sociodemographic and treatment characteristics on bed mobility at discharge. The dummy variable terms for the bed mobility at admission construct were created. The reference category was set automatically at the lowest value of bed mobility (0). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 Table 39. Main Effects Ordered Logit Model: Predictors o f Bed Mobility at Discharge from Physical Therapy (N=325). Independent Variable Unstandardized Coefficient Std. Error Standardized Coefficient Odds Ratio Age -.01 .01 -.01 1.01 Female .31 .24 .14 .73 Married .17 .24 .08 .84 Deyo-Charlson -.16* .09 -.07 1.17 Comorbidity Index Admitted from hospital -1.06*** .38 -.49 2.88 Days since onset .01 .01 .01 .99 Admitted with orders .86 .91 .40 .42 for therapy Total units .01*** .01 .01 .99 o f therapy Therapy units .02 .02 .01 .98 per day Managed care .70*** .24 .32 .50 Dummies for Bed mobility at admission (0=reference) Bed mobility =1 .17 .37 .08 .84 Bed mobility =2 1.51*** .34 .70 .22 Bed mobility =3 2.25*** .36 1.03 .11 Bed mobility =4 2.67*** .84 1.23 .07 * * * * p < .10 p < .001 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 Interpretatioa of the payment source coefficient is problematic because of the ordinal nature of the bed mobiUty measure. However, the odds ratios can be interpreted to some extent and the statistical significance informs later predicted probability analysis. The odds ratio for the managed care independent variable can be interpreted as follows: The odds of scoring total assistance (0) for bed mobility at discharge versus the combined outcomes of all the other levels of bed mobihty are .50 times less for MCO patients than FFS patients, holdmg other variables constant Similarly, the odds of scoring (0) and (1) on bed mobihty at discharge versus the combined outcomes of all other categories are .50 times less for MCO patients than FFS patients. Unfortunately, parsimonious interpretations of the odds ratios are not possible because of the ordinal nature of the dependent variable. Table 40 fists the results of the interaction effects model testing for an explicit relationship between payment source and initial health status. Interaction effects were modeled in Stata using the xi: command option. The dummy variables representing the bed mobility at admission categories were interacted with the payment source variable. The interaction terms were not significant. The effect of payment source on functional disability at discharge does not vary depending on a SNF patient’s functional disability at admission into rehabilitation therapy. Hypotheses 20 and 21 were not supported by the data. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 0 Table 40. Interaction Effects Ordered Logit Model: Predictors o f Bed Mobility at Discharge fix> m Physical Therapy (N=325). Independent Variable Unstandardized Coefficient Std. Error Standardized Coefficient Odds Ratio Age -.01 .01 -.01 LOI Female .35 .24 .16 .70 Married .15 .25 .07 .86 Deyo-Charlson -.15* .09 -.07 1.16 Comorbidity Index Admitted from hospital -1.17*** .39 -.53 3.22 Days since onset .01 .01 .01 .99 Admitted with orders .96 .91 .44 .38 for therapy Total units of therapy .01*** .01 .01 .99 Therapy units per day .02 .02 .01 .98 Managed care .53 .55 .25 .59 Dummies for Bed mobility at admission (0=reference) Bed mobility =1 -.22 .54 .08 1.24 Bed mobility =2 1.27*** .49 .70 .28 Bed mobility =3 2.47*** .51 1.03 .08 Bed mobility =4 3.22*** 1.09 1.23 .04 Interaction terms: Bed mob. (1) x MCO .74 .73 .33 .48 Bed mob. (2) x MCO .47 .65 .21 .63 Bed mob. (3) x MCO -.35 .67 -.16 1.42 Bed mob. (4) x MCO -1.31 1.63 -.60 3.72 * p <.10 *** p < .001 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. I l l The independent effects of the key variables can be reinterpreted using predicted probabiUty “macros” developed for the Stata computer software program. Predicted probability analysis is used to interpret the above ordered logit model. The effect of payment source on the probabihty of reaching certain levels of independence at discharge can only be interpreted while the remaining independent variables are held constant. Prior to generating the predicted probability tables, it is useful to first examine the mean, m inim um , and maximum predicted probabihties over the sample. Table 41 lists the values associated with reaching different levels o f bed mobihty independence. Within the sample, the minimum probabihty of scoring minimal assistance (3) at discharge is .05 and the maximum probabihty is .45, resulting in a range o f .40. Similar results are hsted for the other categories. There is sufficient variation in each category to justify further analysis. When the range is too small to be of substantive interest, further analysis is unnecessary (Long, 1997). Table 41. Predicted Probabihties o f Outcomes Within the Sample for the Ordered Logit Model (N=325). Probabihty of Outcome 0 1 2 3 4 5 6 Minimum .01 .01 .02 .05 .01 .00 .01 Mean .22 .05 .14 .36 .15 .02 .06 Maximum .85 .09 .20 .45 .34 .06 .47 Range .84 .08 .18 .40 .33 .06 .46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 1 2 In predicted probability analysis, the effect of a single variable can be examined while the remaining variables are held constant. The convention is to compute predicted probabilities while setting the other independent variables at their means. Because of the non-normal distribution of several of the independent variables, medians are used as the preferred measure of central tendency. For independent variables measured dichotomously, multiple profiles can be computed (e.g., for single females, married females, single males, etc.). Table 42 lists the predicted probabilities associated with all levels o f bed mobility at discharge for FFS and MCO patients with total assistance (0) and minimal assistance (3) at admission and the following characteristics: 1. Unmarried 2. Female 3. Age 81 4. Deyo-Charlson score of 1.0 5. Admitted from hospital 6. Admitted into SNF with orders for therapy 7. Received 49.7 units of physical therapy 8. Received an average of 4.7 units per day 9. Began therapy 6 days after onset of the condition. The results of the predicted probability analysis demonstrate the effect of utilization review in postacute care. MCO patients (predicted probability = .22) are significantly more likely than FFS patients (predicted probability = .13) to score minimal assistance (3) in bed mobility when entering into rehabilitation in the most dependent state (0). Similarly, MCO patients are less likely than FFS patients to decline in Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11 3 functional status when entering into rehabilitation needing minimal assistance (3) in the bed mobility deficit area. The differences in predicted probabilities between the two payment source groups are significant as denoted in the ordered logit model. Table 42. Predicted Probabilities by Payment Source and Initial Health Status for the Bed Mobility Ordered Logit Model (N=325). Bed Mobility at Discharge 0 1 % ? 4 ? Bed Mobility at Admission = 0 FFS .62 .08 .14 .13 .02 .01 .00 MCO .45 .09 .19 .22 .04 .01 .00 FFS-MCO .17 -.01 -.05 -.08 -.02 .00 .00 Bed Mobility at Admission = 3 FFS .12 .04 .14 .45 .17 .02 .06 MCO .06 .02 .09 .43 .26 .03 .11 FFS-MCO .06 .02 .05 .02 -.09 -.01 -.05 The above analysis used the median number of therapy units (49.7) as a measure of central tendency. Further analysis was conducted to test whether MCO patients faired less well when provided with comparatively fewer therapy units. Table 43 lists the results of a predicted probability analysis setting total therapy units to 10 and fixing the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 other independent variables at their medians. The difTerences between the two groups persist when total therapy units are set at 10. Table 43. Predicted Probabilities by Payment Source and Initial Health Status for the Bed Mobility Ordered Logit Model, Total Therapy Units = 10 (N=325). Bed Mobihty at Discharge 0 1 2 3 4 5 6 Bed Mobihty at Admission = 0 FFS .68 .07 .12 .11 .02 .01 .00 MCO .52 .09 .17 .18 .03 .01 .00 FFS-MCO .16 -.02 -.05 -.07 -.01 .00 .00 Bed Mobihty at Admission = 3 FFS .15 .05 .16 .43 .14 .02 .05 MCO .08 .03 .11 .44 .23 .02 .09 FFS-MCO .07 .02 .05 -.01 -.09 .00 -.04 Predicted probabihty analyses were also conducted for the physical therapy deficit areas transfers and gait-level surfaces. Table 44 hsts the ordered logit model for transfers. Admission into the SNF fix)m a hospital, total therapy units provided, transfer status at admission and payment source were significant predictors o f transfers at discharge. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 Table 44. Main Effects Ordered Logit Model: Predictors o f Transfers at Discharge from Physical Therapy (N=429). Independent Variable Unstandardized Coefficient Std. Error Standardized Coefficient Odds Ratio Age -.01 .01 -.01 1.01 Female .32 .20 .16 .73 Married .01 .21 .01 .99 Deyo-Charlson -.06 .08 -.03 1.06 Comorbidity Index Admitted from hospital -.88*** .32 -.42 2.41 Days since onset .01 .01 .01 .99 Admitted with orders 1.19 .86 .57 .30 for therapy Total units o f therapy .01*** .01 .01 .99 Therapy units per day .02 .02 .02 .98 Managed care .56*** .21 .27 .57 Dummies for transfers at admission (0=reference) Transfers =1 .30 .31 .14 .74 Transfers =2 1.62*** .27 .78 .20 Transfers =3 2.18*** .30 1.04 .11 Transfers =4 2.95*** .62 1.41 .05 * p < .1 0 *** p < .001 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 Table 45 lists the predicted probabilities associated with all levels of transfers at discharge for FFS and MCO patients with total assistance (0) and minimal assistance (3) at admission and the following characteristics; 1. Married 2. Male 3. Age 81 4. Deyo-Charlson score of 1.0 5. Admitted from hospital 6. Admitted into SNF with orders for therapy 7. Received 49.7 units of physical therapy 8. Received an average of 4.7 units per day 9. Began therapy 6 days after onset of condition Table 45. Predicted Probabilities by Payment Source and Initial Health Status for the Transfers Ordered Logit Model (N=429) Transfers at Discharge 0 1 2 3_____ 4 $_____ è Transfers at Admission = 0 FFS .64 .06 .15 .13 .02 .00 .00 MCO .50 .06 .20 .19 .04 .01 .00 FFS-MCO .14 .00 -.05 -.06 -.02 -.01 .00 Transfers at Admission = 3 FFS .14 .03 .17 .44 .16 .02 .04 MCO .09 .02 .12 .45 .23 .03 .06 FFS-MCO .05 .01 .05 -.01 -.07 -.01 -.02 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117 While the differences between payment groups appear to be slightly smaller in the transfers at discharge deficit area, MCO patients are still more likely to improve finm the most dependent category and less likely to deteriorate when admitted at the minimal assistance functional level. The smaller differences may in fact be related to the analysis of the married men whereas previous analyses of bed mobility where of single women. The results of the ordered logit model foi gait-level surfaces mirror the analyses of bed mobility and transfers (see Table 46). Table 47 lists the predicted probabihties associated with all levels of gait-level surfaces at discharge for FFS and MCO patients with total assistance (0) and minimal assistance (3) at admission and the following: 1. Married 2. Female 3. Age 81 4. Deyo-Charlson score of 1.0 5. Admitted firom home 6. Admitted mto SNF with orders for therapy 7. Received 49.7 units of physical therapy 8. Received an average of 4.7 units per day 9. Began therapy 6 days after onset of condition Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 118 Table 46. Ordered Logit Model: Predictors o f Gait-Level Surfaces at Discharge from Physical Therapy (N=407). Independent Variable Unstandardized CoefGcient Std. Error Standardized Coefficient Odds Ratio Age -.01 .01 -.01 1.01 Female .32 .20 .16 .73 Married .03 21 .01 .97 Deyo-Charlson C.I. -.03 .08 -.01 1.03 Admitted from hospital -1.15*** .34 -.56 3.15 Days since onset .01 .01 .01 .99 Admitted w/ orders for TX 1.36 .89 .68 .26 Total units of therapy .01*** .01 .01 .99 Therapy units per day .03 .02 .01 .97 Managed care .60*** .22 .30 .55 Dummies for Gait-level surfaces at admission (0=reference) Gait-level surfaces =l .86*** .34 .29 .42 Gait-level surfaces =2 1.07*** .27 .36 .34 Gait-level surfaces =3 1.79*** .28 .60 .17 Gait-level surfaces =4 3.06*** .64 1.02 .05 * p < .1 0 *** p < .00 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119 Table 47. Predicted Probabilities by Payment Source and Initial Health Status for the Gait-Level Surfaces Ordered Logit Model (N=407) Gait-Level Surfaces at Discharge 0 1 2_____3 4 5 6 Gait-Level Surfaces at Admission = 0 FFS .23 .04 .18 .41 .12 .01 .01 MCO .14 .02 .15 .46 .19 .02 .02 FFS-MCO .09 .02 .03 -.05 -.07 -.01 -.01 Gait-Level Surfaces at Admission = 3 FFS .05 .01 .06 .38 .37 .05 .08 MCO .03 .01 .04 .28 .44 .08 .13 FFS-MCO .02 .00 .02 -.10 -.07 -.03 -.05 The differences between payment groups persist in the above analysis of gait- level surfaces at discharge. Because the predicted probabilities are for married female patients admitted from home, the likelihood of improving from the most dependent category is higher (for both groups). Similarly, the likelihood of deteriorating in function is less compared to predicted probabilities of unmarried female patients admitted from a hospital. Summarv. Medicare FFS beneficiaries and Medicare MCO enrollees had different characteristics at discharge in terms of destination, with a significantly larger percentage Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 2 0 of MCO enrollees discharged home as opposed to being re-hospitalized. Aggregate differences in functional health status at discharge as measured by the ROM scales were not apparent. However, comparing the distributions of scores within groups is problematic because o f the selection effect caused by the MCO utilization review mechanism. The results of the utilization review mechanism were apparent in the predicted probability analysis, which showed MCO enrollees to be more likely to improve from the most dependent category, holding all other independent variables constant. The results of the characteristics at discharge analysis are interpreted more fully in the context of the other findings in the discussion chapter. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 2 1 v n . DISCUSSION AND CONCLUSIONS The findings of the dissertation illustrate differences in treatment and outcomes between Medicare FFS and Medicare MCO patients receiving rehabilitation therapy in a large Southern California SNF. Unfortunately, the results can not be generalized beyond the individual SNF, but the analysis succeeds in shedding light on several important issues surrounding the fast changing postacute care environment for Medicare beneficiaries. The dissertation also exemplifies the possibilities and limitations of applied health outcomes research in postacute settings. The discussion chapter is organized around the three major areas of the conceptual model of the study: characteristics at admission; characteristics of treatment; and characteristics at discharge. The results firom each area are synthesized and interpreted within the context of the larger foci o f the study. Finally, implications for public policy and applied health outcomes research are discussed to highlight the significance of the dissertation. Summarv of Studv Results. Characteristics at admission. The finding that Medicare FFS beneficiaries and Medicare MCO enrollees did not significantly differ in sociodemographic characteristics was not surprising. As a whole, MCO enrollees in the population are generally younger. But this study focused on only those FFS beneficiaries and MCO enrollees requiring SNF rehabilitation services. This selection criteria appears to have had a leveling effect in terms of sociodemographic differences. In addition, the high managed care penetration rate in the Los Angeles/Orange County madcet makes it likely that managed care health plans have a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 2 2 higher percentage of older Medicare beneficiaries (who tend to be female and unmarried) relative to other Medicare managed care markets. The absence o f significant difTerences between FFS and MCO patients who received therapy services compared to those FFS and MCO patients who did not was more surprising. The selection criteria used by physicians and therapists to determine eligibility and/or prognosis for therapy services was not measurable in the data analysis. With the exception of age for FFS beneficiaries, the therapy patients and the non-therapy patients did not significantly differ on any of the sociodemographic or health status measures. The interesting findings related to the analysis of the possible selection effect were the large number of MCO patients who did not receive therapy services (N=l 19) and the large number of patients under age 65 being cared for in the SNF (N=271). Managed care plans are utilizing the SNF as a step-down from or replacement for acute care for enrollees of all ages and therapy services are not necessarily part of the care plan for a sizable percentage o f admitted patients. The findings from the selection analysis underscore the need for a better understanding of how criteria for offering therapy services are operationalized within the SNF setting. The significant differences between FFS beneficiaries and MCO enrollees on the measures of comorbidity and the overall lack of significant differences between the two groups in terms of functional disability scores can be interpreted in several ways. The count of secondary diagnoses in a patient’s medical record may be higher for FFS patients because of incentives on the part of hospitals to list more secondary diagnoses to classify patients into more financially advantageous DRGs. No such incentive exists for Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 123 MCO enrollees hospitalized under capitated risk arrangements between health plans and acute care hospitals. The finding that the FFS group who did not receive therapy services had a higher average number of secondary diagnoses (Mean=3.67) than the MCO group who did not receive therapy services (Mean=3.26) supports the conclusion that comorbidity differences between FFS and MCO patients may be artificial and due to characteristics of the health care system. Alternatively, the ROM functional disability measure may be not be sensitive enough to quantify differences in health status at admission between FFS beneficiaries and MCO enrollees. The average ROM scores at admission in all deficit areas were clustered around moderate assistance (2) for both groups. Since most patients o f both payment source groups scored toward the lower end of each ROM scale at admission, the entire seven category distribution of each scale was not manifested in the non-parametric analysis of the payment source groups’ distributions. The differences between Medicare FFS beneficiaries and Medicare MCO enrollees in terms of pre-admission health care utilization further confound a clear interpretation of health status differences at admission. Both groups were equally likely to be admitted into the SNF firom a hospital, but the lengths of time between onset of the condition and admission into the respective therapy programs were significantly different. MCO enrollees entered into SNF-based therapy programs at earlier points in their episodes of care. The differences between payment source groups were substantial - - ranging firom over nine days for occupational therapy to over twenty days for speech therapy. The differences in acute hospital stays may be an indicator that the FFS SNF Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 patient population was more ill on average than MCO SNF patients, but health system characteristics also may have had an effect. The differences in hospital length of stay may be attributable to managed care utilization review during the acute care hospital stay of enrollees as well as selection effects operating to keep FFS beneficiaries in hospitals longer than MCO enrollees. In very competitive markets like Los Angeles/Orange County, managed care has lowered the costs of health care. In these areas. Medicare FFS remains a financially attractive payer for doctors and hospitals. Thus, while the financial incentives under the hospital PPS fimction to discharge FFS patients out of hospitals “quicker and sicker,” the incentives are not as strong as those created by managed care utilization review and FFS remains attractive to acute providers. The differences in days between onset and admission into therapy can be also attributed to managed care utilization review during the SNF length of stay and the SNF provider’s incentives to enroll FFS beneficiaries in therapy programs at any time during their length of stay in the SNF. The smaller percentage of FFS speech therapy patients admitted into the SNF with a physician’s orders for speech therapy (34.9 percent) compared to MCO speech therapy patients (62.9 percent) is evidence of the SNF’s incentive to enroll patients into speech therapy later in their episodes of care. Characteristics of treatment. The large differences in rehabilitation treatment between FFS beneficiaries and MCO enrollees are unambiguous. Medicare MCO patients received significantly fewer evaluation units, experienced shorter lengths of stay in the rehabilitation program and Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 SNF, received fewer therapy units per day and received fewer total therapy units than Medicare FFS patients. The reasons why MCO patients received less therapy treatment could include any or all of the following: a) MCO patients were on average healthier on than FFS patients and thus needed fewer therapy units; b) therapy to MCO patients was discontinued if they failed to demonstrate continuous improvement; c) SNF-based therapy services for MCO patients were discontinued in favor of home-based therapy provision; or d) due to financial constraints/incentives, MCO patients received less than the optimal number of therapy units and/or FFS patients received more than the optimal number of therapy units. The plausibility of each explanation can be clarified by reviewing how the groups differed in characteristics at discharge. Characteristics at discharge. The large differences between FFS beneficiaries and MCO enrollees in discharge destination offers some explanation for the large therapy treatment differences. FFS patients were likely to be more ill (as measured by the higher percentage o f FFS patients re-hospitalized firom the SNF). It also appears as though managed care plans may have opted to substitute home care for the more expensive SNF care whenever possible (as measured by the higher percentage of MCO patients discharged home firom the SNF and out of the rehabilitation programs). The higher percentage of FFS patients who died Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 126 during their episode of care in the SNF also supports the conclusion that FFS patients were less healthy as a group than MCO patients. The fact that FFS beneficiary and MCO enrollee groups did not differ significantly in their average fimctional disability levels at discharge may be a ROM measurement issue. The scales may not be sensitive enough to measure aggregate differences related to payment source. However, the finding also lends support to the conclusion that therapy services to MCO patients were discontinued when patients failed to demonstrate continuous improvement. MCO patients who demonstrated improvement continued to receive therapy services and eventually contributed a higher than average ROM score to the MCO patient group average. MCO patients who did not respond to therapy were not provided with more units and eventually contributed a lower than average ROM score to the MCO patient group average. Scores firom the two types of MCO patients (and those in between) resulted in a distribution of ROM scores that was similar to the FFS distribution. The FFS distribution was made up of patients who received therapy units in the absence of a strict utilization review mechanism. Thus some FFS patients received therapy units and improved accordingly, while others received therapy units yet failed to demonstrate improvement. Predicted probability analysis was useful in illustrating how the MCO utilization review mechanism influenced the probabilities associated with reaching the various ordinally-scaled levels of functional status at discharge. MCO patients who scored total assistance (0) at admission on the bed mobility ROM scale and were similar on all other sociodemographic and treatment characteristics were significantly less likely (predicted probability = .45) to score (0) at discharge compared to FFS patients with similar Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127 characteristics (predicted probability = .62). MCO patients were more likely than FFS patients to score minimal assistance (3) at discharge under similar conditions (MCO predicted probability = .22, FFS predicted probabiUty = .13). The differences in predicted probabihties across the three physical therapy deficit areas illustrate how the MCO utiUzation review mechanism operates to create a MCO patient population that appears more likely to respond to therapy and demonstrate improved functional status. This finding is not inconsistent with the conclusion that FFS patients who received rehabiUtation therapies were on average less healthy than MCO enrollees who received therapies. The MCO enrollees who received therapies most likely did so because they were healthier and thus more likely to respond to the treatment. Similarly, the lack of significant interaction effects between payment source and initial health status suggests that more explicit determinations regarding the prognoses of MCO patients were completed prior to admission into a program of therapy. In other words, it appears as though MCO patients rated as total assistance (0) at admission were judged likely to respond to ther^y using some other criteria, whereas FFS patients rated as total assistance (0) at admission were not similarly screened and thus were less likely to respond to therapy, holding all other independent predictors constant. Recall the earUer reasons put forth as to why MCO patients received fewer therapy units: a) MCO patients were on average healthier than FFS patients and thus needed fewer therapy units; b) therapy to MCO patients was discontinued if they failed to demonstrate continuous improvement; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 c) SNF-based therapy services for MCO patients were discontinued in favor of home-based therapy provision; or d) due to financial constraints/incentives, MCO patients received less than the optimal number of therapy units and/or FFS patients received more than the optimal number of ther^y units. The characteristics at discharge analysis offers support for each of the first three reasons. The final reason is difficult to quantify in the absence of more complete analyses concerning the fate of all patients who did not receive any therapy services as well as the fate of the large percentage of MCO patients who were discharged home. Did MCO patients continue to receive needed therapy services at home? An definitive answer to that question is needed before the results of this study can be used to conclude that MCOs are more efficient at managing the postacute care of Medicare beneficiaries. The issue of FFS patients receiving unnecessary therapy services is similarly difficult to quantify. It appears as though a large percentage of FFS patients received a sizable number of therapy services yet did not show improvement. What about those patients who at first failed to show improvement but eventually made progress? What was the fate of such patients under the managed care financing arrangement? These questions and others highlight the need to conduct applied health outcomes across time and across multiple settings. ImoHcations for Public Policv and Applied Health Outcomes Research. This study offers a rare portrait of how managed care is altering the postacute care delivery system for Medicare beneficiaries. In many respects it raises more questions than it answers. More research like it is needed to fully understand how payment source Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129 influences treatment and outcomes in SNFs and to inform public policy decisions concerning Medicare managed care reform and SNF reimbursement reform. The growth of managed postacute care is unfolding parallel to the issue o f the SNF prospective payment system which begins implement-ation as o f July, 1998. In the case o f the SNF studied for this dissertation, PPS could be viewed as a threat or an opportunity to distinguish itself. The prospective payment system could represent a threat if the SNF has taken advantage o f FFS reimbursement and cost-shifted to remain competitive in the managed care market. The competitive per diem rates negotiated with the managed care plans may have been made possible by the provision of more therapy services on the FFS side. However, it appears as though the SNF has been successful in providing care on a per diem basis (in terms of functional outcomes and discharges home for MCO patients). Thus, operating in a managed care financing environment may have made the SNF better prepared to excel within the PPS case-mix adjusted per diem reimbursement system. This study demonstrates that applied outcomes research in postacute settings is possible and that valuable information about treatments and outcomes can be modeled using empirical tools such as predicted probability analysis. However, the treatment and outcome data of the SNF was not being usçd to its full potential. The fact that the critical data elements (functional status, treatment units) were not linked to the SNF’s main information system at the patient level was troubling and a significant barrier to the analysis. Fortunately, the implementation of PPS and the increasing demands for outcomes information on the part of MCOs make it likely that such gaps in the information infrastructure of SNFs will not last long. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 130 REFERENCES American Heart Association (1992). 1992 Stroke Facts . Dallas TX. Ariing, G., Rhyther, B. J., Collins, T. M., & Zimmerman, D. (1991). Mental illness and psychotropic medication use in the nursing home. Journal o f Aging and Health, 3 ,455-472. National Committee for Quality Assurance (1997). Quality Improvement Outcomes: Defining "MeaningfuV Improvement.. Washington DC. Baum, H. M., & Manton, K. G. (1987). National trends in stroke-related mortality: a comparison o f multiple cause mortality data with survey and other health data. The Gerontologist. 27(3), 293-300. Brainsky, A., Click, H., Lydick, E., Epstein, R., Fox, K. M., Hawkes, W., Kashner, T. M., Zimmerman, S. I., & Magaziner, J. (1997). The economic cost of hip fractures in community-dwelling older adults: a prospective study. Journal of the American Geriatrics Society. 45(3), 281-7. Brown, R. S., Bergeron, J. W., Clement, D. B., Hill, J. W., &, & Retchin, S. M. (1993). The Medicare Risk Program for HMDs - Final Summary report on findings from the Evaluation - Final Version . Princeton, NJ: Mathematica Policy Research, Inc. Buchner, D. M., & Larson, E. B. (1987). Falls and fractures in patients with Alzheimer-type dementia. Journal of the American Medical Association. 257, 1492-1495. Caper, P. (1988). Defining quality in medical care. Health Affairs. 7(1), 49-61. Charlson, M. E., Pompei, P., Ales, K. L., & MacKenzie, C. R. (1987). A new model of classifying prognostic comorbidity in longitudinal studies: Development and validation. Journal o f Chronic Disease. 40{5), 373-383. Cifii, D. X. (1993). Rehabilitation of the elderly crash victim. In S. Retchin (Ed.), Geriatric Clinics: Medical Considerations for the Older Driver (Vol. 2, pp. 473-483). Counte, M. A., & Glandon, G. L. (1996). Managed care for the elderly: Impacts and future trends. In J. C. Romeis (Ed.), Applying Health Services Research to Long-Term Care. (pp. 113-139). New York: Springer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 Deyo, R. A., Cheikin, D. C., & CioU M. A. (1992). Adopting a clinical comorbidity index for use with the ICD-9CM administrative databases. Journal of Clinical Epidemiology. 45(6), 613-619. Dighe, M. S., Aparasu, R. R., & Rapp^ort, H. M. (1997). Factors predicting survival, changes in activity limitations, and disability in a geriatric post-stroke population. The Gerontologist. 57(4), 483-489. Enthoven, A. C., & Vorhaus, C. B. (1997). A vision of quality in health care delivery. Health Affairs. 16(3), 44-57. Farmer, M. E., White, L. R., & Brody, J. A. (1984). Race and sex differences in hip fracture incidence. American Journal of Public Health. 7 4 ,1374-1380. Felson, D. T., J., A. J., T., H. M., & €., M. R. (1989). Impaired vision and hip fracture: The Framingham Study. Journal of the American Geriatrics Society. 37(6), 495-500. Fishman, P., Von Korfif, M., Lozano, P., & Hecht, J. (1997). Chronic care costs in managed care. Health Affairs. 16(3), 239-247. Freedman, V. A. (1996). Family structure and the risk of nursing home admission. Journals o f Gerontology. 5/(2), S61-9. Friedman, B., & Elixhauser, A. (1993). Increased use of an expensive, elective procedure: Total hip replacements in the 1980s. Medical Care. 31(7), 581-599. Friedman, B., & Elixhauser, A. (1995). The changing distribution of a major surgical-procedure across hospitals: Were supply shifts and disequilibrium important? Health Economics. 4(A), 301-314. Fries, B. E., & Cooney, L. M. (1985). Resource Utilization Groups: A patient classification system for long-term care. Medical Care. 23, 110. Fries, B. E., Schneider, D. P., Foley, W. J., Gavazzi, M., Burke, R., & Cornelius, E. (1994). Refining a case-mix measure for nursing homes: Resource Utilization Groups (RUG-Œ). Medical Care. 52(7), 668-685. Gapay, L. (1997, ). States moves to lift lid ofFHMOs. AARP Bulletin. 38^ 1. Gillen, P., Spore, D., Mor, V., & Freiberger, W. (1996). Functional and residential status transitions among nursing home residents. Journals of Gerontology. 5/(1), M29-36. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 Granger, C. V., Hamilton, B. B., Linacre, J. M., Heineman, A. W., & Wright, B. D. (1993). Peformance profiles of the Functional Independence Measure. Archives of Physical and Medical Rehabilitation, 72, 84-89. Greene, V. L., & Ondrich, J. I. (1990). Risk factors for nursing home admissions and exits: a discrete-time hazard fimction ^proach. Journals o f Gerontology, 45(6), S250-8. Greshman, G. E., Duncan, P. W., Stason, W. B., Adams, H. P., Adehnan, A. M., Alexander, D. N., Bishop, D. S., Oilier, L., Donaldson, N. E., Granger, C. V., Holland, A. L., Kelly-Hays, M., McDowell, F. H., Myers, L., Phipps, M. A., Roth, E. J., Siebens, H. C., Tarvin, G. A., & Trombly, C. A. (1995). Post-Stroke Rehabilitation Guideline Technical Report: Vol 1 . Rockville, MD: Agency for Health Care Policy and Research. Halverson, P. K., Kaluzny, A. D., McLaughlin, C. P., Mays, G. P. (Ed.). (1998). Managed Care and Public Health. Gaithersburg, MD: Aspen Publications. Hawes, C., Phillips, C. D., Mor, V., Fries, B. E., & Morris, J. N. (1992). MDS data should be used for research. The Gerontologist, 32(4), 563-564. Hawes, C., Mor, V., Phillips, C. D., Fries, B. E., Morris, J. N., Steele Friedlob, E., Greene, A. M., & Nennstiel, M. (1997). The OBRA-87 nursing home regulations and implementation of the Resident Assessment Instrument: effects on process quality. Journal of the American Geriatric Society, 45(8), 977- 85. HCFA. (1996 December). Monthly Report of Medicare Prepaid Health Plans. Baltimore MD. Hebert, P. (1997). Treatment, Understanding Health Care Outcomes Research, (pp. 93-119). Gaithersburg, MD.: Aspen Publications. Hoenig, H., Rubenstein, L. V., Sloane, R., Homer, R., & Kahn, K. (1997). What is the role of timing in the surgical and rehabilitative care of community- dwelling older persons with acute hip fracture? Archives of Internal Medicine, 157(5), 513-20. Holbrook, T. L., Grazier, K., Kelsey, J. L., & Stauffer, R. N. (1984). The Frequency of Occurance, Impact, and Cost of Selected Musculoskeletal Conditions in the United States. Chicago: American Academy of Orthopedic Surgeons. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 Holbrook, T. L., Barret-Connor, E., & Wingard, D. L. (1988). Dietary calcium and risk of hip fracture: A 14 year prospective population study. Lancet, 2(8), 1046-1049. Horn, S. D., Sharkey, P. D., & Buckle, J. M. (1991). The relationship between severity of illness and hospital length of stay and mortality. Medical Care, 2 9 ,305-317. Isenberg, S. F. (Ed.). (1998). Managed Care. Outcomes, and Quality: A Practical Guide. New York: Thieme. Kane, R. L. (1995). Improving the quality of long-term care. Journal o f the American Medical Association, 273, 1376-1380. Kane, R. L., Chen, Q., Blewett, L. A., & Sangl, J. (1996). Do rehabilitation nursing homes improve outcomes of care? Journal of the American Geriatrics Society, 44{S), 545-554. Kane, R. L., Finch, M., Blewett, L., Chen, Q., Bums, R., & Moskowitz, M. (1996). Use of post-hospital care by Medicare patients. Journal o f the American Geriatrics Society, 44(3), 242-250. Kane, R. L. (1997). Approaching the outcomes question. In R. L. Kane (Ed.), Understanding Health Care Outcomes Research (pp. 1-15). Gaithersburg, MD: Aspen Publications. Kannus, P., Parkkari, J., Sievanen, H., Heinonen, A., Vuori, I., & Jarvinen, M. (1996). Epidemiology of hip fractures. Bone, 18(\ Suppl), 57S-63S. Kemper, P., Spillman, B. C., & Murtaugh, C. M. (1991). A lifetime perspective on proposals for financing nursing home care. Inquiry, 25(4), 333-44. Kemper, P., & Murtaugh, C. M. (1991). Lifetime use of nursing home care. New England Journal of Medicine, 524(9), 595-600. Kiel, D. P., Eichom, A., Intrator, O., Silliman, R. A., & Mor, V. (1994). The outcomes of patients newly admitted to nursing homes after hip ftzcime. American Journal o f Public Health, 54(8), 1281-1286. Kramer, A. M., Steiner, J. F., Schlenker, R. E., Eilertsen, T. B., Hrincevich, C. A., Tropea, D. A., Ahmad, L. A., & Eckhoff, D. G. (1997). Outcomes and costs after hip fracture and stroke: A comparison of rehabilitation settings. Journal of the American Medical Association, 277(5), 396-404. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 Lamphere, J. A., Neuman, P., Langwell, K-, & Sherman, D. (1997). The surge in Medicare managed care: An update. Health Affairs, id(3), 126-133. Lewin-VHI. (1995). Subacute Care: Policy Synthesis and Market Area Analysis. Submitted to Department of Health and Human Services of the Assistant Secretary for Planning and Evaluation. Washington DC. Linacre, J. M., Heineman, A. W., Wright, B. D., Granger, C. V., & Hamilton, B. B. (1994). The structure and stabihty of the Functional Independence Measure. Archives o f Physical and Medical Rehabilitation, 75, 127-132. Long, J. S. (1997). Regression Models for Categorical and Limited Dependent Variables. Thousand Oaks, CA: SAGE Publications, Inc. Lyons, A. R. (1997). Clinical outcomes and treatment of hip fractures. American Journal of Medicine, 103(2A), 51S-63S; discussion 63S-64S. Manton, K. G., Liu, K., & Cornelius, E. S. (1985). An analysis of the heterogeneity o f U.S. nursing home patients. Journals o f Gerontology, 40(1), 34- 46. Manton, K. G., Cornelius, E. S., & Woodbury, M. A. (1995). Nursing home residents: a multivariate analysis of their medical, behavioral, psychosocial, and service use characteristics. Journals of Gerontology, 50(5), M242-51. Means, K. M. (1990). Falls and Fractures. In F. P. Maloney & K. M. Means (Eds.), Rehabilitation and the Aging Population: Physical Medicine and Rehabilitation State of the Art Reviews. (Vol. 4, pp. 39-48). Medicine, I. o. (1996). Health Outcomes for Older People: Questions for the Coming Decade. Washington, DC: National Academy Press. Meyer, J., Silow-Carroll, S.. & Regenstein, M. (1996). Managed Care and Medicare . American Association of Retired Persons. Washington DC. Miller, R. H., & Luft, H. S. (1997). Does managed care lead to better or worse quality o f care? Health Affairs, 16(5), 7-25. Modan, B., & Wagener, D. K. (1992). Some epidemiological aspects o f stroke: mortality/morbidity trends, age, sex, race, socioeconomic status. Stroke, 23(9), 1230-1236. Moon, M., & Gage, B. (1997). Key Medicare provisions in the Balanced Budget Act of 1997. The Public Policy and Aging Report, 5(4), 1,3-5. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 Mot, V., Intrator, O., Fries, B. E., Phillips, C., Teno, J., Hiris, J., Hawes, C., & Morris, J. (1997). Changes in hospitalization associated with introducing the Resident Assessment Instrument Journal of the American Geriatric Society, 45(8), 1002-10. Morgan, R. O., Vimig, B. A., DeVito, C. A., & Persily, N. A. (1997). The Medicare-HMO revolving door— the healthy go in and the sick go out. New England Journal o f Medicine, 337(3), 169-75. Morgan, R. O., Vimig, B. A., DeVito, C. A., & Persily, N. A. (1997). The Medicare-HMO revolving door— the healthy go in and the sick go out. New England Journal o f Medicine, 337(3), 169-175. Murtaugh, C. M., Kemper, P., & Spillman, B. C. (1990). The risk of nursing home use in later life. Medical Care, 25(10), 952-62. Neu, C. R., & Harrison, S. C. (1988). Posthospital Care Brfore and After the Medicare Prospective System (R-3590-HCFA). Santa Monica, CA; RAND. Nitz, N. M. (1997). Comorbidity. In R. L. Kane (Ed.), Understanding Health Care Outcomes Research (pp. 153-166). Gaithersburg, MD.: Aspen Publications. Government Accounting Office (1996). Approval process for certain services may result in higher Medicare costs: Report to the Ranking Minority Member. Washington DC: Special Committee on Aging, U.S. Senate. Government Accounting Office (1997). Home health and skilled nursing facility cost growth and proposals for prospective payment. Statement of William J. Scanlon, Director. Washington DC: Health Care Financing and Systems Issues, Health, Education and Human Services Division. Joint Commission for Accreditation of Healthcare Organizations, (1997). Joint Commission Announces ORYX: The Next Evolution in Accreditation . Chicago, EL . Phillips, C. D., Hawes, C., & Fries, B. E. (1993). Reducing the use of physical restraints in nursing homes: will it increase costs? American Journal of Public Health, 55(3), 342-8. Phillips, C. D., Morris, J. N., Hawes, C., Fries, B. E., Mor, V., Neimstiel, M., & lannacchione, V. (1997). Association of the Resident Assessment Instrument (RAI) with changes in function, cognition, and psychosocial status. Journal o f the American Geriatric Society, 45(8), 986-93. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 Pol, L. G., & Thomas, R. K. (1992). The Demography of Health and Health Care. New York: Plenum Press. Physician Payment Review Commission (PPRC) (1996). Access to Care in Medicare Managed Care: Results from a 1996 Survey ofEnrollees and Disenrollees. Washington DC. Prospective Payment Advisory Commission (ProPAC) (1992). Medicare's Skilled Nursing Facility Rrform (#691). Chicago: Commerce Clearing House. Retchin, S. M. e. a. (1991). Elderly Patients with Congestive Heart Failure Under Prepaid Care. The American Journal of Medicine, 90(2), 236-242. Retchin, S. M., Brown, R. S., Yeh, S. C., Chu, D., & Moreno, L. (1997). Outcomes of stroke patients in Medicare fee for service and managed care. Journal of the American Medical Association, 278(2), 119-24. Riggs, B. L., & Melton, L. J. (1986). Involutional osteoporosis. New England Journal of Medicine, 314, 1676-1686. Riley, T., Chiang, & Ingber. (1996). Health status of Medicare emollees in HMOs and Fee-for-Service in 1994. Health Care Financing Review, / 7(4), 65-76. Riter, R. N., & Fries, B. E. (1992). Predictors of the placement of cognitively impaired residents on special care units. The Gerontologist, 32, 184-190. Schlenker, R. E., Shaughnessy, P. W., & Hittle, D. F. (1995). Patient-level cost of home health care under capitated and fee-for-service payment. Inquiry, 32(3), 252-70. South Coast Rehabiltation Services (1994). Rehabilitation Outcome Measure (ROM): A Training and Resource Guide. . Laguna Hills, CA: Communicology, Inc. Starr, P. (1997). Smart technology, stunted policy: Developing health iriformation networics. Health Affairs, 16(3), 90-104. National Center for Health Statistics (1992). National Hospital Discharge Survey, Annual Summary, 1989. Vital and Health Statistics (Series 13, No. 109; DHHS Publication No. [PHS] 92-1770). Washington DC: Govermnent Printing OfBce. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 137 National Center for Health Statistics (1996). Advanced Report o f Final Mortality Statistics, Annual Summary, 1993. Vital and Health Statistics (Vol. 44, No. 8, Suppl. 2: DHHS Publication No. [PHS] 96-1120). Washington DC: Government Printing Office. Steiner, A., & Neu, C. R. (1993). Monitoring the Changes in Use of Posthospital Services. Santa Monica, CA: RAND. Stone, R. I., & Katz, R. E. (1997). Thoughts on the future of integrating acute and long-term care. In R. J. Newcomer (Ed.), Focus on managed care and quality assurance: Integrating acute and chronic care. New York: Springer. Teresi, J. A., & Holmes, D. (1992). Should MDS data be used for research? (Editorial). The Gerontologist, 3 2 ,148-149. Verbrugge, L. M. (1989). The twain meet: empirical explanations of sex differences in health and mortality. Journal o f Health and Social Behavior, 50(3), 282-304. Vladeck, B. C. (1980). Unloving Care: The Nursing Home Tragedy. New York: Basic Books, Inc. Vladeck, B. C. (1996). From the Health Care Financing Administration. Journal of the American Medical Association, 275(13), 978. Vladeck, B. C. (1996). Medicare payment for home health agency and skilled nursing facility services. Testimony before the Subcommittee on Health, House Committee on Ways and Means. Washington DC. von Sternberg, T., Hepburn, K., Cibuzar, P., Convery, L., Dokken, B., Haefemeyer, J., Rettke, S., Ripley, J., Vosenau, V., Rothe, P., Schurle, D., & Won-Savage, R. (1997). Post-hospital sub-acute care: An example o f a managed care model. Journal of the American Geriatrics Society, 45, 87-91. Wagner, E. H. (1997). Managed care and chronic illness: Health services research needs. Health Services Research, 32(5), 702-714. Ware, J. E., Jr., Bayliss, M. S., Rogers, W. H., Kosinski, M., & Tarlov, A. R. (1996). Differences in 4-year health outcomes for elderly and poor, chronically ill patients treated in HMO and fee-for-service systems. Results from the Medical Outcomes Study. Journal o f the American Medical Association, 275(13), 1039- 47. Weimer, D. L., & Vining, A.R. (1992). Policy Analysts: Concepts and Practice. Englewood Cliffs, New Jersey: Prentice Hall. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 White, B. L., Fisher, W. D., & Laurin, C. A. (1987). Rate of mortality for elderly patients after fracture of the hip in the 1980s. Journal o f Bone Joint Surgery. 69A, 1335-1340. Wiener, J., & Scaggs, J. (1995). Current Approaches to Integrating Acute and Long-Term Care Financing and Services. Washington DC: American Association o f Retired Persons. Wolinsky, F. D., Callahan, C. M., Fitzgerald, J. F., & Johnson, R. J. (1992). The risk o f nursing home placement and subsequent death among older adults. Journals o f Gerontology. 47(4), SI 73-82. Wolinsky, F. D., Callahan, C. M., Fitzgerald, J. F., & Johnson, R. J. (1993). Changes in ftmctional status and the risks of subsequent nursing home placement and death. Journals of Gerontology. 48(3), S94-101. Wolinsky, F. D., Fitzgerald, J. F., & Stump, T. E. (1997). The effect of hip fracture on mortality, hospitalization, and functional status: a prospective study. American Journal o f Public Health. 87(3), 398-403. Zinn, J. S., Brannon, D., & Mor, V. (1995). Organizing for Nursing Home Quality. Quality Management in Health Care. 5(4), 37-46. Zuckerman, J. D., & Schon, L. C. (1990). Hip Fractures. In J. D. Zuckerman (Ed.), Comprehensive Care o f Orthopedic Injuries in the Elderly (Vol. 23-111). Baltimore: Urban and Schwarzenberg. Zuckerman, J. D. (1996). Hip Fracture. New England Journal o f Medicine. 334, 1519. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 Appendix A Managed Care Contract Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140 PER DIEM COMPENSATION STRUCTURE Provider shall be reimbursed by MCQ in accordance with the per diem rate structure referenced below. Compensation shall be based on the corresponding level of skilled care required by the patient as determined by the MCQ physicians and nurse practitioners or physician assistants. All per diems are to include the following services: Semi-private room Meals (special delivery) Prescription and non-prescription medications not lo exceed AWP $40 per day. (PC, IM, NG, SO, IV) Nursing care Oxygen services and supplies Nutrition services including enteral nutrition and supplies Administration of medications including intramuscular and hydration Intravenous services Laboratory services X-Ray services Medical/in-house supplies Speech, occupational, or physical therapy evaluations (Level II & HI) Case management and discharge planning Standard DME - wheelchair, trapeze, walker, commodes, feeding pump Podiatary care Patient and/or family education Level I Per Diem Includes: Application and maintenance of Buck’s traction Post operative care Administration of NG/G Tube J/Tube feedings, including supplies and services Enteral feeding services Colostomy and Ileostomy Care Respiratory therapy, i.e., inhalation treatment Isolation care (no strict isolation) Chronic tracheostomy care Administration of insulin to unstable insulin dependent diabetic patients Intramuscular injections Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141 Level n Per Diem Includes All Level I Services and also the Following: • Rehabilitation evaluation and one (1) modality (1 hour treatment), five (5) treatments per week (FT, OT, or ST) • Administration of IV therapy (includes medical supplies) • Insertion and maintenance of PlCC line • Maintain central line • Acute tracheostomy care • Stage n i and IV pressure ulcer care including dressing and topical wound treatments • Administration of Chemotherapy (including medical supplies) • TPN admministration Level HI Per Diem Includes All Level 1 and H Services and also the Following: • All rehabilitation modalities as specified below: • PT, OT, ST, firom 60 to 180 minutes per day (7 x week PT; 6 x week OT; 5 X week ST*) • One ( 1 ) physical therapy evaluation per admission and two (2) physical therapy treatments per each twenty-four (24) hour period and/or • One (1) occupational therapy evaluation per admission and one occupational therapy treatment per each twenty-four (24) hour period and/or One (I) evaluation of speech therapy per admission and one (1) speech therapy treatment per each twenty-four (24) hour period. * Any additional need for therapy must have written prior authorization fi’ om the MCO and will be paid at $50.00 per thirty (30) minutes. Each treatment will be 30-60 minutes in length, totaling combined therapy to 60 to 180 minutes per day. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 Appendix B ICD-9/DCG Conversion Tables Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7) C D ■ D O Q . C g a. ■ D C D (/) (/) s 3. 3 " (D (D T3 O Q . C g O 3 T3 O C D Q . O C T3 (D (/) (/) IC D -9'8 Î38' Ü Ô i4 Ô iëô tel i7ô'“ tê t" t92*_ 2Ôt.9 2 2 5 -_ 2 Î Ô 231“ 235 “ 2 3 5 r 235:8" 250.6 25Ô.r Z 50J 263' ■ 2Îr 290' 295' 296* 3 i5 .^ 319'" 323 323 6 331.89 331.9 332 333 333.1 333j 3 3 3 .7 _ 333.82 ICpO Ç oiiyffsiüii Tal)l«s Oe»cil|illiiii/liilu PlilK N ilyulilis. lalu dut Is IllIj^llUSIIIS _ lleonjâsmsi ______ lleo|ilasins llecfiissiii, twiie, inalignaiil ___ lletnHasiii. bialn. inaliuiianl________ IleiHilasMis. iiiallgiiaiil, iieivwis Sy& luiii. üiliui ' • — "W », ^ mXii - I ■ ■ ■ — • — - — « ... . . I luilgliiii's OIseasa _ llculHasiii. laaiii. olliei pails ul iicivtnis sjsluiii. Iniiiyii Caicliiotnam Neu|i|asmni pliaiyn* t* laiyim _ CaitiiKima w HutiplasiH ut pltaiyii» m laiyii» Caieiiioiiia ta lleinilasin m ptiaiyii» tu laiyiin Caiciiioiiia M Heo|)lasiii w pliaiyi» tu laiyi» Caicliioma m Ueoptasni m pliaiyii» oi laiyii» _ DIatieles. iieiudoglcd iiiatilleslalioiis IImIh:|us. vtdli |iuil[iiiuial ihiiilaliay iliwailiih Olaliulus. iiltiui inis|H!i.l(luiiiiiaiilu(u!ilaliiiiib MalimliS Itni, im oIuI h caloila. tuisiiucilioil__ <3tail. iiiis|ieclliuU Deiiiaiilla. piesenite w/ ItIV Sclmu)iliieiila l'syclmses, allecllve lluvukuKiieiilal ilc la y _____ neiaitlalluii. uiispecllleti Enceptialills. vlial EiicepliaWls, olliei.iltielolnlecllon EüSEn!!?!!!!?! PudkitadkN i. dastlliinl clsotvtivio Eitcuiilimlilis, iiltiei causa Eiice|ilialilis, twiS|itclllei| cause___________ Degeneialtuii, bialn. usiiaBy luaitiluslud ht cIù I iIIh khI ÂUheliiiers, Pick's Ulcmeulla) (Icgaiieialkiii. lu alu, sculla I lydiocu|ilialus, couaiHmtcalliiH______ I lydiocepliatus, obsliucUve Oeneneiallou, liiain. In disease classilieil elsewlieie Alanla, ceielnal degeneiallon Degeneialloii iM aln, unspeclllutt Paiklusuus Degeiieiallve Oiaki Disease, iilltui tliseascs ul lusai ganylia liemoi. esseiilial Clmiea. lliuilliiQlon's llysltmla. syiii|ilotiialic Imskui Üyskiiiesla, wolaclal________ DCG 34 84 04 64 239 iô iô_ ^ 482 iô C4 2 1 84 84 2 1 2 " 04 2 04 2. ia ” j.îii 204 200 244 489 _ 43Ô 430 431 42Ô 18 18 io io 20 là 221 Î2 12 _ 12 12 S oiled l)v IC09 Desciliilluii OliK» llisiHiluis ul lliu lleiyous Sysieui IHsaases ul liw i ( ai, lluse ami liuual Oiseaitet ul Ilia Eai. tlose ami I limai Diseases ol liia Eai. Hose ami Tiuoal PaliiokiDtcal f lacltaes lleiveus Syilem Neoplasms lleivotis Syslem Neoplasms llmkikéis Disease tleivous System lleti|ilasiiis Diseases ol llie Eai. lluse ami l liiual Diseases ol llie Eat, Muse amlllwual Diseases ol llie Eai, Nose ami Tliinal Diseases ol llie Eai, dose aiiU llwoal Diseases ul llie Eai. Nose ami Tliioal Ciaiilal ami Peiliilieial Netve Olsoidets w/ CC Peiljilioial Vasi iiiai Disease w/ PC I ilaWes ____ lliiliilioiiai ami Miscellaiieous Melaiaiüc Disuiiicis Boue Diseases ami Specilic Ailleoiiailiies HIV ami Olliei iiia)oi Relaled CwKliliou Psycl loses _ Psycl loses _ _ ClilkNiooil, Mental Disoiileis, Delayeil Dcv OiUBiitc Disliuliaiices ami Mental llelaitlallcii Ciaiital ai K l Peiiplieial Heive DIsoideis w > CC Ciaiilal ami Peil|ilieial Heive Disotileis wl CC Ciaiilal aid Peii(ilieial Heiva DIsoiileisw/CC Ciaillâiaid Peii|4ieiai Heive Disuiileis w/ CC liai vous System liileclkiii Degeiieiallve lleivous System Disoitleis Degenei alive lleivous System Disoitleis Degeiieiallve lleivous System Disouleis Degeiiei alive Degeiieiallve Heivoiis System Disoitleis tleivous System DisoiUeis Degenei alive Heivotis System Disoiileis Degeiieiallve tleivous System Disoiileis Degeiieiallve lleivous System Disoitleis Degenei alive tleivous System Disoideis Degeiieialivo lleivous System DisoiUeis _ Degeiieiallve I Degenei alive ï B Heivotis Syslem Disoitleis I Heivous System Disoitlois Degenei alive Neivous System Disouleis Oegeiieialive lleivous System D jso^is tlegeiieia'.ive lieivous System ïtlsonleis_ 6 C D ■ D O Q. C g Q. ■ D C D C /) C /) 8 ci' 3 3 " C D C D ■ D O Q. C a O 3 " O O C D Q. ■ D C D C /) C /) IC D -9 's ICD9 Conve(siun Tables O bsci||iIIuii/IiiIo D CG Soiled by IC D 9 Descilplloii 334- 335" 337.9 Disease, S|Hiiiiceieliellai (7lieilicii'.h's Alania) De||eiieialiim. aiileiKe Ih h m i eP. (A iiiyiiliii|iliii. 1 .ilri.il !li li iiK.e. Disunlei, iieivuiis aysteiii. aiitimuiiik: (slHnililei ImihI swiilnniir) 13 12 lÛ M iiH^nu Scleinsis and CeieliePai Alaaia De||eneiallve lleivnns Syslem Dlsindiis CiaiPal and l'eippieial îleive Disonleis w/ CC 340" M iilli|ite Scleiusls 13 MuHi|Pe Scleiusis and Ceiebellai Alasla 341* Disease, ilemylmaliiiÿ. eus 13 MuPiple Scleioala and Ceiebclai AIaala Degeiieiallve tleivous Syslem Dlscndois 342* Keiiiipleoia. Iioiiiljiaiesis 12 343.9 344.0* Ceielnal l'alsy, Pilaiilile, iiiisnecilieil QiiaUiijilegla 34 9 ................... miiei OIsuideia ul llie Neivoiia Syslem Spinal Disouleis 344.1 l'aïajilegia. uiispecllieil 9 Spinal DIsoideis 344.2 Diplegia, uppei tmlis 9 Spinal Disouleis 348.1 Aiwsla ((lainage lo (lie hiaiii. aiwMc) 34 Olliei Disoideis cl llie Neivous Syslem 348.3 Eiieeplielopalliy, inispecilieit 10 Noiispeclln: Ceieinovasculei Disoideia wl CC 348.5 Edcnia, InaPi 23 Non liaunmllc Sliipoi and Coma 348.8 Disease ol llie bialn, olliei, calcilicallon, liingiis 16 Nonspecilic Ceieinovasculai DIsoideisw/CC 349.9 3 si .............. 353* Disease, iieivoiis ^sleiii, inis|ieeilieil llcHa l'alsy Disinilei, iicive luol, |ile«iis 16 ill 18 Honspecillc CeielnovasctPai DIsoideisw/CC Ciaiilil ainl l'eilipieiul Neive llisindeis w/ CC Cl ai Pal ami l'eiiiPieial Nui vu IHsonlois w/ CC 354* Moiioiieiiiltls, ii|i|ici hiiiti 10 CiaiPal and Peiiplieial Neive Olsmdcis w/ CC 357 Polyiieuillls. aciPe kileclive (GuiPaiii Oaiie Syinlionic) 20 Neivous Syslem Inlecllon 358 Myasllieiiia Giavis, 12 Degeiieiallve Neivous Syslem DIsoideis 358.1 Myasllieiila Oiavis, 12 DegeneiBllve Neivous Syslem Disoideis 359* DysliüfPiy, iiiiisctildi. olliei iiiyiijialWes 34 Olliei DIsoideis ol llie Neivous Syslem 369* llkiideiiees, piolouml visual PiipaPiiieiri 46 Olliei Dlsoideis ul llie Eye 386 Menleie's OIseï le 65 DyseqiPIPnhnii 388.43 Abnoiinalily, atnlPoiy peiceplion 73 Olliei Eai, Nose and lliioal Diagnoses 389* loss, heaiHig 73 Olliei Eai, Nose and îliioal Diagnoses 410* liilaiclion, aciPe, inyocaidial 144 Olliei CbciPalwy Syslem Diagnoses w/ CC 428* 429.4 43Ô ' ClwMik lle a ilF a P iiie___ Dislinbailee, lieail, liiin Iknial, înig Iciiii clleel ni eaiiliai: siiiiii iy llciiioiihage, siibaiacluiukl 121 124 14 CkciPaliny Disindeis w/ AciPe Myocaidial CPciPalii^ Dlsindeis cnceyp AciPe Simclllc Ceieinovasculai DIsoideis (escepi 1IA) Specllk Ceiebiovasculai Dlsindeis (e«ce|p 1IA) 431 lleinoiiliage, kiliaeiaiPal 14 432* 1 leinoiiliage. Iiiliaciaiilal w/ceiebial Inhiclimi 14 Specilk CeieUovasculai DIsoideis (escepl IIA) 433 Occlusion and slenosia w/o ceielnal liilaiellon 15 liansleip IsclienPc Allacli 433.01 Occlusion and slenosls, basUai aileiy wlceieUal Pilaiclion 14 S|iecillc CeieUuvasciPai OIsoideia (eaceiP TIA ) 433.1 Occlusion end slenosia w/u ceielnal Pilaielnn 15 1 lansleip IsclieiiPc Allack 433.11 Ucelusion and slenosia, ceiolid aileiy w/ceiainsi inlaiclnni 14 Specilic Ceiebiovasculai Dlsindeis (eaceiP TIA ) 433.2 Occlusion and slenosia «v/o ceiebial Pilaicllon 15 liansleip IsclieuPc Allack 433.21 Occliision and slenosia, veilebial ailery w/ceiebial liilaielion 14 Speclln: CeieUovasciPai Dlsindeis (eacepi TIA ) 433.3 Occlusion and slenosia w/o ceiebial bilaiclion 15 liansleip Isclioniic Allack 43131 Occlusion and slenosia, inuPIple eileiles w/ceiebial Inlaiclkni 14 Specilic Ceiebiovasculai DIsoideis (eacepl TIA ) 433.8 Occhiakni and slenosls w/o ceiebial inlaiclion 15 TianslenI Iscliemlc Allack 433.81 OccPislon and slenosls, speclliod pieceietnal aileiles w/ceieinal inlaiclnni 14 Specllk Ceieinovasculai Disouleis (escejp IIA) 433.9 Occlusion and slenosls w/o ceiebial Pilaiclkni 15 liansleip Ischemk Allack 433.91 Occlusion and slenosls, iinspecilied pieceieUal ailoiies w/ceielual iiiloicliini 14 Specllk Ceieinovasculai Dlsindeis (e«ce|p 1 IA ) C D ■ D O Q. C g Q. ■ D C D C /) C /) 8 ■ D 3. 3 " C D C D ■ D O Q. C a O 3 " O O C D Q. ■ D C D C /) C /) ICD9 Conversion Tables S oiled by ICD9 IC D -9’8 De(cil|illoii/liilo D C G OesGilpllun 434 I hicHiitioiis w/o ceielNBl liilaicliu» 15 liansleni Iscliemk Allack 43401 1 hlIN IlbO tlS. C elC lN U l. W /i:«IL 'IH dl U llaitiH H I 14 Specilic Ceieinovasculai DIsmdeis{escepl IIA) 434 1 l'iiilmW am w/u C fiulH O l /uldiiltiM i 15 1 lanslenl Iscliemlc Allack 434.11 EltiMisiii. ce/eixal, w/ ceiebial iiilaiclioii 14 Specific Ceiebiovasciilai DIsmdeis (escepl TIA ) 434.9 Occlusion, unspecllicd, w/o ceielnal liilaicilun 15 liansleni Iscliemlc Allack 434.91 Occliiskm. inisjiecjlieil. ceielnal ailciy 14 Specific Ceieinovasculai Disindeis (escepl IIA) 435" Isclieiiila. ceielnal. Iiaiisieiil 15 liansleni Isclnmilc Allack 436 Ceiebiovasciilai, aciilo 14 Specific Ceieinovasculai DIsmdeis {escepl IIA) 437 Aillnoscleiosls, ceielnal 18 Hmispeclllc Ceiebiovasciilai DIsmdeis w/ CC 437' Allieiuscleiosls, ceiebial 14 S|ieclllc Ceieinovasculai Dismdeis(esce)4 IIA) 437.1 Disease, ceiebiovasciilai, Ischemic 18 Nmispeclllc Ceiebiovasculai Olsoideis w/ CC 437.3 Aiiemysm. naniiijiliiieil 14 Specific Ceieinovasculai Dismdeis (escepl IIA) 437.8 Disease, ceiebiovasciilai, ollie/ 18 Nmispeclllc Ceiebiovasculai Dismdeis w/ CC 437.9 Disease, ceieinovasculai, iinspecllled ia Nonspecific Ceiebiovasculai Dismdeis w/CC 438 443 ■ 4519 453.8 CVA w/lale elleci Heyiiaiiife Syitilnniiw l'IiiiH iH Is and linonilnniliellnlni, iiiisiiei.llled biles Einliolisin and lluonilmsls, ulliei simlclled silcs 12 24Ô iiii is Deyeiieiallve lleivoiis Syslem Dismileis Cmniei.llvu Dlsiniloie I'oi^iIh iiuI Vasbiil.ll ikseiibe w/ Cl: TiansienI Isciiemis Aliack 453.8 Embolism and llwombosis, olliei S|ielciled sites 130 Peiiplieial Vasciilai Disease w/ CC 453.9 Emlnjlisin and llnomlwsls, uns|ielciled silvs 15 1 laiiskml Ihcliemlc Allack 453.9 Emlxilism and llwombosis. inis|iek.ilcd siles 130 Peiiplieial Vascuku Disease w/CC 454* Vailcose veins, lowei eidiemlllas ww/o iHceis 130 Peil|ilieial Vasciilai Disease w/ CC 457.1 Edema, lyiii|ilieilciiia 283 Mnnn Skin Dismdeis 459.9 Disease, cl cubdmy, uiispecllied (nickidntg edema) 132 Aillieioscleiosis w/ CC 485.9 Inlecllon, uppei lesplialuiy. aciile, inispecllled sile 08 Oimis Media and URI 488 Oioncliills, seule 98 Dionclikls and Asllniia 480' PneiiiiKM ila, vlial 88 Sknpla PneiNiimila w/ CC 401' Pneumonia, imeumococcal 89 Skiiple Pneunimila w/ CC 482 Pneumonia, Oacleilal, due lo Klebsiella pneumoniae 79 Respkalmy Inlecllmis w/ CC 482.1' Pneiwnonia, bacleilal, pseudomonas 89 Simple Pneummila w/ CC 482.3' Pneumonia, bacleilal, SliejHococciis 89 Simple Pneumonia w/ CC 482.8' Pneumonia, mganlsm specified. 89 Simple Pneummila w/ CC 482.9' Pneumonia, bacleilal, iinspecllled 89 Simple Pneummila w/ CC 485 OioncliopiieiHnoiila, mganlsm unspecified 89 Simple Pnetniionia w/o CC 488 Pneumonia, mganlsm iinspecllled 89 Simple Pneummila w/o CC 487 Inlliieiua w/piicummila 89 SImiite Pimummila w/o CC 491.2' DioiKliHIa, clumilc, olisliucllve 08 COPD 492.' Empliysema 80 COPD 493' Aslluna, clwonic 88 COPD 493.9 Asllwiia, unspecllled 98 Bimiclillls and Asllnna 498 CORD 88 COPD 507' Aspkalkm Pneumonbi 79 Respkalmy Inlaclknis w/ CC 513 Abscess, lung and medlasllnum 79 Respkalmy InlecHons w/ CC 514 Congeslkm, kmg, hyposlello 87 Pleinal Elluslon w/o CC d: C D ■ D O Q . C g Q . ■ D C D C /) C /) 8 ci' 3 3 " C D C D ■ D O Q . C a O 3 " O O C D Q . ■ D C D C /) C /) ICD-9’s ICOü Coiiveisioli Tables Descilption/lnlu 5T7.8 518 5Î8.4 Disease, king. In olliei diseases classilled elsewlieie Collapse, laiknonaiy Acute edeina ol liwig, leispecllied 518.5 Pulinonaiy kisiilllclency, lollowlng liaiinia, siiigciy 518.81 Respbqaloiy laSias 5 1882 Olliei pulinonaiy kisulllclency fiiol classilied elsewlieie 518.89 Disease, odiei, liiiig 519.9 584 585 Disease, lesplialoiy, iinspecilicd iïenal lalkee, acnie, unspecllleil Rénal lalkae, cliionic 588* Renal lakuie, leispecllled 881* Cetulills and abscess, loe, llngei 682* Ceaulills and abscess, otiiei llian loe oi lingot 707* Ukei, skIn, cleonlc 710* 7Î42 ................ 7Î5* ■ 7Î6i4* Disease, cotaiecllve tissue, dilinse (Pnlyiiiynsilis) RlictiitialuldAilliillls (Tslcoaillaosis and allied disonleis Ailliiopalliy, olliet oi iinspecllled, Ailluilis 718.5* Ailleopalliy, olliei m iinspecilied, Ailliiills 718.8* Ailleopalliy, olliei oi unspecllled, Ailluilis 718.8* 718,9* n 7 .7 Ailliiopalliy, olliei oi unspecllled, Aillinlis Ailliiopalliy, olliei oi unspecllled, Aitliiilis, ImihI Cliiomlioiiialacla, palella 717.9 DeiangenienI, knee, unspecllled 718.2* 718 3^ 71831 Dlsoidei, cailUage DIsotdei, |oinl locallon iinspecilieil DIslocallon, teciatenl, |olnl, slioiiklei tegiun 718.35 DIslocallon, tecutteni, peMs 718.4* 718.85 Conliaclwe, joinl (IncliM leiq liaml, aiin, 1E ) Jokil kislsUkly, knee, kiwci leg 718.88 Joint InslabMIy, peMs, llilgli 718.87 Jokil kislabmiy, ankle, lool 718.9* OetangeinenI, K H M 718.95 Jolnl InslaUMy, knee, lowet leg 719.0* Elluslon, )olnl 719,4 l'ain, jolnl 719.5 Silliness, jolnl, olliet 719.8 Syinploni, olliei, iclenati<e lo joliil 719.7 719.8 Disoidet, jokil, watilng dllllcu«y Calcilicallon, jokil, unspecllled 719.8 DIsotdei, olliei, unspecllled, jolnl 719.9 DIsonlei, unspecllled, jolnl 720 SponddyMIs ankylosing 721* Spondylosis and alked disotdeis 92 ïüî Û L L 87 07 ÜCG SoMed by ICD9 U«»cil|illuii litleislilMl Imifl I>ise»s8 Olliei Rtt|iùJloiy Syileiii Oiafliiotes nieiiial Elliislui w/o CC Plewal Elliiskm w/o CC 70 iôr i ô i ÎÎ8 i i ô ] 3 Î6 277 2 7 7 ' Olliei n etpualoiy Syslem Ptopnote» Deiial raHui« Reliai railme Ruiiairailine CetuMIa 2 7 1 ^ 240 2 4 0 ' 244 240 240f 246 248 248 253 253 258 2 5 6 ' 253 256 2 5 6 ' 2 5 8 ' 258 256 25^ 258 258 24r 247 247 U ]_ 2 4 7 ^ 247 247 24^ 243 PleiHol Ellinloii w/o CC Reaiilialuiy liilucliona w/ CC Olliei R etiiéaloiy S yaleiii Oiafliiotea CelliiW la Skin Ulceia Comiecllve DIsoideia CwNicLlIvu DiM eilcia Houe filsoases and Specilic Aillwopallile:_____ Honspecillc Ailliiopallilea___________________ Nonspecilic Ailluopalliles _ _ _____________ Nonspecilic Aillwopalliles__________________ Nonspecilic AillMopalliles Nonspecilic Aillaopalliles ■ ■ ■ — # ■ ■ — . ■ — ^ 1 ■ > II» - ■ I M M ■ — M ■ 1 I > > ■ - ■ — - - — - - f lacliwes, Siaakis, DIslocallons, HE. LE, escepl liand & lool_____ Fiacliees, Spialns, OIslocallons, UE, lE, escepl liand Â looi__________ Olliei Miisciiloslielelal Sya & Connecllve Tissue Diagnoses Odiei Miisciilusiiclelal Sys 8 Connecllve Hssiie niapnoses Fiacliaes, Siealiis. OIslocallons. UE, lE. escepl liand 8 lod Qlliei Miisciiloslielelal Sys 8 Connecllve Tissue Diagnoses Olliei Miisciiloslielelal Sys i Connecllve Tisane Olapiioses Ulliei Mnsciiluskiilelal Sys & Connecilye Tissue Dlapiioses Olliei Mnsculoskelelal Sys 8 Connecllve 7Issue Diagnoses Olliei Musciiloskelelal Sys & Connective Tissue Diagnoses Ôiliei Muscukiskelelal Sys & Connecllve Tissue Diagnoses Olliei Mnsculoskelelal Sys 8 Connecllve lissne Diagnoses Olliei M nsculoskelelal Sys 8 Connective 3Issue Diagnoses SIqiis end Syiiip Mnsculoskelelal Sys 8 CoiMictlIve I Issno Signa and Syiiip. Musculoskelelal Sys 8 Connecilye I Issue Signs and Syinp Muscukiskelelal Sys & Connecllve Tissus Signs and Syinp Mnsculoskelelal Sys & Connecllve 11ssue Signa and Syiiip Musculoskelelal Sys 5 Connecllve Tissue Signs and Syinp Musculoskelelal Sys ^ CoieiecllvB Tissue Signs aiM l Sy inp Musculoskelelal Sys S Connecllve Tissue Connecllve Disonleis Medical Back PioUeins 147 cl .21 « I ill O a, s l i i C l n i i»'si I I I w ! = a sis I I î l H «i 1 1 Si S m » . 'P " » ' W 'O V «a iS i <0 vwivm'v.&mnoiaiw : CJI r t I r ï . «n i « I o • o • o % :w V w 3 m # « C N r N i C H e n laisimivimiv V « A 'V I * A " S la «viv * WB C. >:a 8 : | C O l U o a lain v r» a alkiaisio d mimimimaïaia ^ icw iioia; aiaïaïai a e b id |d >d - d > d 0 ^ I 0 ^ 0 ! ■ 0 ^ Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 il#! C ) u i s. 5 s S 5 5 W « n i f » » ( o i o o t e o i o i o t 0 t < ^ i o i o t c e t o i c o < m > a ' o ey ( V w * W i C N #v: V W l v n ë *r y h » > p»». (S < t o I h » t (>9 1 <n t »«> m i c m i m ICO i w i t A i < o eo io io i(n ((O icD itA te5 > (S io to > to c n i C Q I S i 'S I I I 2 ! 2 O i > 0 3 c o f^ (o » * a |m îÛ 3 iQ 3 O ieiOtO O jO tO ld co ia> m IO Q » I* c y io 4 ifn i^ iiA lc o lh » O B iO t' « O ( M c N ir y lc y l« > C D I C D > e o Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149 î '■s • S co n i I : h = • “ > i I J D 3 = -ï <sii " 9 ~ Cl u c • <n i r t if«-1 IV 'V IA ir n ia rn I f '* IV C M 1 ^ 11 ! I I V « l Vf V I , e > a i<0i v l V ' O ■ V ex I V I I 1 3 l ! i l ' i i i ilïîliâi l: Ji- - J i I I I n a = m ê 11 M l î i f * il i t l î l ï ( Jf! I 5 1 ï P i il s 5 a I ri O A 'A A 'Â « A S'A:2)3lÂlÂir* A ' A ■ A * A I A I A ' A l O i A ^ p*. «X im im A ï o • A >ex i c x l e x A IA i A A A A I A «A I A ' A « C > ' 0 ' A » A l A i a » » A ' A A « A « Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IMAGE EVALUATION TEST TARGET ( Q A - 3 ) A y V . / % 1 . 0 l.l 1.25 Lâ |Z8 Uâ B M 2.0 i m 1-4 I I I 1.6 150mm V « P . y / / o / '/j. /APPLIED A irVMGE . Inc 1653 East Main Street Roctiester. NY 14609 USA Phone: 716/482-0300 . ^ ^ ^ 5 Fax: 716/288-5989 0 1993. Applied image. Inc.. AU Rights Reserved Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Assessment of prognostic comorbidity in hospital outcomes research: Is there a role for outpatient pharmacy data?

PDF

An analysis of health risk selection and quality of care under Medicare fee -for -service and Medicare managed care health care systems

PDF

Business and medicine at the crossroads: An ethical assessment of for-profit managed care.

PDF

A randomized controlled trial of provider independent interventions to increase mammography screening among HMO members who are overdue for a mammogram

PDF

A comparative analysis of management control in public and private health care administrative organizations

PDF

Bringing social worlds together: Information systems as catalysts for new interactions in health care organizations

PDF

Evidence of age-based rationing of health care to the elderly: The effect of age and death on the use of health care services by elderly enrollees in an HMO

PDF

Care management for the uninsured: A force field analysis of the business case

PDF

Balancing community benefit and economic survival in nonprofit hospitals: A comparative perspective on charitability and tax-exemption policies in competitive health care markets in Pennsylvania ...

PDF

Effects of a formulary expansion of the use of SSRIs and health care services by depressed patients in the California Medicaid program

PDF

Futile medical treatment: an analysis of physician decision-making

PDF

Age-related blunting of growth hormone secretion during exercise is not due solely to increased somatostatin tone

PDF

A comparative study of costs and quality between case managed and non-case managed hospitals

PDF

Clinical capital equipment acquisition: A study of executive decision-making

PDF

Client characteristics, service characteristics, and psychosocial outcomes in the community treatment of schizophrenia

PDF

Alternate models of women's health care policy in the United States

PDF

A critical analysis of the society comedies of Henry Churchill De Mille and their contribution to the American theater

PDF

Evaluation of the Care Advocate Program: Bridging managed care and home community -based services

PDF

Assessing the cost implications of combined pharmacotherapy in the long term management of asthma: Theory and application of methods to control selection bias

PDF

Context is everything: An empirical study of institutions, instability and economic growth

Asset Metadata

Creator Angelelli, Joseph James (author)

Core Title An analysis of postacute treatment and outcome differences between Medicare fee-for-service and managed care

Degree Doctor of Philosophy

Degree Program Gerontology and Public Policy

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag Gerontology,health sciences, health care management,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor [illegible] (committee chair), [illegible] (committee member), DeJong, Fred (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-373300

Unique identifier UC11353845

Identifier 9919008.pdf (filename),usctheses-c17-373300 (legacy record id)

Legacy Identifier 9919008.pdf

Dmrecord 373300

Document Type Dissertation

Rights Angelelli, Joseph James

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA