Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Chronic eye disease epidemiology in the multiethnic ophthalmology cohorts of California study
(USC Thesis Other)
Chronic eye disease epidemiology in the multiethnic ophthalmology cohorts of California study
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Chronic Eye Disease Epidemiology in the Multiethnic Ophthalmology Cohorts of California Study by Dominic Joseph Grisafe II A Dissertation Presented to the FACULTY OF THE USC KECK SCHOOL OF MEDICINE UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (EPIDEMIOLOGY) August 2020 Copyright 2020 Dominic Joseph Grisafe ii Dedication For my sister, Kathryn Grisafe. 1992–2020 iii Acknowledgements I would like to acknowledge the scientific contributions and mentorship of my committee members including Wendy J. Mack, PhD for instruction of epidemiological and statistical methods; Lourdes A. Baezconde-Garbanati, PhD for insight into racial/ethnic disparities and community health; Dr. Benjamin Xu, MD, PhD for guidance in ophthalmologic disease processes and treatments; Cecilia Patino-Sutton, MD, PhD for education of clinical trials, quality of life, and epidemiological concepts; and my advisor Roberta McKean-Cowdin, PhD for the many hours discussing everything from project details to academic development. Many others were essential in developing scientific ideas and implementing analyses including Meredith Franklin, PhD; Bruce Burkemper, PhD; Grace Richter, MD; Darryl Nousome, PhD; Malcolm Barrett, MPH; Farzana Choudhury, PhD; Xuejuan Jiang, PhD; Mina Torres, MS; and Rohit Varma, MD. I would also like to thank Eileen M. Crimmins, PhD for supporting me through her predoctoral training grant, Multidisciplinary Research Training in Gerontology (NIA T32 AG000037) from the NIH National Institute on Aging. Thank you to Mary Trujillo, Sherri Fagan, Renee Stanley, Linda Hall, and Roland Rapanot for assisting with administrative requirements. The collection of data was an enormous undertaking by several people listed above as well as Alicia Fairbrother-Crisp, MPH; Tengiz Adamashvili; Carlos Lastra, MD; Elizabeth Corona; YuPing Wang; M. Roy Wilson, MD; Julia A. Haller, MD; Helen Hazuda, PhD; Eve J. Higginbotham, MD; Joanne Katz, ScD; Maryann Redford, DDS; Xinzhi Zhang, MD, PhD; Jacqueline Douglass; Jaimie Barrera; Judith Linton; Kisha Milo; and many others. This work was supported by several grants including the NIH National Eye Institute Grant U10 EY-023575 and an unrestricted grant from the Research to Prevent Blindness, New York, New York. All mistakes are my own. Code for analyses will be available upon journal publication at https://github.com/dgrisafe/Dissertation. iv Table of Contents Dedication ........................................................................................................................................ ii Acknowledgements ......................................................................................................................... iii Table of Contents ............................................................................................................................ iv List of Tables ................................................................................................................................... vi List of Figures................................................................................................................................ viii Abbreviations .................................................................................................................................. xi Abstract ......................................................................................................................................... xiv Chapter 1: Introduction ................................................................................................................... 1 Paper 1: Impact of Visual Field Loss on Vision-Specific Quality of Life in the African American Eye Disease Study ..................................................................................................... 5 Paper 2: Variability of Visual Field Loss on Vision-Specific Quality of Life in the Multiethnic Ophthalmology Cohorts of California Study ............................................................................... 7 Paper 3: The Association of Traffic-Related Air Pollution with Reduced Blood Perfusion through Peripapillary Capillaries of the Retina in African Americans ......................................... 9 Background ............................................................................................................................... 11 Chapter 2: Impact of Visual Field Loss on Vision-Specific Quality of Life in the African American Eye Disease Study ....................................................................................................... 27 Abstract...................................................................................................................................... 27 v Introduction ................................................................................................................................ 29 Methods ..................................................................................................................................... 31 Results ....................................................................................................................................... 36 Discussion ................................................................................................................................. 54 Chapter 3: Variability of Visual Field Loss on Vision-Specific Quality of Life in the Multiethnic Ophthalmology Cohorts of California Study............................................................... 57 Abstract...................................................................................................................................... 57 Introduction ................................................................................................................................ 60 Methods ..................................................................................................................................... 62 Results ....................................................................................................................................... 68 Discussion ................................................................................................................................. 99 Chapter 4: The Association of Traffic-Related Air Pollution with Reduced Blood Perfusion through Peripapillary Capillaries of the Retina in African Americans ........................................ 108 Abstract.................................................................................................................................... 108 Introduction .............................................................................................................................. 111 Methods ................................................................................................................................... 114 Results ..................................................................................................................................... 120 Discussion ............................................................................................................................... 145 References .................................................................................................................................. 149 vi List of Tables Table 1: Crude prevalence of Visual Field Loss and Open Angle Glaucoma in Population- Based Racial-Ethnic Cohorts ........................................................................................................ 19 Table 2: Definitions of open angle glaucoma diagnosis in MOCCAS population-based studies ........................................................................................................................................... 21 Table 3: Sociodemographic differences between the analytic cohort and AFEDS participants excluded from analysis ............................................................................................. 38 Table 4: Sociodemographic and Clinical Characteristics of Participants in AFEDS ................... 40 Table 5: Sociodemographic and Clinical Characteristics of Participants in AFEDS by VFL severity categories ........................................................................................................................ 41 Table 6: Sociodemographic differences between the VFL severity cohort and AFEDS excluded participants .................................................................................................................... 43 Table 7: Item Response Theory Diagnostics for Task and Well-being Composite Models........ 44 Table 8: Linear Regression β Coefficients for the Association Between VSQOL and VFL in the BSE in AFEDS ........................................................................................................................ 48 Table 9: ANCOVA Assessing the Relationship Between VSQOL and Various Categories of Severity of VFL in the AFEDS ...................................................................................................... 51 Table 10: VSQOL survey instrument: NEI-VFQ-25 items, CTT subscales, and IRT domains ... 65 Table 11: Sociodemographic and Clinical Characteristics of Participants in MOCCaS (n = 14,570)*,† ....................................................................................................................................... 71 Table 12: Linear Regression for VSQOL IRT Composite Scores on VFL in the BSE in the MOCCaS (n = 14,570)*................................................................................................................. 77 Table 13: Linear Regression for the Association of VSQOL on VFL in the BSE among Latinos in MOCCaS ...................................................................................................................... 80 vii Table 14: Linear Regression for the Association of VSQOL on VFL in the BSE among Chinese Americans in MOCCaS .................................................................................................. 81 Table 15: Linear Regression for the Association of VSQOL on VFL in the BSE among African Americans in MOCCaS .................................................................................................... 82 Table 16: Linear Regression for the Association Between CTT Driving Difficulties and VFL in the BSE in MOCCaS* ............................................................................................................... 93 Table 17: Comparing Sociodemographics of Included and Excluded Adults ≥ 40 Years in the AFEDS .................................................................................................................................. 122 Table 18: Sociodemographic and Clinical Variables of 1,009 Adults ≥ 40 Years in the AFEDS ........................................................................................................................................ 124 Table 19: Regression of VAD on NO2 Exposure the Week before the Exam for 1,009 Adults ≥ 40 Years in the AFEDS ............................................................................................................ 142 Table 20: Regression of VAD on PM2.5 Exposure 6 Months before the Exam for 1,009 Adults ≥ 40 Years in the AFEDS ................................................................................................ 144 viii List of Figures Figure 1: Prevalence of open-angle glaucoma (OAG) among US eye-care recipients................ 2 Figure 2: Study populations for the Multiethnic Ophthalmology Cohorts of California Study..... 12 Figure 3: Country of birth for participants of the Multiethnic Ophthalmology Cohorts of California Study ............................................................................................................................. 13 Figure 4: Conceptual Model for Vision-Specific Quality of Life ................................................... 15 Figure 5: Visual field loss categorization by severity and laterality ............................................. 18 Figure 6: Flow diagram for AFEDS .............................................................................................. 37 Figure 7: LOWESS plot of predicted VSQOL IRT composite scores from linear regression on VFL in the BSE ........................................................................................................................ 46 Figure 8: LOWESS plot of predicted VSQOL CTT driving difficulties from linear regression on VFL in the BSE|| ....................................................................................................................... 47 Figure 9: Comparing VSQOL Effect Sizes in AFEDS Participants in each VFL Severity Category and those with No VFL.................................................................................................. 53 Figure 10: Flow Diagram for MOCCaS ........................................................................................ 69 Figure 11: Item response theory diagnostics and factor analysis ............................................... 73 Figure 12: LOWESS plot of predicted NEI-VFQ-25 IRT composite scores from linear regression on VFL in the BSE ...................................................................................................... 78 Figure 13: Linear regression beta coefficients of NEI-VFQ-25 IRT and CTT on VFL in the BSE by cohort ............................................................................................................................... 83 Figure 14: LOWESS plot of predicted NEI-VFQ-25 CTT role function scores from regression on VFL in the BSE ...................................................................................................... 85 Figure 15: LOWESS plot of predicted NEI-VFQ-25 CTT mental health scores from regression on VFL in the BSE ...................................................................................................... 86 ix Figure 16: LOWESS plot of predicted NEI-VFQ-25 CTT dependency scores from regression on VFL in the BSE ...................................................................................................... 88 Figure 17: LOWESS plot of predicted NEI-VFQ-25 CTT near vision scores from regression on VFL in the BSE ........................................................................................................................ 89 Figure 18: LOWESS plot of predicted NEI-VFQ-25 CTT peripheral vision scores from regression on VFL in the BSE ...................................................................................................... 90 Figure 19: Linear regression beta coefficients of all QOL scales on VFL in the BSE by cohort ............................................................................................................................................ 91 Figure 20: LOWESS plot of predicted NEI-VFQ-25 CTT driving difficulties scores from regression on VFL in the BSE ...................................................................................................... 94 Figure 21: Linear regression of NEI-VFQ-25 CTT driving difficulties on VFL in the BSE by age group ...................................................................................................................................... 95 Figure 22: Regression of VSQOL IRT composite domains on VFL in the BSE excluding visual impairment* ......................................................................................................................... 97 Figure 23: Regression of VSQOL IRT composite domains on VFL in the BSE excluding self-reported depression*.............................................................................................................. 98 Figure 24: Images of the Retina and Retinal Blood Vessels..................................................... 113 Figure 25: Map of air pollution monitors for NO2 and PM2.5 and participants from the AFEDS ........................................................................................................................................ 117 Figure 26: Flow Diagram for OCTA imaging in AFEDS ............................................................ 121 Figure 27: Univariate linear regression for the association of vessel area density and predictors .................................................................................................................................... 126 Figure 28: LOWESS Plots of Covariates with Significant Univariate Associations with Vessel Area Density.................................................................................................................... 127 Figure 29: Daily NO2 and PM2.5 averaged from 2014–2018 in Los Angeles County ................ 129 x Figure 30: Temporal Trends in daily maximum NO2 and daily mean PM2.5 for Downtown Los Angeles Monitor ................................................................................................................... 130 Figure 31: Generalized Additive Model of NO2 and PM2.5 from 2014–2019 in Los Angeles County ......................................................................................................................................... 131 Figure 32: Observed and Predicted Air Pollution Exposures from 2014–2018 at Monitors in Los Angeles County .................................................................................................................... 132 Figure 33: Conceptual models of the relationship of RBF and TRAP for hierarchical linear models ......................................................................................................................................... 134 Figure 34: Multivariable hierarchical linear regression for VAD on TRAP of increasing exposure duration ....................................................................................................................... 137 Figure 35: LOWESS plot of predicted VAD from regression of TRAP exposure in the AFEDS ........................................................................................................................................ 140 xi Abbreviations 12-Item Short Form Survey (SF-12) Activities of Daily Living (ADL) Activities of Daily Vision Scale (ADVS) Advanced Glaucoma Intervention Study (AGIS) African American Eye Disease Study (AFEDS) Age-Related Macular Degeneration (AMD) Angle Closure Glaucoma (ACG) Assessment of Disability Related to Vision (ADREV) Best Corrected Visual Acuity (BCVA) Better Seeing Eye (BSE) Body Mass Index (BMI) Chinese American Eye Study (CHES) Classical Test Theory (CTT) Compressed Assessment of Ability Related to Vision (CAARV) Directed Acyclic Graphs (DAGs) Early Treatment of Diabetic Retinopathy Study (ETDRS) Effect Size (ES) False Negative (FN) False Positive (FP) xii Fixation Losses (FL) Generalized Additive Models (GAMs) Glare Sensitivity (GS) Glaucoma Symptom Scale (GSS) Glycated Hemoglobin (HbA1c) Instrumental Activities of Daily Living (IADLs) Integrated Visual Field (IVF) International Classification of Diseases, 9th Revision (ICD-9) Item Response Theory (IRT) Locally Weighted Scatterplot Smoothing (LOWESS) Los Angeles Latino Eye Study (LALES) Mini-Mental State Examination (MMSE) Modified Glaucoma Symptom Scale (MGGS) Multiple Imputation with Chained Equations (MICE) National Eye Institute Visual Functioning Questionnaire-25 (NEI-VFQ-25) Nitrogen Dioxide (NO2) Non-Hispanic White (NHW) Open-Angle Glaucoma (OAG) Optic Nerve Head (ONH) Optical Coherence Tomography Angiography (OCTA) Particulate Matter with Aerodynamic Diameter Less Than 2.5 µm (PM2.5) xiii Parts Per Billion (ppb) Pattern Standard Deviation (PSD) Pattern Visual Evoked Potentials (PVEP) Primary Open-Angle Glaucoma (POAG) Radial Peripapillary Capillaries (RPCs) Retinal Nerve Fiber Layer (RNFL) Short Portable Mental Status Questionnaire (SPMSQ) Socioeconomic Status (SES) Spaeth/Richman Contrast Sensitivity (SPARCS) Spectral Domain Optical Coherence Tomography Angiography (SD-OCTA) Stereoacuity (SA) Systolic Blood Pressure (SBP) Traffic Related Air Pollution (TRAP) Vessel Area Density (VAD) Vision Specific Quality of Life (VSQOL) Visual Acuity (VA) Visual Angle (VAng) Visual Field Loss (VFL) Visual Impairment (VI) Worse Seeing Eye (WSE) xiv Abstract Chronic diseases of the eye are becoming more prevalent as the average age of the population continues to climb. Most non-acute eye diseases present in adults aged 40 years or older. Epidemiologic studies over the last several decades have observed that certain eye pathologies are more common among certain racial/ethnic groups. For example, glaucoma is more prevalent in Latinos and African Americans, age-related macular degeneration is more common in non-Hispanic Whites, and diabetic retinopathy is more prevalent in Chinese Americans. How people are impacted by chronic eye diseases might also be related to race/ethnicity. Furthermore, degree of visual function—not just prevalence—may prove to be more informative and precise measures for characterizing disparities in eye disease. A functional perspective of vision focuses on the experiences of individuals suffering from chronic eye diseases, and not on diagnoses of diseases made by panels of expert physicians. The impact of eye pathologies has been shown to be largely mediated through vision loss. For example, a patient with early stage glaucoma who has not suffered loss of vision—visual impairment (VI)—is likely not impaired by their disease. VI may be defined by numerous measures of vision, and all of them should be characterized to understand how chronic eye disease affects patients. Visual acuity (VA) is a metric of central, high acuity vision that is measured using an illuminated chart the patient reads aloud to the healthcare provider. VA of 20/40 or worse is a standard definition of VI that is commonly used in both the clinic and research. However, there are other, less familiar measures of VI. Visual field loss (VFL) measures the entire visual field— both central and peripheral vision—and can be used to locate specific patterns of vision loss in a patient’s visual field. However, a total score can also be used to provide a global assessment of visual ability separately for each eye. VFL is quantified on a logarithmic scale as the xv difference in vision from a standard population of healthy adults of the same age. One definition of VI using VFL is -2 decibels (dB) in the better-seeing eye (BSE) of total mean deviation (MD) from an age-adjusted standard population. Racial/ethnic disparities in chronic eye disease should be assessed using both VFL and VA as well as other measures of vision such as contrast sensitivity. Considering continuous measures of visual function improves our understanding of the extent of vision loss for each racial/ethnic group as well as the dimensions of vision that are most affected. For example, population-based cohort studies have demonstrated that VI defined by VFL is more common in racial/ethnic minorities compared to non-Hispanic Whites. To see the full picture of ophthalmic illness, however, we must look beyond objective measures of visual impairment. Modern research investigates patients’ perspectives of their quality of life (QOL) to fully understand the impact of disease and to design effective treatments. Health-related quality of life (HRQOL) is a latent construct with numerous definitions, but overall is used to measure the value of life for several domains including the ability to perform tasks and socioemotional well-being. HRQOL has been shown to vary among racial/ethnic groups for cancer patients after adjusting for demographic factors and socioeconomic status (SES). Cultural differences in beliefs of how health is related to illness and death have been identified as explanations for racial/ethnic differences in HRQOL among terminal patients. Race/ethnicity is related to how people find meaning during the end of life. Might similar differences exist in how people find value related to their vision when suffering from chronic eye disease? Vision-specific QOL (VSQOL) has been employed by ophthalmologists and investigators for several decades to capture the value of life related to one’s visual ability. VSQOL is another quantitative measure of visual function—like VA and VFL—but focuses on patients’ subjective experiences. Because chronic eye diseases are predominantly disabling only when they cause xvi VI, the association of VI and VSQOL is essential to understanding how chronic eye disease affects patients. VSQOL has been used as an outcome variable in clinical trials to assess whether ophthalmologic interventions are effective in treating chronic eye diseases. In addition, VSQOL has also been used at the population level to assess the perception of vision loss and how it impacts daily life in the community. This information is useful for prioritizing public health resources to improve domains of visual health that are most meaningful for multiethnic groups. Of course, this is only relevant if race/ethnicity modifies the relationship between visual impairment and VSQOL. Although VSQOL is a subset of QOL, differences have been identified in the literature. Only VSQOL, but not HRQOL, appears to be affected by vision loss. VSQOL is specific to vision but not to health in general, which may explain why broader measures of QOL are not sensitive to VI. Furthermore, VSQOL is measured using instruments that were developed and validated for multiethnic populations. Given that VSQOL is a more specific domain than HRQOL and has demonstrated construct validity in diverse populations, do cultural beliefs and other racial/ethnic factors modify the association between patients’ subjective experiences and quantitative measures of visual impairment? Variation in the prevalence of chronic eye diseases exist by race/ethnicity. Do people experiencing the same quantitative loss in vision have similar magnitudes and patterns of reduced QOL specific to their vision, regardless of their race/ethnicity? VSQOL has been associated with VA, VFL, and other continuous measures of visual impairment among non-Hispanic Whites, Latinos, and populations of African descent outside the US. But no studies have assessed these associations in large groups of African Americans or cross-cultural studies of the most populous racial/ethnic groups in the United States (US). The primary aim of the first two papers in this dissertation are to elucidate how VSQOL may be differentially impacted by VFL for major US racial/ethnic groups. In the first paper we validate xvii modern psychometric measures of VSQOL in the largest, population-based cohort of chronic eye disease ever conducted in African Americans. In this study, we assess whether clinically meaningful differences in VSQOL are associated with similar magnitudes and patterns of VFL in non-Hispanic White and Latino populations. The second paper is a pooled analysis concerned with harmonizing methods to assess whether similar associations between VFL and VSQOL exist for Latinos, Chinese Americans, and African Americans. The second study concludes our assessment of whether differences exist in how vision loss impacts VSQOL by race/ethnicity. From here we move on to investigate why differences in chronic eye diseases exist in the first place, and whether air pollution might be an important explanatory exposure. Recall that Latinos and African Americans have a higher prevalence of open angle glaucoma (OAG), which is the leading cause of irreversible blindness globally. However, the scientific community has not identified why this disparity exists. Several genetic studies have found evidence that a limited number of genetic markers may increase the risk of OAG among participants with African Ancestry. But small effects do not fully explain the relatively large racial/ethnic difference. In the third paper of this dissertation we identify a modest association between traffic-related air pollution (TRAP) and reduced blood flow to retinal vessels implicated in OAG in a cohort of African Americans. The pathophysiology of OAG involves the death of retinal ganglion fiber cells, possibly due to deleterious effects of the retinal microvasculature surrounding the optic nerve head (ONH). TRAP has been shown to cause damage to microvasculature in other areas of the body, leading to cardiovascular disease, stroke, vascular Alzheimer’s disease and overall mortality. Also, exposure to TRAP has been shown to be greater for impoverished people and people of color in the US. Optical coherence tomography angiography (OCTA) is a noninvasive imaging methodology that was used to measure retinal blood flow in the vessels surrounding the ONH. These findings complement recent studies of large hospital biobanks that identified xviii TRAP measured at patient addresses was associated with changes in the thickness of retinal nervous tissue associated. We propose that the next population-based study of eye disease should implement OCTA to more precisely assess TRAP exposure and retinal blood flow in a larger cohort of participants. This dissertation characterizes multiethnic disparities in domains of chronic eye disease epidemiology including subjective experience related to vision, objective visual function, anatomical measurements of the eye and environmental exposures. Study participants are from the Multiethnic Ophthalmology Cohorts of California Study (MOCCaS), which is composed of population-based cohorts of Latino, Chinese American, and African American participants living in Los Angeles County. Technical abstracts are available later in this dissertation at the beginning of each chapter. The first paper proposes and assesses the psychometric validity of a VSQOL instrument in a large cohort of African Americans for the first time. Associations are identified between patterns of VFL and VSQOL related to tasks and socioemotional wellbeing. The second paper identifies racial/ethnic differences in VFL and VSQOL across all three racial/ethnic groups using harmonized methods. The final paper investigates a burgeoning hypothesis that TRAP is associated with OAG as measured by blood perfusion to vessels in the eye. 1 Chapter 1: Introduction Eye conditions and diseases are expected to increase in the next 3 decades as the US and global populations age. The total economic burden of vision loss and eye disorders in the US in 2013 was estimated to be $139 billion.1 In 2015, the number of US adults aged 40 years and older that were visually impaired reached 3.22 million and the number who were blind was 1.02 million; by 2050 these numbers are expected to more than double.2 But not all individuals are affected equally. In the US older adults, women, Latinos, and African Americans endured a disproportionate amount of visual impairment.2–4 For example, open-angle glaucoma (OAG) had a greater prevalence and incidence in African American patients (Figure 1).5 Research is vital for understanding how vision loss and eye pathologies affect physical health, mental health, and quality of life (QOL) among major US subpopulations. Once these disparities are better understood, efforts to reduce visual impairment should prioritize specific domains of health that are most impacted within each group. 2 Figure 1: Prevalence of open-angle glaucoma (OAG) among US eye-care recipients A: Prevalence of OAG B: Incidence of OAG* Participants were 40 years and older who used eye services between Jan 1, 2001 and Dec 31, 2007. US eye-service data was collected retrospectively from the i3 InVision Data Mart database (Ingenix, Eden Prairie, MN).5 *Hazard models were adjusted for age, sex, US region of residence, education, household net worth, diabetes mellitus, hypertension, hyperlipidemia, sleep apnea, migraine, cataract, pseudophakia or aphakia, diabetic retinopathy, macular degeneration, and overall health Visual field loss (VFL) is the deterioration of sight in the visual field, including both central and peripheral vision. Visual impairment encompasses both VFL and visual acuity (VA), a measure of central vision, both of which have been associated with decreased vision-specific QOL (VSQOL). However, most investigations of visual impairment and VSQOL focused on VA.6–10 Existing population-based cohorts studying VFL found associations with greater falls,11– 13 hip fractures,14 and ankle fractures15 as well as decreased driving16–19 and worse VSQOL.20–25 Investigations linked to DMV registries demonstrated the association of VFL with less driving at night26 and that motor vehicle accidents were twice as common among individuals with binocular VFL.27 Hospital and clinic-based studies found VFL was related to reduced physical activity28 and less travel from home,29,30 as well as further supported the associations of VFL with more frequent falls,31,32 increased hip fractures,33 and lower VSQOL.34–41 But clinical results may not represent the impact VFL has on VSQOL in the broader US population. Worse VFL was associated with lower VSQOL in a population-based sample of Latinos21–23 and in multicultural populations including a relatively small number of African Americans.20,24 However, 3 no population-based studies have explored how VFL impacts self-reported VSQOL differentially across major US racial and ethnic groups. It would be reasonable to hypothesize that a loss in VFL would impact VSQOL for all people. But characteristics specific to racial/ethnic groups likely modify the magnitudes of effect on different domains of VSQOL. Completing vision-specific tasks and socioemotional well-being are the main dimensions of VSQOL,42 a patient reported outcome (PRO) measured by validated surveys.43–45 Previous publications have found vision-specific PROs are affected by variations in socioeconomic status,46 depression,47 and personality.48 Many studies of VSQOL and VFL have adjusted for these variables, which likely modify the relationship between VFL and VSQOL. However, country of origin, acculturation measures, race and ethnicity may serve as useful proxies for cultural differences that also affect this relationship yet defy quantification. To date, no population-based studies have included large numbers of individuals from major US racial/ethnic groups to explore the potential differential impact of VFL on perceived VSQOL. However international studies of middle aged and older adults have investigated pathologies of diseases that lead to VFL. VFL is associated with various disease states in different racial/ethnic populations. OAG was the most common cause of VFL observed in a non-Hispanic White population. Incident VFL was 7.4 (95% CL 6.3-8.5) per 1000 person-years in a population-based study of Netherlanders aged 55 years and older.49 Among the 240 eyes with VFL, the most common etiologies of VFL were open-angle glaucoma (24.2%), stroke (13.8%), and age-related macular degeneration (11.7%). In a population-based study of rural and urban Chinese living near Beijing aged 40 years and older, incidence of VFL was much greater compared to non-Hispanic Whites, possibly due to less access to cataract extraction surgery. Incident VFL was 15.3 (95% CL 13.2- 17.1) per 1000 person-years.50 Among the 235 eyes with VFL, the most common diseases and conditions were cataracts (24.9%), glaucoma (8.4%), and diabetic retinopathy (4.8%). However, 4 prevalence and incidence of VFL are difficult to compare across multi-ethnic populations due to differences in diagnosis criteria and definitions of visual field loss. Furthermore, the associations of VFL with VSQOL in US populations should be explored in groups with a common disease state such as OAG. 5 Paper 1: Impact of Visual Field Loss on Vision-Specific Quality of Life in the African American Eye Disease Study The primary aim of the first project is to characterize the association between VFL and VSQOL in a nationally-representative, population-based cohort of African Americans aged 40 years and older.51 I hypothesize that in this cross-sectional study, VFL in the better- (BSE) and worse seeing eyes (WSE) will be associated with decreased task and well-being domains of VSQOL. A similar analysis has been conducted in the Los Angeles Latino Eye Study,21 which has been improved upon in the present study by imputing missing covariates using multiple imputation with chained equations,52 excluding unreliable VFL measurements,53 and implementing more robust models of VSQOL.42 VSQOL was measured using the National Eye Institute Visual Functioning Questionnaire- 25 (NEI-VFQ-25) instrument, which was analyzed by two separate scoring methods. Item response theory (IRT) modeling was used to produce two unidimensional domains of VSQOL that better satisfied statistical assumptions: vision-related task and well-being.54 Classical test theory (CTT) scoring was also included to provide a more granular perspective of VSQOL; for example, the 13 items of the NEI-VFQ-25 in the IRT task domain were organized into separate subscales for color vision, distance vision, driving difficulties, near vision, peripheral vision, and vision-related role function. VSQOL outcomes were regressed on VFL in the BSE and WSE using multivariable linear regression adjusting for clinical and sociodemographic covariates. However, considering VFL in one eye may not accurately represent how VFL affects individuals who normally see while using both eyes. Participants with different patterns of VFL in each eye may have deficits in different VSQOL domains, such as driving difficulties, vision-related task, and vision specific well-being. I hypothesized that individuals with bilateral VFL would have the strongest relationship with all domains of VSQOL. Analysis of covariance was used to assess 6 least squared mean VSQOL in participants with various VFL categories compared to those with no VFL. Aims 1. To characterize any statistical and clinically meaningful association between VFL and VSQOL in a population-based cohort of African Americans 2. To elucidate how domains of VSQOL differ by laterality and severity of VFL 3. To propose loadings and assess the validity of IRT in NEI-VFQ-25 Hypotheses 1. There exists a statistically significant and clinically meaningful (≥ 5-point) inverse association between VFL and VSQOL for both task and well-being domains in African Americans 2. The previous association will be greatest among participants with bilateral, severe VFL 3. Task and well-being are two valid, unidimensional measures of VSQOL measured with the NEI-VFQ-25 and assessed using IRT. The task domain includes items relating to driving difficulties, distance vision, peripheral vision, near vision, vision-related role function, and color vision. The well-being domain includes items relating to mental health, general vision, social function, dependency, and ocular pain. 7 Paper 2: Variability of Visual Field Loss on Vision-Specific Quality of Life in the Multiethnic Ophthalmology Cohorts of California Study The aim of the second projects is to assess the association of VFL with VSQOL across population-based cohorts of African Americans, Chinese Americans, and Latinos. Various eye diseases are more prevalent for major US racial/ethnic groups. Limited studies have investigated how communities may be impacted differently by visual impairment resulting from chronic eye diseases. Furthermore, methods have varied across existing analyses, which have occurred internationally over two decades.20,21,24,25 A direct comparison across major US subpopulations will enable better understanding of which VSQOL domains are most impacted by VFL for each race/ethnicity. If VSQOL domains are differentially affected by VFL for each racial/ethnic group, a better understanding of these domains might allow public health efforts to prioritize interventions to address the greatest need within each community. If the data illustrate driving difficulties are a greater issue for African Americans with VFL, but mental health is more important for Chinese Americans, for example, then appropriate community interventions should target these domains. If domains are similarly impacted across all cohorts, then the impact of VFL on VSQOL may be independent of race/ethnicity, and all public health efforts may be similar for all communities. QOL has been shown to differentially impact older adults. The oldest old can have greater well-being scores and lower physical, task scores compared to younger adults.55 Assessing whether a similar interaction exists for adults with respect to VSQOL has not been addressed within the context of chronic eye disease. Assessing how domains of VSQOL may be modified by age is an important step in deepening an understanding of how visual impairment impacts older adults. 8 Aims 1. To investigate any statistical and clinically meaningful associations between VFL and VSQOL, and how these may vary by racial/ethnic group in a population-based, pooled- analysis of major US racial/ethnic groups. 2. To identify subscales of VSQOL most impacted by VFL across racial/ethnic groups 3. To elucidate whether associations of VFL and VSQOL vary by age Hypothesis 1. There exists a statistically significant and clinically meaningful (≥ 5-point) inverse association between VFL and VSQOL for both task and well-being domains, which varies in magnitude for the most populous US racial/ethnic groups 2. Driving difficulties related to vision are most impact by VFL for all racial/ethnic groups 3. Among participants ≥ 65 years old, VFL has larger associations with task domains and smaller with well-being domains related to vision 9 Paper 3: The Association of Traffic-Related Air Pollution with Reduced Blood Perfusion through Peripapillary Capillaries of the Retina in African Americans This cross-sectional study assesses whether retinal blood perfusion—as measured by optical coherence tomography angiography (OCTA), vessel area density (VAD)—is negatively associated with increased exposure to traffic-related air pollution (TRAP). Air pollution has been associated with vascular disease including stroke, heart attack, and vascular Alzheimer’s Disease. The retina is a highly vascular organ, which has not been appreciated in environmental studies of air pollution and chronic disease. The eye may be an important window into assessing the relationship between air pollution exposure and the microvasculature throughout the body. This is a two-part analysis. First, spatiotemporal generalized additive models were used to prediction NO2 and PM2.5 air pollution exposure over time periods of increasing duration prior to the OCTA eye exam. The second part of the study was an epidemiological evaluation of the relationship of VAD and each air pollutant in a population of African Americans living in Inglewood, California. African Americans have historically been disproportionately impacted by environmental injustices, making this cohort a poignant population for assessing health impacts of air pollutions. There has been no consensus on why African Americans endure a greater prevalence of glaucoma compared to non-Hispanic Whites. This study begins to address whether air pollution may be a unique exposure that could eventually explain this disparity. 10 Aims 1. How does TRAP—NO2 and PM2.5—impact the retinal microvasculature of the peripapillary area in African Americans? 2. To assess whether any association between TRAP and retinal blood flow differs for acute and chronic TRAP 3. To understand whether TRAP varies in exposure for participants living less than 10 km apart. Hypotheses 1. There is a significant, inverse association of TRAP and Vessel Area Density (VAD) of the radial peripapillary capillaries (RPC) in African Americans. 2. Both acute and chronic TRAP exposures are inversely associated with VAD as measured by OCTA. 3. TRAP exposure can be predicted for Los Angeles County using a spatiotemporal model with an R2 ≥ 60%. 11 Background Study Populations AFEDS, CHES, and LALES are population-based cross-sectional cohorts in Los Angeles County. AFEDS included 6,347 subjects aged 40 years and older residing in 32 US census tracts in Inglewood, California; CHES included 4,582 subjects 50 years and older from 10 US census tracts in Monterey Park, California; LALES included 6,357 subjects 40 years and older from 6 US census tracts in La Puente, California (Figure 2). Data were collected from interviews and comprehensive clinical eye examinations between 2014 and 2018 in AFEDS, 2010 and 2013 in CHES, and 2000 and 2003 in LALES. Data collection in multi-ethnic population-based cohort studies of eye disease occurred in Inglewood (African American Eye Disease Study), Monterey Park (Chinese American Eye Study), and La Puente, California (Los Angeles Latino Eye Study). Detailed descriptions of cohort populations are available for AFEDS,51 CHES,56 and LALES.57 12 Figure 2: Study populations for the Multiethnic Ophthalmology Cohorts of California Study Data collection in multi-ethnic population-based cohort studies of eye disease occurred between 2000 and 2018 in Inglewood (African American Eye Disease Study), Monterey Park (Chinese American Eye Study), and La Puente, California (Los Angeles Latino Eye Study). In brief, eligible residents were identified by door-to-door census and invited to participate in an interview conducted in their home followed by a comprehensive clinical eye examination by study ophthalmologists at the local eye clinic. Home interviewing was conducted after informed consent to gather demographic factors, history of ocular and medical conditions, and access to ocular and medical care. A comorbidity score was calculated as a summation of twelve self- reported medical conditions.58–60 Participants who completed the in-home questionnaire also completed an additional brief interview at the local clinic before the eye examination, which included information on quality of life and visual function. The University of Southern California Medical Center Institutional Review Board approval was obtained. All study procedures adhered to the recommendations of the Declaration of Helsinki. 13 Figure 3: Country of birth for participants of the Multiethnic Ophthalmology Cohorts of California Study Data collection for multi-ethnic population-based cohort studies of eye disease occurred between 2000 and 2018 in Inglewood (African American Eye Disease Study), Monterey Park (Chinese American Eye Study), and La Puente, California (Los Angeles Latino Eye Study). Acculturation Many study participants immigrated from outside the US (Figure 3). People who moved to the country later in life may have different cultural perspectives towards vision and health in general.61 People may also have varying ability to access health care resources and different familial support systems. The cultural and demographic variations may affect the impact that VFL has on VSQOL. In an attempt to capture some of these differences, we include country of origin in statistical models. Acculturation is measured by validated surveys that assess primary 14 language spoken and adherence to cultural traditions as well as preferences for foods, music, and film. Acculturation was measured using instruments for Chinese Americans in CHES,62 and for Mexican Americans in LALES.63 There was no measure of acculturation for African Americans in AFEDS, who were largely born in the US. 15 Figure 4: Conceptual Model for Vision-Specific Quality of Life ↑ indicates increase in the level of the variable. ↓ indicates decrease in the level of the variable. + indicates presence of a factor. – indicates absence of a factor. ‡ Data on these variables are not collected. Conceptual Model Certain variables likely confound the association between VFL and VSQOL. Eye diseases have been shown to impact VSQOL almost entirely through mediation of visual impairment.64 Risk factors for developing eye disease include psychosocial attributes, personal health factors, and health care access and utilization. Sociodemographics likely affect all of the above factors. A conceptual model of the relationship between potential predictors of eye disease, visual impairment and VSQOL was developed for MOCCAS with input by the external advisory panel for the NIH funded cohorts included in this analysis (Figure 4). Variables in this model that were considered of potential relevance but not measured in all 3 population-based cohorts are indicated in red. The directions of associations are indicated where relevant. 16 Visual Acuity Visual acuity (VA) is a measure of central, high acuity vision. One definition of visual impairment (VI) is VA of 20/40 or worse, which has been associated with many adverse outcomes. VI is associated with increased mortality,65,66 but eye diseases including open-angle glaucoma do not appear to be a risk factor independent of VI.67 VI is also associated with increased complaints of memory and confusion in adults over 60 years old.68 Finally, VI is strongly associated with VSQOL.6 The procedure to assess distance presenting visual acuity (VA) has been described previously.69 Vision was measured for each eye with presenting correction at 4 meters (m) when needed using standard Early Treatment Diabetic Retinopathy protocols with a modified distance chart trans-illuminated with a chart illuminator (Precision Vision).70 Presenting distance visual acuity (VA) was measured for both left and right eyes with existing refractive correction. If the participant could not read 55 letters at 4 m in either eye (equivalent to Snellen fraction 20/20), an automated refraction was performed, using the Humphrey Automatic Refractor (Carl Zeiss Meditec, Dublin, CA), followed by subjective refraction. After refraction, the eye was retested to measure the best-corrected VA (BCVA). If the participant was unable to read 20 letters at 4 m (equivalent to Snellen fraction 20/100), measurement and subjective refraction were done at 1m. Presenting VA was calculated as the sum of all letters read correctly and converted to a logarithm of the minimum angle of resolution (LogMAR) score. Visual acuity loss was defined as presenting visual acuity of 20/40 or worse based on the U.S. definition of visual impairment (VI). 17 Visual Field Loss Visual field (VF) is a measure of the entire visual field, which extends laterally from the point of fixation, or central vision, and is measured for a given visual angle (VAng) in degrees. VAng depends on both the size of the object being viewed and its distance from the viewer. A VAng of 24º from the fixation of central gaze is frequently tested, although wider ranges are possible.71 15 different perimetric machines have been developed to measure VFL, of which the Humphrey Field Analyzer (HFA) (Carl Zeiss Dublin, CA) is used most often in research and clinical practice.72 The Standard Swedish Interactive Thresholding Algorithm (SITA) algorithm takes 3-7 minutes to complete per eye and shows a visual stimulus in one of 54 points in the visual field. The patient is instructed to stare straight ahead without shifting their gaze and click a trigger each time they detect a stimulus. Sensitivity scores are automatically generated as the total deviation from the standard age-adjusted population in the ability to detect light of a given intensity at each point. Mean deviation (MD) is a weighted average of the total deviation of each point and is reported in decibels (dB).73 VFL severity was quantified as a continuous and categorical variable (defined as -2 dB of MD or worse). Smaller—more negative—scores indicate greater VFL. Continuous VFL was assessed as MD stratified by the better- and worse-seeing eye. VFL severity categories were based on previous studies (Figure 5).21,74 VFL was categorized into severity levels of magnitude (moderate/severe VFL < -6 dB; -6 dB > mild VFL > -2 dB; no VFL > -2 dB) and laterality (VFL in both, one, or neither eye). 18 Figure 5: Visual field loss categorization by severity and laterality VFL was categorized into severity levels of magnitude (moderate/severe VFL < -6 dB; -6 dB > mild VFL > -2 dB; no VFL > -2 dB) and laterality (VFL in both, one, and neither eye). The prevalence of VFL in the US (Table 1) has been defined as -2 dB of VFL or worse, and was reported as 25% for Latinos.21 By this same definition, preliminary analysis of AFEDS and CHES has found 30% and 33% VFL prevalence for African Americans and Chinese Americans, respectively. However VFL definitions have varied in other population-based studies, and prevalence has not been reported in any standardize way. These analyses will be able to establish better estimates of VFL prevalence in the US. 19 Table 1: Crude prevalence of Visual Field Loss and Open Angle Glaucoma in Population-Based Racial-Ethnic Cohorts Study Race or Ethnicity Prevalence of VFL (MD < -2 dB) in the BSE Prevalence of Definite OAG CHES* Chinese American 33% (1,463 of 4,582) *** AFEDS** African American 30% (1,310 of 6,347) *** LALES** Latino 25% (1,305 of 5,213)21 4.7% (291 of 6,142)73 BES** African American 4.2% (100 of 2,395)74 BES ** Non-Hispanic White 1.1% (32 of 2,913)74 Visual Field Loss (VFL), Mean Deviation (MD), Better Seeing Eye (BSE), Open-Angle Glaucoma (OAG), Chinese American Eye Study (CHES), African American Eye Disease Study (AFEDS), Los Angeles Latino Eye Study (LALES), Baltimore Eye Study (BES) *Participants are 50 years and older **Participants are 40 years and older ***Glaucoma patients are yet to be graded for the Chinese American Eye Study and the African American Eye Disease Study Monocular vs Binocular Visual Field Measurement Unilateral VFL in the BSE has been shown to be as strong an indicator of vision specific quality of life (VSQOL) as integrated and binocular VFL in population-based and clinical studies.75,76 Therefore VFL in the BSE is the primary exposure in this study. VFL in the WSE has shown weaker associations with VSQOL and was therefore analyzed but considered a secondary exposure.35 Test Reliability VF reliability assesses the quality of the VF measurements. Some participants may be anxious, and repeatedly click the trigger even when they are not detecting stimuli; this is reported as false positives (FP), and is the most important index of VF reliability.53 False negatives (FN) measure inattentive patients by showing the brightest possible stimulus in a field point that was previously determined to be visible to the patient. However FN are more common in patients with glaucoma, and usually are not considered when excluding unreliable 20 measurements.73 Fixation losses (FL) assess whether the patient maintained their gaze straight ahead. FL are counted by positive sensitivity detections in the physiological blind spot.77 If a participant indicates they are detecting stimuli in the blind spot, it is assumed they have shifter their gaze and are using central vision instead of peripheral vision. However, excluding fixation losses has been shown to have little impact on MD,53 and is not used as a reliability measure in glaucoma and related diseases. VF measurements were considered unreliable if they had more than 15% false negatives, or 15% false positives. This was determined after discussions with ophthalmologists, although the literature reports various cutoffs for these measurements.75,78,79 If a measurement was considered unreliable the visual field test was repeated up to 2 times. Open Angle Glaucoma Open angle glaucoma (OAG) is the most common cause of both irreversible blindness and incident VFL in a population with access to cataract surgical resources.49 For this reason, OAG is presented here as an example from which to assess the context of VFL. The 10-year cumulative incidence of OAG has been reported from 2.2% to 4.4% in non-Hispanic White and African-descent populations.80 The prevalence of OAG In the US (Table 1) was 1% for non- Hispanic Whites and 5% for Latino populations.81 However, there is no standard diagnosis of OAG, making OAG and glaucomatous VFL difficult to compare across public health studies.82 A description of OAG diagnosis guidelines used in population-based MOCCAS cohort studies are reported in Table 2. A consensus diagnosis from 2 of 3 ophthalmologists is usually used to diagnoses OAG in population-based studies.56 Repeated perimetry is the most accurate method of assessing the progression of OAG, although it is often supported by imaging of the optic nerve head.73 21 Table 2: Definitions of open angle glaucoma diagnosis in MOCCAS population-based studies 8.3.4 Primary Open-angle Glaucoma (POAG) Owing to the heterogeneity of definitions previously employed in population-based research into glaucoma, we will adopt two definitions: the ISGEO “Standardized” scheme, and the expert panel approach Definite OAG requires: • The presence of an open angle, congruent, characteristic or compatible glaucomatous visual field abnormality; and • Evidence of characteristic or compatible glaucomatous optic disc damage in at least one eye after ophthalmologic exclusion of other possible causes. Probable OAG requires the presence of an open angle and one of the following in at least one eye: • A congruent characteristic or compatible glaucomatous visual field abnormality and the absence of optic disc data; or • Characteristic or compatible glaucomatous optic disc damage and the absence of visual field data; or • A combination of visual field and optic disc abnormalities that are characteristic of or compatible with glaucoma in the opinion of the glaucoma specialist. Other glaucoma will include those individuals who have evidence of definite or probable glaucoma and who have evidence of secondary glaucoma such as neovascular glaucoma or traumatic glaucoma. Ocular hypertension will include those individuals with intraocular pressure > 21 mmHg, either horizontal or vertical cup/disc ratio ~ 0.7, and the absence of optic disc damage or abnormal visual field test results. For the expert panel approach, a two-step process will be used to determine the diagnosis of OAG. First, two glaucoma specialists will evaluate all clinical history, including any history of glaucoma, treatment of glaucoma, family history of glaucoma; history of treatment and management for other ocular diseases, including cataract, diabetic retinopathy, and age- related maculopathy; and examination data, including visual acuity, Van Herrick test results, gonioscopy results, evaluation of the anterior and posterior segments of the eye, clinical optic disc evaluation, clinical fundus evaluation, optic disc photographs, and visual fields. Second, the two glaucoma specialists will determine the presence or absence of OAG using guidelines specified below. The two specialists will grade both optic disc photographs and visual fields, independently and masked, to the gradings of the other glaucoma specialist. In determining the diagnosis of glaucoma, the specialists will classify each eye of each person with particular consideration to the Humphrey visual field test results and evaluation of the optic disc photographs. If the two glaucoma specialists agree on the diagnosis, that diagnosis will be assigned to each specific eye of each person. Third, in the event of disagreement, a third glaucoma specialist will assess the data; an agreement between two of the three glaucoma specialists will be used to assign the diagnosis for each eye of each participant. Additionally, the Principal Investigator will perform a confirmatory review of all cases of OAG (Figure E-6). • 8. Specific Criteria for Diagnosis of Open-Angle Glaucoma, used by the expert panel: Definite Open Angle Glaucoma: 1. Open angle, two or more reliable, congruent abnormal visual field tests (Humphrey C24 SITA standard and/or full threshold C 24-2) and optic disc damage both characteristic of glaucoma. 2. Open angle, one or more abnormal visual field tests (Humphrey C 24 SITA standard and/or Humphrey full threshold C 24-2), and optic disc damage that are either characteristic of or compatible with glaucoma. 22 OAG is a chronic progressive eye disease. Some forms of OAG are thought to be caused by a blockage of aqueous humor drainage from the eye, leading to increased intraocular pressure within the eye, and then structural changes to the optic nerve. The increased pressure damages retinal axonal nerve fibers, which transmit signals from light waves to the brain. Peripheral nerves are damaged first, which leads to characteristic patterns of visual field loss and optic nerve changes known as “cupping” which are visible on eye examination using an ophthalmoscope, a common, non-invasive clinical assessment. OAG and glaucomatous VFL have been associated with many adverse events, including depression83 and worse VSQOL.84,85 Decreased VSQOL has been demonstrated in other types of glaucoma as well, but OAG has been associated with lower deficits in vision-related social function and other specific VSQOL domains.85 Further population-based studies should aim to clarify how VFL impacts VSQOL in those with and without OAG. How much VFL is associated with disability in glaucoma patients? Moderate VFL has been previously considered between -6 dB and -12 dB,74 but glaucoma patients appear to be disabled even at this level. OAG patients with disabilities completing vision-specific tasks measured by the 25 Item National Eye Institute Visual Function Questionnaire (NEI-VFQ-25) had an average of -6.0 dB of MD of VFL in the BSE, compared to -2.5 dB non-disabled patients.79 It appears that more modest deficits in VFL are associated with clinically meaningful disability. 23 Quality of Life Quality of life (QOL) is the “degree to which persons perceive themselves able to function physically, emotionally, mentally, and socially”.86 Health-related QOL (HRQOL) is a subset of QOL, and vision-specific QOL (VSQOL) is a subset of HRQOL. Two validated survey instruments were used to assess HRQOL and VSQOL in population-based cohorts. Health-Related Quality of Life HRQOL was measured using the Medical Outcomes Study 12-item Short-Form Health Survey (SF-12, version 1).87 This was a measure of QOL related to overall health, and provides separate component scores for physical and mental HRQOL. Standard US norm-based SF-12 Physical Component Summary (PCS) and Mental Component Summary (MCS) scores were calculated using data from the SF-12, so that a score of 50 (standard deviation of 10) was the average among adults in the US.88 Larger scores indicate greater HRQOL. VFL has been shown to have minimal impact on HRQOL.21 Vision-Specific Quality of Life VSQOL was measured using the National Eye Institute Vision Functioning Questionnaire-25 (NEI-VFQ-25).89,90 The NEI-VFQ-25 has traditionally been analyzed using classical test theory (CTT).21 Item response theory (IRT) has been the preferred analysis method in recent ophthalmic literature.91,92 We performed both CTT and IRT to utilize the benefits of each. VSQOL Research has Improved Medicine Studies of VSQOL have justified the medical cost that insurance companies incur in funding cataract surgeries in the second eye.93–96 Health economists weigh the cost of an intervention by measuring quality-adjusted life years (QALYs)—the increase in number of years lived with complete health.93 Cost of cataract surgery in the second eye was $2,727/QALY, which was 24 considered affordable relative to common medical interventions and was further justified by the large increase in VSQOL that patients gained.97 The EAGLE study is a randomized clinical trial that used VSQOL as a co-primary outcome to assess the clinical impact and cost-effectiveness of different treatments for primary angle- closure glaucoma—clear lens replacement versus laser peripheral iridotomy (LPI).98 Compared to LPI at 3 years after surgery, clear lens replacement was found to have greater HRQOL, improved cost-effectiveness, lower intraocular pressure (IOP), and greater VSQOL. Self- reported HRQOL was measured by the European Quality of Life-5 Dimensions (EQ-5D), which scores the extent of problems—none, some, or extreme—for 5 domains including mobility, self- care, usual activity, pain or discomfort, and anxiety or depression.99,100 Cost effectiveness was assessed by QALYs, which were calculated from longitudinal ED-5D scores. VSQOL was assessed using the 25-item National Eye Institute Visual Functioning Questionnaire (NEI-VFQ- 25),89,90 and the Glaucoma Utility Index (GUI).101 Compared to LPI, the clear lens replacement group had a statistically significant NEI-VFQ-25 score that was more than 5-points in magnitude, which is considered a clinically meaningful difference.10 Furthermore the clear-lens replacement group required less IOP and glaucoma medications, but there was no difference in VFL between the two arms. The EAGLE study demonstrates the utility of PROs: HRQOL and VSQOL can be clinically-meaningfully different among treatment groups even when measures of visual impairment—VFL and VA—are not. VSQOL has Potential to Personalize Medicine The broadest goal of visual clinical practice is for people to reach old age in good health and without visual disability. VSQOL brings an intangible human element to this overarching priority of vision care. Quality of life may vary by region, culture, and within the same person over repeated visits. Incorporating VSQOL into medicine has the potential to tailor care to each patient, even as their attitudes, priorities, and needs change over multiple visits. VSQOL allows 25 individual priorities to be considered in clinical decision making, allowing a personalized approach to vision care. NEI-VFQ-25: Classical Test Theory CTT analysis of the NEI-VFQ-25 has been validated for various eye diseases in many populations.37,47,102–104 Using CTT will allow for comparisons with existing literature, particularly from the LALES.21–23 CTT subscales are also useful for identifying which domains of VSQOL have the strongest associations with VFL among study participants. The NEI-VFQ-25 was analyzed using CTT. Each item was scaled from 0 to 100, with 100 indicating the greatest VSQOL score. 25 items were grouped into 11 vision-specific subscales relating to well-being and task-oriented domains of VSQOL; a general health item assessed overall self-perceived health. Each subscale included 1 to 4 items. A mean was taken of all items in each subscale to produce a 0 to 100 subscale score. An overall composite score was calculated by taking the mean of all 11 vision-specific subscales (excluding general health). CTT relies on many assumptions.105 Less than 5% of the study population should be missing responses to any items; all item responses should be represented without a skewed distribution; items should not be unequally distributed towards the smallest (floor effect) or largest possible responses (ceiling effect); items should not correlated strongly (≤ 0.80) with each other; and items should correlate somewhat (≤ 0.30) with the overall composite score.91 CTT also assumes differences between item responses are equal, for example the difference between “none of the time” and “a bit of the time” is equivalent to the difference between “most of the time” and “all of the time”. Furthermore, in CTT scores are dependent upon each population because the abilities of the participants cannot be independently scored from the difficulty of the items; therefore results cannot be fairly compared for different groups administered the same CTT instrument. Finally, CTT assumes a constant standard error for all 26 ability levels, which is known to be false. For the violation of many of the above assumptions, an alternative scoring analysis is often used in recent publications. NEI-VFQ-25: Item Response Theory (Graded Response Model) IRT assumes participants each have an underlying latent trait value, VSQOL, and estimates the probabilities that individuals will respond to each item given their trait value. IRT models classify individuals with varying VSQOL scores along a linear continuum of NEI-VFQ-25 item difficulty. Less difficult items differentiate individuals with lower VSQOL, and more difficult items separate those with greater VSQOL.106,107 IRT analysis relies on more modest assumptions than CTT.105 Unlike CTT, IRT does not assume differences between item responses are equal or that standard errors are constant across all trait values. Furthermore, IRT also scores item difficulty independent from individuals’ abilities, allowing for direct comparisons across different populations. IRT analysis of the NEI- VFQ-25 also assumes a single underlying domain is being measured. Previous literature has shown the NEI-VFQ-25 is unidimensional when the instrument is separated into a functional task domain and a socioemotional well-being domain.42,108 The NEI-VFQ-25 was analyzed using the graded response model, a 2-parameter IRT model for ordinal items on a Likert scale.109,110 IRT was used to produce two unidimensional composites: task and well-being. The task composite score was calculated from 13 items of the vision-related role function, distance vision, driving difficulties, peripheral vision, near vision, and color vision subscales. The well-being composite was calculated from 12 items of the vision- related dependency, general vision, vision-related mental health, ocular pain, and vision-related social functioning subscales. General health was not included in either composite. Task and well-being composite scores were calculated using SAS PROC IRT, which has been shown to perform similarly to more common IRT software packages.111 27 Chapter 2: Impact of Visual Field Loss on Vision-Specific Quality of Life in the African American Eye Disease Study Abstract Purpose To examine the association between visual field loss (VFL) and self-reported vision-specific quality of life (VSQOL) in adult African Americans Design A cross-sectional, population-based cohort study. Participants African Americans 40 years and older living in 32 US census tracts in Inglewood, California Methods African Americans aged 40 years and older completed comprehensive interviews and ophthalmic examinations from 2014–18. VSQOL was quantified using the National Eye Institute Visual Function Questionnaire 25 (NEI-VFQ-25). VFL was assessed in both eyes using the Humphrey SITA Standard 24-2 test. VFL was scored as decibels (dB) of mean deviation (MD). Multivariable linear regression and analysis of covariance were used to determine the adjusted relationship between VSQOL and VFL. Main Outcome Measures VSQOL was measured for each participant using the NEI-VFQ-25, which was scored by item response theory (IRT). IRT yielded two domains including a functional task composite and a 28 socioemotional well-being composite. Classical test theory (CTT) was also used to evaluate NEI-VFQ-25 11 subscales with respect to existing literature. All scores were scaled from 0-100. Results 6,347 AFEDS participants had clinical eye exams, of which 4,187 completed reliable VFL measurements and QOL surveys. Worse VFL severity was significantly associated with older age, more comorbidities, unemployment, income ≤ $20,000, < high school education, self- reported depression, and visual acuity loss (p trend < 0.01). All IRT and CTT domains of the NEI-VFQ-25 had significantly lower mean, covariate-adjusted scores (p < 0.001). Each 1 dB MD of VFL in the BSE was associated with 0.69 (95% CI 0.59, 0.78) lower task and 0.61 (95% CI 0.51, 0.72) lower well-being composite scores. Of the CTT subscales, driving difficulties (β = 1.14, 95% CI 1.03, 1.25), mental health (β = 0.59, 95% CI 0.50, 0.68), and role function (β = 0.57, 95% CI 0.47, 0.67) had the strongest associations with VFL. Conclusions In this population-based sample of African American adults, a 7–8 dB loss of VF in the BSE was associated with a clinically meaningful (5-point) loss in self-reported ability to complete vision related tasks and emotional well-being. Driving difficulties may arise even earlier near 4 dB of VFL. 29 Introduction The number of people impacted by eye disease is expected to increase in the next 3 decades with the aging of the US population. The total economic burden of vision loss and eye disorders in the US in 2013 was an estimated $139 billion.1 In 2015, the number of US adults aged 40 years and older that were visually impaired reached 3.22 million and this number is expected to double by 2050.2 In the US older adults, women, and African Americans endure a disproportionate amount of visual impairment (VI).2–4 Research is necessary to characterize how VI affects physical health, mental health, and vision-specific quality of life (VSQOL) among major US subpopulations. While it is likely that all racial/ethnic groups experience a loss in quality of life with VI, the magnitude of the impact with increasing severity of VI has not been characterized, nor has the impact of vision loss on different aspects of daily activities and well- being. VI encompasses both visual acuity (VA) and visual field loss (VFL). Existing literature has found worse VFL is associated with less physical activity, walking, and traveling from home,29 and with more frequent falls,12,32 hip fractures,14 and automobile accidents.27 With respect to VSQOL, VFL has been associated with worse self-reported quality of life in a population-based sample of Latinos21–23 and in multicultural populations including African Americans.20,24,35–37 But no well-powered studies exist specifically to evaluate the question in African Americans. Many studies have found similar associations in glaucoma patients recruited from clinics,34–39,75,112 but these results may not represent the impact VFL has on VSQOL in the broader population. This is the first study to assess how VFL is associated with VSQOL in a large sample of African Americans. Specifically, we studied participants in the African American Eye Disease Study (AFEDS), a population-based cohort of adult African Americans 40 years and older residing in Inglewood, California, USA. We hypothesized that in AFEDS, the magnitude of association 30 between VFL and VSQOL may differ from those previously reported for other race/ethnicities due to perceived and real differences in sociodemographic or cultural experiences. 31 Methods AFEDS is a population-based, cross-sectional cohort of 6,347 subjects aged 40 years and older residing in 32 US census tracts in the city of Inglewood, California within Los Angeles County. Data were collected from interviews and comprehensive clinical eye examinations between 2014 and 2018. Demographics of AFEDS participants were similar to those of African American populations in Los Angeles County, California, and the United States.51 A detailed description of data collection methods have been published elsewhere,51 however a brief review of data collection and statistical methodology are provided below. The University of Southern California Medical Center Institutional Review Board approval was obtained. All study procedures adhered to the recommendations of the Declaration of Helsinki. Sociodemographic Assessment Eligible residents were identified by door-to-door census. Participants were interviewed in their homes and completed a comprehensive clinical eye examination by study ophthalmologists at the local eye clinic. Home interviewing was conducted after informed consent to gather demographic factors (age, sex, education, marital status, employment status, country of birth), history of ocular and medical conditions (e.g. history of diabetes, history of hypertension, comorbidities, history of eye diseases), and access to ocular and medical care (e.g. insurance, income, usual place of medical care, barriers to care, glasses, time since previous medical and eye care). A comorbidity score was calculated as the sum of twelve self- reported medical conditions.58–60 A combination of self-reporting, physical examination, and blood testing were used to determine hypertension and diabetes status. Information on quality of life and visual function was collected at the clinical eye exam. 32 Visual Acuity Assessment The procedure to assess distance presenting visual acuity (VA) has been described previously.69 Vision was measured for each eye with presenting correction at 4 meters (m) when needed using standard Early Treatment Diabetic Retinopathy protocols with a modified distance chart trans-illuminated with a chart illuminator (Precision Vision).70 Visual acuity loss was defined as presenting visual acuity of 20/40 or worse based on the U.S. definition of visual impairment (VI). Visual Field Loss and Visual Acuity Assessment Visual field testing evaluated each participant’s ability to detect objects in their peripheral vision. Visual fields for each eye were assessed using the Swedish Interactive Threshold Algorithm (SITA) Standard C24-2 test (Carl Zeiss Humphrey Field Analyzer II 750 Dublin, CA). VF was measured using mean deviation (MD) in decibels (dB).73 Measurements considered unreliable (≥ 15% false negatives or false positives) were omitted.53 VFL severity was quantified as a continuous (defined as -2 dB or more extreme) and categorical variable. Smaller (more negative) scores indicate greater VFL. Continuous VFL was assessed as MD stratified by the better- and worse-seeing eye. Unilateral VFL in the BSE has been shown to be as strong an indicator of vision specific quality of life (VSQOL) as integrated and binocular VFL,75,76 and therefore VFL in the BSE is the primary exposure in this study. VFL severity categories were based on previous studies.21,74 Participants were classified as having no VFL (MD > -2 dB in both eyes), unilateral mild VFL (-6 dB ≤ MD ≤ -2 dB in the worse eye, MD > -2 dB in the better eye), unilateral moderate to severe VFL (MD < -6 in the worse eye, MD > -2 dB in the better eye), bilateral mild VFL (-6 dB ≤ MD ≤ -2 dB in both eyes; or MD < -6 dB in the worse eye, -6 dB ≤ MD ≤ -2 in the better eye), and bilateral moderate to severe VFL (MD < -6 dB in both eyes). 33 Quality of Life The 12-Item Short Form Survey (SF-12) and the National Eye Institute Visual Functioning Questionnaire-25 (NEI-VFQ-25) were administered by a trained interviewer before the clinical examination. Health-related quality of life (HRQOL) was measured using the Medical Outcomes Study 12-item Short-Form Health Survey (SF-12, version 1).32 Standard US norm-based SF-12 Physical Component Summary (PCS) and Mental Component Summary (MCS) scores were calculated so that a score of 50 (standard deviation of 10) was the average among adults in the US.33 Larger scores indicate greater HRQOL. Vision specific quality of life (VSQOL) was measured using the National Eye Institute Vision Functioning Questionnaire-25 (NEI-VFQ- 25).89,90 NEI-VFQ-25: Classical Test Theory CTT analysis of the NEI-VFQ-25 has been validated for various eye diseases in many populations.37,47,102–104 CTT was completed in the current analysis to allow comparisons of AFEDS with existing literature, particularly from the LALES.21–23 Each item was scaled from 0 to 100, with 100 indicating the greatest VSQOL. 25 items were grouped into 11 vision-specific subscales relating to well-being and task-oriented domains of VSQOL. A mean was taken of all items in each subscale to produce a composite score. NEI-VFQ-25: Item Response Theory (Graded Response Model) A second analysis of VFL and NEI-VFQ-25 data was conducted using a graded response model, a 2-parameter IRT model for ordinal items on a Likert scale.109,110 IRT models classify people with varying VSQOL scores along a linear continuum of NEI-VFQ-25 item difficulty, where less difficult items differentiate participants with lower VSQOL, and more difficult items separate those with higher VSQOL.106,107 IRT was used to produce two unidimensional composites: task and well-being.42,108 The task composite score was calculated from 13 items of 34 the vision-related role function, distance vision, driving difficulties, peripheral vision, near vision, and color vision subscales. The well-being composite was calculated from 12 items of the vision-related dependency, general vision, vision-related mental health, ocular pain, and vision- related social functioning subscales. Task and well-being composite scores were calculated using SAS PROC IRT.111 Statistical Analysis Differences in continuous and categorical covariables were compared using Wilcoxon rank sum, Fisher’s exact tests and Bonferroni-adjusted Chi-squared tests. Continuous covariates were compared across severity categories of VFL using analysis of variance and are shown as means and standard deviations and evaluated using Tukey pairwise comparisons. Categorical variables are shown as frequencies and percentages and include: less than a college education (16 years) [yes/no], unemployed [yes/no], health insurance [yes/no], vision insurance [yes/no], VA loss of 20/40 or worse [yes/no], and depression in the last 4 weeks [yes/no]. Tests for trend were performed using the Wilcoxon rank sum test for continuous variables and the 2-sided, exact Cochran-Armitage test for categorical variables. Multivariable linear regression was used to assess VSQOL composite and subscale scores on VFL in the BSE and WSE. Covariates were selected that were previously identified in the literature as predictors of VSQOL or were considered potentially related to both VFL and perceived VSQOL. Models were adjusted for age, number of comorbidities, sex (female), education (< 4 years of college), working status (unemployed), income (≤ $20,000), has health insurance (yes), has vision insurance (yes), visual acuity loss (20/40 or worse), and depression (last 4 weeks). Missing covariates were imputed using multiple imputation with chained equations (MICE).47, 48 Locally weighted scatterplot smoothing (LOWESS) plots with 95% confidence limits were produced for predicted VSQOL outcomes. Beta coefficients of VFL in the BSE in the linear regression models were multiplied by 5 units of VSQOL to obtain the 35 corresponding change in VFL for each composite and subscale. A 5 unit change in the NEI- VFQ-25 was considered clinically meaningful, as it has been associated with a 2-line deficit in VA.10 Analysis of covariance (ANCOVA) was used to calculate adjusted mean scores of VFL categories based on laterality (unilateral or bilateral) and severity (mild or moderate/severe). Effect sizes (ES) were calculated as the difference in adjusted mean scores (between each severity level of VFL and no VFL) divided by the standard deviation for the no VFL group. ES from 0.20 to less than 0.50 were considered small effects, 0.50 to less than 0.8 were medium effects, and 0.80 or greater were large effects.113 All analyses were performed using SAS software 9.4 (SAS Institute, Cary, North Carolina, USA). All data visualization was produced using ggplot2 package (Hadley Wickham, Springer- Verlag New York) for R markdown. 36 Results Description of Study Cohort Of the 7,957 people identified as eligible, the final AFEDS cohort included 6,347 (80.0%) participants who completed the home questionnaire and clinical eye examination. The final analytic cohort for this analysis was composed of 5,153 participants (Figure 6) after excluding those who were missing NEI-VFQ-25 or SF-12 scores (n = 390), and had missing or unreliable VFL measurements in both eyes (n = 804). Analyses of the driving difficulties subscale were based on participants who reported they were currently driving or had driven in the past, which was 4,594 in the analytic cohort and 3,780 in the VFL severity categories with complete data. Participants in the analytic cohort (n = 5,153) were similar to AFEDS participants that were excluded from the analysis (Table 3). Relative to excluded participants, those in the analytic cohort had fewer comorbidities (2.4 versus 2.6) and were less likely to be unemployed (54% versus 61%), to earn less than $20,000 per year (29% versus 38%), and to have VA loss (8% versus 14%) (P < 0.001). The participants in the analytic cohort were also younger (mean age 60.8 versus 61.6 years, P = 0.02) and less likely to be female (62% vs 66%, P = 0.03). Differences in measured sociodemographic variables were small for income and VA loss (ES = 0.21) and negligible for comorbidities and unemployment (ES < 0.20).113 There were no statistically significant differences among the two groups in education, health insurance, vision insurance, or depression status. 37 Figure 6: Flow diagram for AFEDS AFEDS = African American Eye Disease Study; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; CTT = Classical Test Theory; SF-12 = 12-Item Short-Form Health Survey; VFL = Visual Field Loss ‖The NEI-VFQ-25 instructs participants to omit items related to driving difficulties if they report they are not currently driving and had not driven in the past. Participants with missing subscale scores for driving difficulties are included in the analytic cohort. Analysis of driving difficulties include 4,594 participants. §Participants with missing or unreliable visual field loss measurements in one or both eyes were excluded from analyses of visual field loss severity categories; these analyses were of n = 4,187 participants. Eligible (n = 7,957) Excluded • Not completing home questionnaire (n = 842) • Not completing comprehensive eye examination (n = 1,610) • Not completing clinic questionnaire (n =1,927) AFEDS Cohort (n = 6,347) Excluded Incomplete Outcome • Complete NEI-VFQ-25 o Classical Test Theory Subscales Composite (n = 190) Color vision (n = 216) Dependency (n = 276) Distance vision (n = 204) Driving difficulties‖ (n = 1,001) General health (n = 183) General vision (n = 191) Mental health (n = 191) Near vision (n = 198) Ocular pain (n = 193) Peripheral vision (n = 214) Role function (n = 265) Social function (n = 208) • Complete SF-12 o Mental component score (n = 278) o Physical component score (n = 278) Excluded Incomplete and Unreliable Exposure§ • Missing visual field measurements (n = 694) • Unreliable visual field measurements o False positive error rates > 15% (n = 179) o False negative error rates > 15% (n = 216) Analytic Cohort (n = 5,153) 38 Table 3: Sociodemographic differences between the analytic cohort and AFEDS participants excluded from analysis Sociodemographic and Clinical Characteristics Analytic Cohort* (n = 5,153) Excluded (n = 1,194) P-Value† Effect Size‡ Age, mean years [SD] 60.77 (11.04) 61.61 (12.49) 0.021 0.07 Comorbidities, mean [SD]§ 2.35 (1.91) 2.62 (2.14) < 0.001 0.13 Female sex 3213 (62.4%) 786 (65.8%) 0.026 0.07 Unemployed 2725 (54.4%) 703 (61.3%) < 0.001 0.14 Income ≤ $20,000 1187 (28.5%) 360 (38.3%) < 0.001 0.21 Education < 16 years 3292 (65.6%) 771 (67.6%) 0.226 0.04 Health Insurance: Yes 4625 (91.5%) 1047 (90.8%) 0.449 0.03 Vision Insurance: Yes 3347 (68.3%) 731 (66.3%) 0.211 0.04 Visual Acuity Loss: Yes‖ 400 (7.8%) 172 (14.4%) < 0.001 0.21 Depressed: Yes¶ 305 (5.9%) 70 (7.0%) 0.193 0.05 AFEDS = African American Eye Disease Study; SD = Standard Deviation; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; SF-12 = 12-Item Short-Form Health Survey *Only 5,153 (81.2%) of all AFEDS participants had reliable visual field measurements in at least one eye and were not missing NEI-VFQ-12 subscale or SF-12 component scores †Wilcoxon rank sum test was used to compare non-normal continuous variables. Fisher’s exact test was used to compare categorical variables. ‡Effect sizes were calculated as standardized mean differences between the analytic cohort and the excluded AFEDS participants by subtracting the means of the two groups and dividing by the overall standard deviation. §Number of self-reported comorbidities (diabetes, arthritis, stroke/brain hemorrhage, high blood pressure, angina, heart attack, heart failure, asthma, skin cancer, other cancer, back problems, hearing problems and other major health problems). ‖Visual Acuity Loss was defined as presenting visual acuity 20/40 or worse. ¶Depression was scored using the SF-12 item “Have you felt downhearted or blue a good bit of the time or more during the past 4 weeks?” Participants were considered depressed if they reported “A good bit of the time”, “Most of the time”, or “All of the time”. 39 Sociodemographic and clinical characteristics of AFEDS participants in the analytic cohort are available in Table 4 and stratified by VFL severity categories in Table 5. More than 95% of this group had complete sociodemographic information, except for 995 (19%) participants with missing income. Excluding missing data, this group had a mean age of 60.8 years (standard deviation [SD] 11.0 years) and had 2.4 self-reported comorbidities on average (SD 1.9). Most were women (62%), did not complete college (66%), had health insurance (92%), and had vision insurance (68%). Positive trends were found with increasing severity of VFL (P-trend < 0.01) for sociodemographic and clinical characteristics (older age, higher number of comorbidities, unemployment, income less than $20,000, VA loss, and depression). Having less than a college education [< 16 years] was also more likely with worsening VFL severity categories (P = 0.02). There were no significant trends with sex, health insurance or vision insurance. 40 Table 4: Sociodemographic and Clinical Characteristics of Participants in AFEDS Socio-demographic and Clinical Characteristics Analytic Cohort (n = 5,153)* Trend Test◊ Age, Mean Years (SD) 60.77 (11.0) < 0.001 Comorbidities, Mean (SD)§ 2.35 (1.91) < 0.001 Female Sex 3213 (62%) 0.546 Unemployed 2725 (54%) < 0.001 Income ≤ $20,000 1187 (29%) < 0.001 Education < 16 Years 3292 (66%) 0.016 Health Insurance: Yes 4625 (92%) 0.051 Vision Insurance: Yes 3347 (68%) 0.640 Visual Acuity Loss: Yes‖ 400 (8%) < 0.001 Depressed: Yes¶ 305 (6%) < 0.001 AFEDS = African American Eye Disease Study; SD = Standard Deviation; VFL = Visual Field Loss *Data are presented as mean (SD) for continuous variables (age and comorbidities); categorical variables are presented as frequency counts with percentages (%) of participants for each category of visual field loss (VFL) severity; percentages exclude participants with missing responses); the number missing is 142 (2.8%) for unemployment, 995 (19.3%) for income, 138 (2.7%) for education, 100 (1.9%) for insurance, 253 (4.9%) for vision insurance, and 7 (0.1%) for visual acuity loss. §Number of self-reported comorbidities (diabetes, arthritis, stroke/brain hemorrhage, high blood pressure, angina, heart attack, heart failure, asthma, skin cancer, other cancer, back problems, hearing problems and other major health problems). ‖Visual Acuity Loss was defined as presenting visual acuity 20/40 or worse. ¶Depression was scored using the SF-12 item “Have you felt downhearted or blue a good bit of the time or more during the past 4 weeks?” Participants were considered depressed if they reported “A good bit of the time”, “Most of the time”, or “All of the time”. ◊Test for trend was performed by Wilcoxon rank sum test for continuous variables, and 2- sided, exact Cochran-Armitage test for categorical variables. Table 5: Sociodemographic and Clinical Characteristics of Participants in AFEDS by VFL severity categories Sociodemographic and Clinical Characteristics Visual Field Loss Severity Categories (n = 4,187)† P-Value‡ No VFL (n = 2,356) Unilateral Bilateral Mild (n = 795) Moderate to Severe (n = 102) Mild (n = 775) Moderate to Severe (n = 159) Age, Years Mean (SD) 58.35 < 0.001 60.92 (10.86)b 63.29 (9.45)b,c 64.35 (12.00) c 67.04 (12.21)d < 0.001 Comorbidities§ Mean (SD) 2.12 < 0.001 2.43 (1.89)b 2.42 (1.94)a,b 2.55 (1.94)b 3.13 (2.21)c < 0.001 Female Sex 1455 0.046 502 (63%)a 58 (57%)a,b 499 (64%)a 78 (49%)b 0.546 Unemployed 1064 < 0.001 430 (56%)a,b 56 (56%)a,b 478 (63%)b 129 (84%)c < 0.001 Income ≤ $20,000 431 < 0.001 179 (28%)a 22 (26%)a 207 (33%)a,b 54 (45%)b < 0.001 Education < 16 Years 1449 0.512 514 (67%)a 59 (60%)a 506 (67%)a 109 (70%)a 0.016 Health Insurance: Yes 2103 1.000 715 (92%)a 91 (90%)a 705 (92%)a 149 (96%)a 0.051 Vision Insurance: Yes 1520 1.000 535 (71%)a 71 (72%)a 513 (69%)a 101 (67%)a 0.640 Visual Acuity Loss: Yes‖ 71 < 0.001 46 (6%)a 6 (6%)a,b 105 (14%)b 52 (33%)c < 0.001 Depressed: Yes¶ 19 0.152 21 (12%)a 21 (8%)a 26 (20%)a 47 (80%)a < 0.001 AFEDS = African American Eye Disease Study; SD = Standard Deviation; VFL = Visual Field Loss †Only 4,187 (81.3%) of the analytic cohort had reliable VF measurements in both eyes and could be categorized into VFL severity categories. Due to missing covariate data, not all VFL severity categories sum to the total (n = 4,187); the number missing is 107 (2.6%) for unemployment, 807 (19.3%) for income, 103 (2.5%) for education, 72 (1.7%) for insurance, 189 (4.5%) for vision insurance, and 3 (0.1%) for visual acuity loss. ‡P-values based on analysis of variance for continuous variables and χ2 tests for categorical variables (with Bonferroni-adjusted pairwise comparisons). ANOVA revealed significant differences across the visual field loss categories for all continuous variables. Bonferroni-adjusted χ2 tests revealed significant differences for unemployment, annual income, and visual acuity loss. 41 42 NEI-VFQ-25 Analyses We found the NEI-VFQ-25 had a positively skewed response distribution (Table 6). High measures of internal consistency and unidimensionality were observed for the IRT graded response models for both the task and well-being composites (Table 7). Cronbach’s alpha was 0.89 for task (13 items) and 0.81 for well-being (12 items) composites, demonstrating high inter- item correlation for each composite score. IRT task and well-being model latent traits were unidimensional; a single factor explained 70.9% of the variance for the task composite and 67.2% for the well-being composite. Test information curves (Table 7) revealed the NEI-VFQ-25 was most informative for task and well-being VSQOL scores below the mean. 43 Table 6: Sociodemographic differences between the VFL severity cohort and AFEDS excluded participants Sociodemographic and Clinical Characteristics VFL Severity Cohort* (n = 4,187) Excluded (n = 2,160) P-Value† Effect Sizes‡ Age, mean years (SD) 60.4 (10.90) 61.96 (12.05) < 0.001 0.14 Comorbidities, mean (SD)§ 2.3 (1.87) 2.6 (2.10) < 0.001 0.15 Female sex 2592 (61.9%) 1407 (65.1%) 0.007 0.07 Unemployed 2157 (52.9%) 1271 (61.2%) < 0.001 0.17 Income ≤ $20,000 893 (26.4%) 654 (38.0%) < 0.001 0.25 Education < 16 years 2637 (64.6%) 1426 (68.8%) 0.069 0.09 Health Insurance: Yes 3763 (91.4%) 1909 (91.3%) 0.330 0.01 Vision Insurance: Yes 2740 (68.5%) 1338 (66.8%) 0.308 0.04 Visual Acuity Loss: Yes‖ 280 (6.7%) 292 (13.6%) < 0.001 0.23 Depressed: Yes¶ 235 (5.6%) 140 (7.1%) 0.587 0.06 VFL = Visual Field Loss; AFEDS = African American Eye Disease Study; SD = Standard Deviation; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; SF-12 = 12-Item Short-Form Health Survey; MD = Mean Deviation *Only 4,187 (66.0%) of all AFEDS participants had reliable visual field measurements in both eyes, therefore could be categorized into VFL severity categories, and were not missing NEI-VFQ-12 subscale or SF-12 component scores. VFL was classified as none (MD > −2), unilateral mild VFL (−6 ≤ MD ≤ −2 in the worse eye), unilateral moderate to severe VFL (MD < −6 in one eye, MD > −2 in the other eye), bilateral mild VFL (−6 ≤ MD ≤ −2 in both eyes, or −6 ≤ MD ≤ −2 in one eye, MD < −6 in the other eye), bilateral moderate to severe VFL (MD < −6 in the both eyes). †Wilcoxon rank sum test was used to compare non-normal continuous variables. Fisher’s exact test was used to compare categorical variables. ‡Effect sizes were calculated as standardized mean differences between the VFL severity cohort and the excluded AFEDS participants by subtracting the means of the two groups and dividing by the overall standard deviation. §Number of self-reported comorbidities (diabetes, arthritis, stroke/brain hemorrhage, high blood pressure, angina, heart attack, heart failure, asthma, skin cancer, other cancer, back problems, hearing problems and other major health problems). ‖Visual Acuity Loss was defined as presenting visual acuity 20/40 or worse. ¶Depression was scored using the SF-12 item “Have you felt downhearted or blue a good bit of the time or more during the past 4 weeks?” Participants were considered depressed if they reported “A good bit of the time”, “Most of the time”, or “All of the time”. 44 Table 7: Item Response Theory Diagnostics for Task and Well-being Composite Models Eigenvalues of the Polychoric Correlation Matrix Task Composite Well-Being Composite Eigenvalue Difference Proportion Cume Eigenvalue Difference Proportion Cume 1 9.210 8.390 0.709 0.709 8.067 7.065 0.672 0.672 2 0.820 0.140 0.063 0.772 1.001 0.044 0.083 0.756 3 0.680 0.157 0.052 0.824 0.958 0.398 0.080 0.836 4 0.523 0.151 0.040 0.864 0.560 0.028 0.047 0.882 5 0.372 0.113 0.029 0.893 0.532 0.271 0.044 0.927 6 0.258 0.032 0.020 0.913 0.262 0.091 0.022 0.948 7 0.226 0.036 0.017 0.930 0.171 0.033 0.014 0.963 8 0.190 0.017 0.015 0.945 0.138 0.033 0.012 0.974 9 0.174 0.022 0.013 0.958 0.105 0.009 0.009 0.983 10 0.152 0.004 0.012 0.970 0.096 0.026 0.008 0.991 11 0.148 0.018 0.011 0.981 0.070 0.031 0.006 0.997 12 0.131 0.015 0.010 0.991 0.039 0.003 1.000 13 0.115 0.009 1.000 Model Fit Statistics Task Composite Well-Being Composite Log Likelihood -22940 -24803 AIC (Smaller is Better) 46011 49730 BIC (Smaller is Better) 46436 50136 LR Chi-Square 9044 7150 LR Chi-Square DF 1220703059 351562437 Cronbach’s Alpha 0.886 0.810 Test Information Curves Task Composite Well-Being Composite 45 Association of VSQOL and VFL In this cohort of African Americans, a negative, linear association was found between VFL and VSQOL (more negative MD values were associated with lower predicted quality of life scores) after adjusting for sociodemographic and clinical covariates. This association was present when VFL was measured as either continuous MD or categorized as unilateral or bilateral severity levels. Continuous Visual Field Loss LOWESS plots reveal strong negative, linear associations of predicted NEI-VFQ-25 composite scores with VFL in the better seeing eye (Figure 7 illustrates IRT composites, and Figure 8 for CTT subscales). Predicted task scores were consistently higher than well-being scores across the entire range of MD of VFL. NEI-VFQ-25 IRT task (β = 0.69) and well-being (β = 0.64) composites had stronger adjusted associations with VFL than the CTT composite (β = 0.53) (Table 8). Of the CTT subscales, driving difficulties (β = 1.14), vision-related mental health (β = 0.61), and general- vision (β = 0.61) had the strongest associations with VFL. The SF-12 physical component score was not significantly associated with VFL. Associations were attenuated for all HRQOL scores with VFL in the WSE. A 5-point loss in IRT task composite was associated with a 7.3 dB decrement in MD and a 5-point loss in IRT well-being composite was associated with a 7.8 dB decrement in MD. Driving difficulties had the strongest association with VFL; a clinically important loss was associated with 4.4 dB of MD. 46 Figure 7: LOWESS plot of predicted VSQOL IRT composite scores from linear regression on VFL in the BSE LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; IRT = Item Response Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye The LOWESS smoothing parameter is 0.6. Gray bars represent 95% confidence limits of the predicted NEI-VFQ-25 IRT composite scores. Linear regression models were adjusted for age, number of comorbidities, sex (female), education (< 4 years of college), working status (unemployed), income (≤ $20,000), has health insurance (yes), has vision insurance (yes), visual acuity loss (20/40 or worse), and depression (a good bit of the time or more in the last 4 weeks). 47 Figure 8: LOWESS plot of predicted VSQOL CTT driving difficulties from linear regression on VFL in the BSE|| LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye The LOWESS smoothing parameter was 0.6. Gray bars represent 95% confidence limits of the predicted NEI-VFQ-25 CTT driving difficulties scores. The linear regression model was adjusted for age, number of comorbidities, sex (female), education (< 4 years of college), working status (unemployed), income (≤ $20,000), has health insurance (yes), has vision insurance (yes), visual acuity loss (20/40 or worse), and depression (a good bit of the time or more in the last 4 weeks). ||Scores could be generated for only 4,594 of the participants who reported that they were currently driving or had driven in the past. 48 Table 8: Linear Regression β Coefficients for the Association Between VSQOL and VFL in the BSE in AFEDS Vision-Specific Quality of Life Measures VFL in Better Seeing Eye P-Value β coefficient (95% CI)* Clinically Meaningful Difference in MD (dB)** NEI-VFQ-25 Item Response Theory IRT Task Composite† 0.69 (0.60, 0.78) 7.3 < 0.001 IRT Well-Being Composite‡ 0.64 (0.54, 0.74) 7.8 < 0.001 Classical Test Theory CTT Composite§ 0.53 (0.48, 0.58) 9.4 < 0.001 Driving Difficulties|| 1.14 (1.03, 1.25) 4.4 < 0.001 Vision-Related Mental Health 0.61 (0.53, 0.70) 8.2 < 0.001 General Vision 0.61 (0.51, 0.71) 8.2 < 0.001 Near Vision 0.60 (0.52, 0.68) 8.3 < 0.001 Peripheral Vision 0.59 (0.52, 0.66) 8.5 < 0.001 Vision-Related Role Function 0.57 (0.48, 0.66) 8.8 < 0.001 Vision-Related Dependency 0.55 (0.48, 0.61) 9.2 < 0.001 Distance Vision 0.49 (0.43, 0.56) 10.1 < 0.001 Vision-Related Social Function 0.46 (0.42, 0.51) 10.8 < 0.001 General Health 0.43 (0.29, 0.58) 11.6 < 0.001 Color Vision 0.29 (0.24, 0.33) 17.4 < 0.001 Ocular Pain 0.17 (0.09, 0.25) 29.8 < 0.001 SF-12 Mental Component Score 0.08 (0.04, 0.12) 60.9 < 0.001 Physical Component Score -0.03 (-0.06, 0.01) 0.104 n = 5,153; VSQOL = Vision-Specific Quality of Life; VFL = Visual Field Loss; BSE = Better Seeing Eye; AFEDS = African American Eye Disease Study; 95% CI = 95% confidence interval; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; MD = Mean Deviation; IRT = Item Response Theory; CTT = Classical Test Theory; SF-12 = 12-Item Short-Form Health Survey *VFL is presented as mean deviation score in decibels; vision-specific quality of life is assessed by the NEI-VFQ-25; and health-related quality of life is assessed by the SF-12. Data are presented as coefficient (95% CI). The SF-12 and NEI-VFQ-25 scores are adjusted for age, gender, education, employment status, income, acculturation, co-morbidities, health insurance, vision insurance, and visual acuity impairment. **Regression coefficients were transformed per 5-point difference in HRQOL score, a clinically significant difference in VSQOL score. 49 †IRT Task Composite was calculated from a graded response theory model of 13 items from near vision, distance vision, driving, color vision, peripheral vision, and role difficulties subscales. ‡IRT Well-Being Composite was calculated from a graded response model of 12 items from general vision, dependency on others, mental health, ocular pain, and social functioning subscales. §Composite score is an un-weighted mean of the 12 subscale scores (excluding general health). ||Scores could be generated for only 4,594 of the participants who reported that they were currently driving or had driven in the past. 50 Severity Categories of Visual Field Loss Clinically meaningful differences in VSQOL scores (≥ 5 points) were only observed for participants with bilateral moderate to severe VFL (Table 3). Comparisons of bilateral moderate to severe VFL and no VFL were clinically meaningful for driving difficulties (18.5), IRT task composite (10.8), IRT well-being composite (9.5), general vision (8.8), vision-related role function (8.6), general health (8.1), vision-related mental health (7.7), near vision (7.5), peripheral vision (7.3), CTT composite (7.0), distance vision (6.1), and vision-related dependency (5.9 points) scores. No clinically meaningful differences were observed between those with no VFL and bilateral mild VFL, unilateral moderate to severe VFL, or unilateral mild VFL. Effect sizes of comparisons in VSQOL were largest for participants with bilateral moderate to severe VFL (Figure 9). Task-based subscales of the NEI-VFQ-25 had large to medium ES for comparisons between bilateral moderate to severe VFL and no VFL, including driving difficulties (1.44), the IRT task composite (0.80), and peripheral vision (0.78). Well-being domains had medium ES for bilateral moderate to severe VFL, including social function (0.73), dependency (0.70), and the IRT well-being composite (0.64). ES were small or negligible for comparisons between those with no VFL and bilateral mild VFL, unilateral moderate to severe VFL, or unilateral mild VFL. Table 9: ANCOVA Assessing the Relationship Between VSQOL and Various Categories of Severity of VFL in the AFEDS Vision-Specific Quality of Life Measures Adjusted Mean Scores (SE)* Visual Field Loss Severity Categories (n = 4,187)† P-Value§ No VFL (n = 2,356) Unilateral Bilateral Mild (n = 795) Moderate to Severe (n = 102) Mild (n = 775) Moderate to Severe (n = 159) NEI-VFQ-25 Item Response Theory IRT Task Composite¶ 89.2 (0.3)a 87.6 (0.5)b 87.9 (1.3)a,b 85.8 (0.5)b 78.4 (1.1)c < 0.001 IRT Well-Being Composite◊ 75.1 (0.3)a 73.5 (0.5)b 71.7 (1.4)a,b,c 71.3 (0.5)c 65.6 (1.2)d < 0.001 Classical Test Theory CTT Composite‡ 94.7 (0.1)a 93.7 (0.2)b 93.8 (0.6)a,b 92.9 (0.2)b 87.7 (0.5)c < 0.001 Driving Difficulties|| 95.5 (0.3)a 93.6 (0.5)b 95.2 (1.3)a,b 93.1 (0.5)b 77.0 (1.3)c < 0.001 Vision-Related Mental Health 91.3 (0.2)a 90.2 (0.4)a,b 88.6 (1.2)a,b 88.8 (0.4)b 83.6 (1.0)c < 0.001 General Vision 75.1 (0.3)a 73.1 (0.5)b 72.4 (1.5)a,b,c 70.2 (0.5)c 66.3 (1.2)d < 0.001 Near Vision 95.2 (0.2)a 94.1 (0.4)a,b 94.7 (1.1)a,b 92.7 (0.4)b 87.6 (0.9)c < 0.001 Peripheral Vision 98.0 (0.2)a 97.0 (0.3)b 96.0 (0.9)a,b 96.6 (0.3)b 90.7 (0.8)c < 0.001 Vision-Related Role Function 96.8 (0.3)a 96.2 (0.5)a 96.4 (1.3)a 95.8 (0.5)a 88.2 (1.1)b < 0.001 Vision-Related Dependency 99.0 (0.2)a 98.3 (0.3)a 98.7 (0.8)a 98.2 (0.3)a 93.1 (0.7)b < 0.001 Distance Vision 97.3 (0.2)a 96.6 (0.3)a,b 96.8 (0.9)a,b 95.8 (0.3)b 91.3 (0.7)c < 0.001 Vision-Related Social Function 99.2 (0.1)a 98.6 (0.2)a 98.3 (0.6)a 98.5 (0.2)a 94.8 (0.5)b < 0.001 General Health 58.3 (0.4)a 57.0 (0.7)a,b 55.5 (2.1)a,b,c 54.7 (0.8)b,c 50.2 (1.7)c < 0.001 Color Vision 99.2 (0.1)a 99.0 (0.2)a 99.5 (0.6)a 99.1 (0.2)a 96.9 (0.5)b < 0.001 Ocular Pain 94.9 (0.2)a 94.2 (0.4)a 95.4 (1.1)a 93.9 (0.4)a 90.5 (0.9)b < 0.001 SF-12 Mental Component Score 60.6 (0.1)a 59.9 (0.2)a,b 60.9 (0.6)a,b 59.8 (0.2)b 59.5 (0.5)a,b 0.006 Physical Component Score 39.8 (0.1)a 39.9 (0.2)a 39.8 (0.4)a 39.8 (0.2)a 39.3 (0.4)a 0.729 ANCOVA = analysis of covariance; VSQOL = Vision-Specific Quality of Life; VFL = Visual Field Loss; AFEDS = African American Eye Disease Study; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; IRT = Item Response Theory; CTT = Classical Test Theory; SF-12 = 12-Item Short-Form Health Survey; MD = Mean Deviation 51 *Shown are the adjusted mean (standard error) NEI-VFQ-25 and SF-12 scores. The covariates for adjustment include age, gender, education, employment status, income, comorbidities, health insurance, vision insurance, and presenting visual acuity 20/40 or worse. †VFL was classified as none (MD > −2), unilateral mild VFL (−6 ≤ MD ≤ −2 in the worse eye), unilateral moderate to severe VFL (MD < −6 in one eye, MD > −2 in the other eye), bilateral mild VFL (−6 ≤ MD ≤ −2 in both eyes, or −6 ≤ MD ≤ −2 in one eye, MD < −6 in the other eye), bilateral moderate to severe VFL (MD < −6 in the both eyes). §Analysis of covariance was used to compare adjusted mean scores across the different levels of unilateral and bilateral VFL. The P- value in the last column corresponds to the ANCOVA type III sums of squares f-test across the VFL groups, adjusted for multiple comparisons. For each row, means with different letters (a–d) across the VFL categories are statistically significantly different from one another after adjusting for multiple comparisons using the Tukey-Kramer method (p < 0.05). ANCOVA using Tukey multiple comparisons revealed significant differences across the VFL categories for all scales except the physical component score of the SF- 12. ‡Composite score is an un-weighted mean of 11 of the 12 NEI-VFQ-25 subscale scores (except general health). ||Scores could be generated for only 3,780 of the participants who reported that they were currently driving or had driven in the past; the sample size was 2,214, 707, 94, 657, and 108 for the five VFL categories, respectively. ¶IRT Task Composite was calculated from a graded response theory model of 13 items from near vision, distance vision, driving, color vision, peripheral vision, and role difficulties subscales. ◊IRT Well-Being Composite was calculated from a graded response model of 12 items from general vision, dependency on others, mental health, ocular pain, and social functioning subscales. 52 53 Figure 9: Comparing VSQOL Effect Sizes in AFEDS Participants in each VFL Severity Category and those with No VFL n = 4,187; ES below 0.20 are negligible and not shown. ES from 0.20 to less than 0.50 are considered small, 0.50 to less than 0.80 are medium, and 0.80 or more are large. ES were calculated from ANCOVA models as the difference in adjusted mean QOL scores for each VFL severity category and the no VFL category, divided by the standard deviation of QOL score in the no VFL group. ES are shown for the NEI-VFQ-25 CTT composite, the IRT task and well-being composites, all 11 CTT subscales, and the general health item; CTT subscales are grouped by task or well-being and ordered by descending ES in the bilateral moderate/severe comparison. The SF-12 component scores are also shown. VFL severity was stratified into five categories: no VFL (mean deviation [MD] > 2 decibels [dB] in both eyes), unilateral mild VFL (-6 dB <MD< -2 dB in the worse eye); bilateral mild VFL ( 6 dB < MD < 2 dB in both eyes; unilateral moderate to severe VFL (MD<-6 dB in one eye, MD > 2 dB in the other eye; or 6 dB < MD < 2 dB in one eye, MD < 6 dB in the other eye), and bilateral moderate to severe VFL (MD < 6 dB in both eyes). *NEI-VFQ-25 classical test theory and item response theory composite scores are marked for emphasis. ||Scores could be generated for only 4,594 of the participants who reported that they were currently driving or had driven in the past. 54 Discussion In this study of African Americans, lower VSQOL scores were associated with worse VFL in the BSE. A clinically-significant difference in the IRT task and well-being composites was associated with 7 to 8 dB of MD. CTT analysis revealed driving difficulties (clinically-significant difference at 4.4 dB) and vision-related mental health (8.2 dB) are two specific subscales of task and well-being VSQOL most impacted by VFL in African Americans. VSQOL showed a dose-response with VFL severity categories, which was abrupt for task VSQOL and gradual for well-being VSQOL. IRT analysis revealed a small reduction in task VSQOL for those with bilateral mild VFL, but a large deficit for those with bilateral moderate to severe VFL; there was no difference in task VSQOL for those with unilateral VFL. However, there was a small difference in well-being VSQOL for those with unilateral moderate to severe and bilateral mild VFL, and also a medium effect in those with bilateral moderate to severe VFL. Clinicians should inform patients with unilateral or bilateral mild VFL that they may have mild reductions to their socioemotional well-being, especially their mental health. But those with more than 6 dB of VFL in both eyes may have larger losses in their ability to function and complete vision-related tasks. These patients should make transportation arrangements other than driving. No large, population-based cohort studies of African Americans were previously available to assess the relationship between VFL and VSQOL. However, several US studies have found that worse VFL is associated with lower VSQOL in multiethnic cohorts that include African Americans.20,24,25,43 The Barbados Eye Study has demonstrated that lower VSQOL is associated with primary OAG in a population of Caribbean adults of African descent.84 Population-based studies that have evaluated the impact of VFL on VSQOL or function have focused on non- Hispanic White13,14,16,20,24–26,43,114 and Latino populations21–23. Population-based studies reporting 55 on VFL among glaucoma patients have found similar negative associations with VSQOL among non-Hispanic White and Latino populations.24,75 Multiethnic comparisons from the National Health and Nutrition Examination Survey (NHANES) have observed VFL to be associated with worse VSQOL. Data from NHANES was used to evaluate the impact of VFL on 8 individual items of the NEI-VFQ-25. They found a higher percentage of participants reported at least some difficulty with vision related task or well- being items with increasing severity of VFL; the strongest relationship with VFL was driving difficulties during the day.20 Additionally, African Americans were more than three times as likely to have severe VFL compared to non-Hispanic Whites. The Salisbury Eye Evaluation Study (SEES) found that after adjusting for all other measures of visual impairment, missing twice as many points of central visual field was associated with 1.37 (95% CI 1.19, 1.58) greater odds of scoring in the worst tertile of VSQOL, as measured by the overall score of the Activities of Daily Vision Scale (ADVS).25 In SEES, African Americans had worse VFL compared to non-Hispanic Whites across all age strata. Instead of categorizing VSQOL, the Los Angeles Latino Eye Study (LALES) reported on the linear relationship of VFL with continuous measures of the NEI-VFQ- 25. In the LALES, a longitudinal, population-based study of Latinos, a 5 dB difference in VF was associated with a clinically meaningful (5-point) difference in NEI-VFQ-25 composite scores.21 In the present study of African Americans, a larger magnitude of VFL (7–8 dB) was associated with a clinically meaningful difference in VSQOL. Comparisons of these findings should follow a pooled analysis of the 2 cohorts to allow for common statistical techniques and confounder adjustment. Few population-based studies have evaluated the association between VFL and VSQOL in glaucoma patients of African descent. The SEES found African Americans were twice as likely (2.20, 95% CI 1.26–3.84) as non-Hispanic Whites to report worse VSQOL.24 In the BES, all 56 participants were of African descent and 19% were diagnosed with OAG.10 Glaucoma patients had significantly lower NEI-VFQ-25 subscale scores for vision-specific mental health, social functioning, and distance vision after adjusting for demographics and VA, however VFL was not accounted for in the models. Clinic-based studies have also examined how glaucomatous VFL affects VSQOL in populations that include African Americans. A multi-center study found 8 subscales of the NEI- VFQ-25 were negatively associated by VFL in glaucoma patients, a third of whom were African American; driving difficulties had the strongest linear association with VFL.35 Another study in Alabama was two-thirds African American and observed worsening VFL was associated with a decrease in most NEI-VFQ-25 subscales.37 Interestingly, findings were similar for both White and African American patients. The AFEDS cohort is highly educated, and 90% have access to health insurance. Therefore findings may be less generalizable to African American populations of varying socioeconomic status. However baseline assessment found that sociodemographics of the AFEDS cohort were similar to African Americans in Los Angeles, California, and the country overall.51 Longitudinal studies of VFL in population-based samples of African Americans are necessary to determine whether patterns observed in this cross-sectional data are predictive of individual experiences. 57 Chapter 3: Variability of Visual Field Loss on Vision-Specific Quality of Life in the Multiethnic Ophthalmology Cohorts of California Study Abstract Importance Visual impairment (VI) may differentially affect people from diverse backgrounds, similarly to racial/ethnic variations in the prevalence of eye diseases. Vision-specific quality of life (VSQOL) assess subjective experience of vision-related task and well-being. We aim to investigate how people of various racial/ethnic backgrounds may be differentially impacted by visual field loss (VFL)—a common measure of VI. We also assessed whether differences of association varied in participants younger and older than 65 years. Objective To assess how VFL impacts VSQOL in Latinos, Chinese Americans and African Americans aged 40 years and older. Design Pooled analysis of three cross sectional, population-based studies of eye disease Setting Three cities in Los Angeles County with large self-identified minority populations; Latinos from La Puente, Chinese Americans from Monterey Park and African Americans from Inglewood, California. 58 Participants 6,142 Latinos, 4,582 Chinese Americans, and 6,347 African Americans from Los Angeles County Exposure Total mean deviation of visual field loss in the better-seeing eye Main Outcomes and Measures VSQOL outcomes were measured by the 25 Item National Eye Institute Visual Functioning Questionnaire (NEI-VFQ-25) and analyzed by item response theory (IRT) for task and well- being domains as well as analysis by classical test theory (CTT) that allowed ranking of 11 subscales of VSQOL. Results Latinos had the strongest associations between VFL and all outcomes of VSQOL, which statistically significantly larger than associations in African Americans for task and well-being domains. Associations in Chinese Americans were not significantly different from Latinos but were significantly smaller than African Americans for well-being VSQOL. Using common cutoffs for VFL, moderate VFL was associated with clinically meaningful differences in functional task VSQOL for all racial/ethnic groups. Moderate VFL was also clinically related to lower socioemotional well-being VSQOL for Latinos and Chinese Americans, but severe VFL in African Americans. Driving difficulties had the largest association with VFL in Latinos and African Americans but was ranked the third largest for Chinese Americans. Driving difficulties was the only measure of VSQOL that was modified by age in its association with VFL. Vision- related mental health was also highly associated with VFL for all racial/ethnic groups. Larger associations with VFL were observed for vision-related dependency in Latinos, for role function in Chinese Americans, and for near and peripheral vision in African Americans. 59 Conclusions and Relevance Racial/ethnic differences in the association of VFL and VSQOL have implications for clinical trials, which should include a diverse patient population if interested in applying findings to the general US population. Despite these differences, there is evidence that driving and mental health are important to all racial ethnic groups, and that moderate VFL may have a clinical impact on all people’s VSQOL. One exception is African Americans did not report clinically meaningful losses in well-being VSQO until severe VFL occurred, which may be due to a legacy of mistrust of the healthcare system due to historical injustices. 60 Introduction The prevalence of chronic eye disease is expected to increase as the US population ages over the next 30 years. Following this shift in the demographic distribution from younger to older age groups, the financial burden of visual impairment (VI) and blindness is projected to double in the US by 2050.2 Another consequence of the growing prevalence of VI and blindness will be a decline in vision specific quality of life (VSQOL) including ability to complete daily tasks around the home and work place, social and emotional well-being, independence, mobility, and cognitive function. Recent data suggests that sensory loss from VI may contribute to cognitive decline as early as middle age.115 VI and blindness due to various eye pathologies may impact subpopulations at differing rates and severity. More specifically, losses in visual acuity (VA) or visual field (VF) due to eye disease are known to vary by race/ethnicity, sex, and age.3,116 For example, Latinos experience a disproportionate amount of diabetic retinopathy,117 African Americans a disproportionate amount of proliferative diabetic retinopathy and open-angle glaucoma,4,118 while Chinese Americans have a greater prevalence of myopic degeneration.119 The potential impact of VI on visual function as measured by subjective patient reported outcomes (PRO) has not been explored across racial/ethnic groups in large, population-based samples. PRO have been used in clinical settings to inform medical decision-making, facilitate communication between patients and providers, and test the efficacy of rehabilitation programs.120 Additionally, PRO measured VSQOL is increasingly used as an outcome in clinical trials,121,98,122 to inform whether interventions are sufficiently safe and efficacious to be made available to the public.123 Among existing clinical studies, few have investigated whether race/ethnicity plays an important role in the association of eye disease and associated VI with patient-reported outcomes (PRO).124–127 If these differences result in health disparities in VSQOL among US populations, then it is essential to have normative data from diverse 61 racial/ethnic populations to appropriately interpret visual function data from clinical trials and to understand the priorities and concerns of different communities before providing eye services. Using data from the Multiethnic Ophthalmology Cohorts of California Study (MOCCaS), we provide the first population-based evaluation of the association between measures of VF (due to any cause) and VSQOL in a multiethnic sample of US adults. A limited number of population- based studies have been conducted on VF and VSQOL, however these were limited to small numbers of participants representing individual minority groups, or analysis of one population at a time, or based on definitions and clinical protocols from more than 20 years prior.20,21,24,25,128 MOCCaS is a pooled analysis of adults 40 years and older living in Los Angeles County, composed of data using modern definitions and clinical protocols from three population-based cohort studies of Latinos,57 Chinese Americans,56 and African Americans.51 We hypothesized that independent of visual acuity and sociodemographic and clinic factors (1) race/ethnicity would be an important effect modifier of the association between VFL and VSQOL when evaluated by composite scores constructed for vision-related tasks and social-emotional well- being; (2) driving difficulties from the NEI-VFQ-25 VSQOL scale would have the strongest association with VFL for all racial/ethnic groups; and (3) greater age would act as an effect modifier, such that participants ≥ 65 years of age would have greater deficits in VSQOL for the same amount of VFL compared to their younger counterparts. 62 Methods The MOCCaS is a pooled analysis of 17,071 subjects residing in Los Angeles County, California who were studied during three population-based, cross-sectional cohort studies: the Los Angeles Latino Eye Study (LALES),57 the Chinese American Eye Study (CHES),56 and the African American Eye Disease Study (AFEDS).51 In brief, LALES participants were studied from 2000 to 2003 and resided in the city of La Puente; CHES participants were studied from 2010 to 2013 and resided in Monterey Park; AFEDS participants were studied from 2014 to 2018 and resided in Inglewood. LALES and AFEDS participants were aged 40 years and older, and CHES participants were aged 50 years and older. Data were collected from interviews and comprehensive clinical eye examinations, which were standard across all three groups. Demographics of MOCCaS participants were similar to their respective racial/ethnic populations in Los Angeles County, California, and the United States. Approval was obtained by the University of Southern California Medical Center Institutional Review Board. All study procedures adhered to the recommendations of the Declaration of Helsinki. Sociodemographic Assessment Eligible residents were identified by door-to-door census and invited to participate. Participants were interviewed in their homes and later completed a comprehensive clinical eye examination by study ophthalmologists at the local eye examination clinic. Home interviewing was conducted after informed consent to gather demographic factors (age, sex, education, employment status, country of birth), history of medical conditions (e.g. history of diabetes, history of hypertension, and comorbidities), and access to ocular and medical care (e.g. medical insurance, income, etc.). A comorbidity score (ranging from zero to thirteen) was calculated as a summation of thirteen self-reported medical conditions, including diabetes mellitus, arthritis, stroke or brain hemorrhage, hypertension, angina, heart attack, heart failure, asthma, skin cancer, other cancers, back problems, deafness or hearing problems, and other comorbidities. 63 Participants who completed the in-home questionnaire also completed an additional brief interview at the time of clinical eye examination that included information on quality of life and visual function. Visual Function Assessment VA measures high acuity, central vision. Binocular VA was assessed with presenting correction at 4 meters using standard Early Treatment Diabetic Retinopathy protocols with a modified distance chart illuminator (Precision Vision).70,129 If needed, an automated refraction was performed, using the Humphrey Automatic Refractor (Carl Zeiss Meditec, Dublin, CA), followed by subjective refraction. VA loss was defined as VA of 20/40 or worse based on the U.S. definition of visual impairment (VI). LogMAR score is a common linear transformation of VA, which was used in regression models. Visual field testing assesses participant ability to detect objects in their entire visual field— both centrally and peripherally. Visual fields for each eye were assessed using the Swedish Interactive Threshold Algorithm (SITA) Standard C24-2 test (Carl Zeiss Humphrey Field Analyzer II 750 Dublin, CA). VF was measured using total mean deviation (MD)—in decibels (dB)—from the age-adjusted standard population.73 Visual field testing was repeated up to 2 times if the measurement was unreliable. Measurements with 15% or more false negatives or false positives were considered unreliable and were excluded from analysis.130 Smaller (more negative) scores indicate greater VFL. Continuous VFL was stratified into the better- (BSE) and worse-seeing eye (WSE); results are only presented for the BSE, which is strongly related to binocular VFL.76 64 Vision-Specific Quality of Life VSQOL was measured using the National Eye Institute Visual Functioning Questionnaire-25 (NEI-VFQ-25) survey instrument,89 which was administered by a trained interviewer prior to the clinical examination. Classical test theory (CTT) is a method for scoring the NEI-VFQ-25, which allows for assessment of specific subscales of VSQOL, but relies on strict assumptions—equal differences between response categories, at least 95% complete data, constant standard error—that may not always be met. One advantage of CTT is the ability to evaluate multiple subscales that may improve communication between clinicians and patients as to potential benefits of clinical interventions in areas of specific vision related tasks or social interactions. Item response theory (IRT) has been favored in the recent ophthalmic literature.91,92 IRT is a robust statistical method that relies on fewer statistical assumptions and is used to analyze broad, unidimensional domains of VSQOL. We consider IRT composite domains as the primary outcomes for the current analysis, from which we base our quantitative claims. In addition, we include CTT subscales as secondary outcomes to qualitatively assess which subdomains of VSQOL are more strongly associated with VFL. IRT domains and CTT subscales were standardized from 0 to 100, with higher scores indicating greater VSQOL. NEI-VFQ-25: Item Response Theory (Graded Response Model) IRT models were used to classify participants with varying magnitude of VSQOL latent scores along a linear continuum (low to high) of NEI-VFQ-25 item difficulty.131,132 The graded response model (GRM) is a 2-parameter IRT model for ordinal items on a 5-point, Likert scale. The GRM was used to produce two unidimensional composites: task and well-being. The functional, task composite score was calculated from 12 items, and the socioemotional well- being composite was calculated from 12 items (Table 10). General health was not included in either domain. Composite scores were calculated using the ltm package for R software.133 65 Table 10: VSQOL survey instrument: NEI-VFQ-25 items, CTT subscales, and IRT domains Summary CTT Subscale IRT Domain Task Domain Items Accomplish less than you liked? Vision-Related Role Function Task Limited in how long could work? Vision-Related Role Function Task Difficulty noticing objects while walking Peripheral Vision Task Difficulty read ordinary print Near Vision Task Difficulty work needs see well up close Near Vision Task Difficulty find things on a crowded shelf Near Vision Task Difficulty read street signs/store names Distance Vision Task Difficulty go down steps in dim light Distance Vision Task Difficulty going out to see movies, plays Distance Vision Task Difficulty picking out/matching own clothes Color Vision Task Difficulty driving at daytime in familiar places Driving Difficulties Task Difficulty driving at night Driving Difficulties Task Difficulty driving in difficult conditions (bad weather, rush hour, freeway, city traffic)* NA NA Well-Being Domain Items How much do you worry eyesight? Vision-Related Mental Health Well-Being Feel frustrated a lot due eyesight Vision-Related Mental Health Well-Being Have much less control due eyesight Vision-Related Mental Health Well-Being Worry about doing things embarrass myself Vision-Related Mental Health Well-Being At present time, your eyesight is General Vision Well-Being Stay home most of time due eyesight Vision-Related Dependency Well-Being Rely too much on what others tell me Vision-Related Dependency Well-Being Need a lot of help from others Vision-Related Dependency Well-Being Difficulty see how people react to things Vision-Related Social Function Well-Being Difficulty visiting people in their home/party Vision-Related Social Function Well-Being How much of pain/discomfort in eyes? Ocular Pain Well-Being Pain in eyes keep you from doing you like? Ocular Pain Well-Being General Health Item In general, would you say your health is? 1 Excellent, 2 Very good, 3 Good, 4 Fair, 5 Poor General Health NA VSQOL = Vision-Specific Quality of Life; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; IRT = Item Response Theory; CTT = Classical Test Theory *The item for difficult driving conditions was not asked in LALES, and is not included in scoring 66 NEI-VFQ-25: Classical Test Theory Each CTT subscale was calculated by calculating the mean score of related NEI-VFQ-25 items (Table 10).89 CTT subscales that relate to the IRT task domain include driving difficulties, vision-related role function, near vision, peripheral vision, distance vision, and color vision; CTT subscales that relate to the IRT well-being domain include vision-related mental health, vision- related dependency, vision-related social functioning, general vision, ocular pain. Statistical Analysis Data for the statistical analyses were restricted to those from participants with complete and reliable VFL measurements in the BSE and complete QOL outcome variables. Sociodemographic variables were compared across racial/ethnic cohorts; tests for equal means in a one-way layout and Pearson’s χ2 tests were performed to assess differences in continuous and categorical covariables, respectively. Differences in sociodemographic variables across cohorts were assessed by Bonferroni-adjusted pairwise comparisons. Hierarchical, multivariable linear regression was used to assess the outcomes of VSQOL on the exposure of VFL in the BSE. Regression equation were fit for the NEI-VFQ-25 outcomes: both task and well-being IRT composites, and the 11 CTT subscales. Covariates were selected that were previously identified in the literature as predictors of VSQOL or were considered potentially related to both VFL and perceived VSQOL. Models were adjusted for age (years); education (highest grade completed); number of comorbidities; sex (female); born in the US (yes); employment status (working); income (≥ $20,000); health insurance (yes); depression (a good bit of the time or more in the last 4 weeks); and presenting, binocular LogMAR VA. US birth was used as a proxy for acculturation because it was from all three cohorts. Hierarchical regression models included a (1) main effects model, (2) an interaction model for VFL and race/ethnicity, and (3) an interaction model for VFL, race/ethnicity and age dichotomized at 65 years. Nested models were compared using the Multivariate Wald test.134 Multiple imputation 67 with chained equations (MICE) produces unbiased and accurate regression coefficients from data that is missing at random. The mice package for R was used to create and analyze 10 multiply imputed data sets.134 Locally weighted scatterplot smoothing (LOWESS) plots with 95% confidence limits were produced for predicted VSQOL outcomes of each regression model. Beta coefficients of VFL in the BSE from the regression models were multiplied by 5 units of VSQOL to obtain the corresponding change in VFL for each composite and subscale. A 5-unit change in NEI-VFQ-25 was considered clinically meaningful, as it has been associated with a 2- line deficit in VA.10 Two sensitivity analyses were conducted for subgroups of the study population; first, for those that did not have visual impairment, and second, for those that were not depressed. As this is a pooled analysis of three cross-sectional studies, I was unable to assess longitudinal change of VSQOL domains with VFL. Words such as “lower”, “decrease”, and “decrement” were used to reflect cross-sectional relationships among measures of interest. All analyses were performed using R statistical software, version 3.6.1.135 All statistical tests were performed with a type 1 error rate of 0.05. All data visualization was produced using the ggplot2 package for R.136(p2) 68 Results Participants The MOCCaS included 17,071 participants from the LALES, the CHES, and the AFEDS (Figure 10). 14,570 (85.3%) participants had complete outcome and exposure data and were included in the analytic cohort. 6.3% of participants were excluded due to missing VSQOL outcomes, which were mostly incomplete in the LALES due to the VSQOL instrument implementation in year 2 of the LALES study. Individuals with missing data for the driving difficulties items were not excluded (27.2% of participants) because the questionnaire was designed with a skip pattern over these questions when the person was a non-driver or who stopped driving for reasons other than vision loss.89 Finally, 8.6% of participants had missing or unreliable visual field measurements, of which 65% (953) were missing from AFEDS. 69 Figure 10: Flow Diagram for MOCCaS MOCCaS Latino Chinese American African American Population-based cohorts LALES CHES AFEDS n = 17,071 6,142 4,582 6,347 Excluded Incomplete Outcome NEI-VFQ-25 CTT Composite 970 5.7% 762 12.4% 18 0.4% 190 3.0% Color vision 1,020 6.0% 771 12.6% 33 0.7% 216 3.4% Dependency 1,066 6.2% 767 12.5% 23 0.5% 276 4.3% Driving difficulties‖ 4,644 27.2% 2,279 37.1% 1,357 29.6% 1,008 15.9% Distance vision 1,068 6.3% 768 12.5% 96 2.1% 204 3.2% General vision 972 5.7% 763 12.4% 18 0.4% 191 3.0% Mental health 972 5.7% 763 12.4% 18 0.4% 191 3.0% Near vision 1,008 5.9% 764 12.4% 46 1.0% 198 3.1% Ocular pain 973 5.7% 762 12.4% 18 0.4% 193 3.0% Peripheral vision 1,023 6.0% 769 12.5% 40 0.9% 214 3.4% Role function 1,047 6.1% 764 12.4% 18 0.4% 265 4.2% Social function 996 5.8% 767 12.5% 21 0.5% 208 3.3% General health 946 5.5% 737 12.0% 26 0.6% 183 2.9% Excluded Incomplete and Unreliable Exposure Visual field mean deviation 1,465 8.6% 166 2.7% 346 7.6% 953 15.0% Analytic Cohort (% Complete) 14,570 (85.3%) 5,248 (85.4%) 4,151 (90.6%) 5,171 (81.5%) MOCCaS = Multiethnic Ophthalmology Cohorts of California Study; LALES = Los Angeles Latino Eye Study; CHES = Chinese American Eye Study; AFEDS = African American Eye Disease Study; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; CTT = Classical Test Theory ‖The NEI-VFQ-25 instructs participants to omit items related to driving difficulties if they report they are not currently driving and had not driven in the past. Participants with missing subscale scores for driving difficulties are included in the analytic cohort. 70 Descriptive data All statistical tests comparing sociodemographics across cohorts within the MOCCaS were statistically significant (p < 0.001) due to the large sample size, with the exception of no differences between Chinese Americans and African Americans for age, sex, or self-reported depression (Table 11). Participants were 59 years old on average; the mean ages of Chinese and African American participants were 6 years older than the mean age of Latinos. Most (61%) of participants were female, but Latinos in this analysis were slightly more likely to be male. The mean highest grade of education completed was 11.5 years; African Americans completed more education (14.3 years) than either Chinese Americans (12.2 years) or Latinos (8.2 years). African Americans had the greatest number of self-reported comorbidities (2.0), compared to Latinos (1.5), and Chinese Americans (1.2). The vast majority of African Americans (91.0%) were born in the US, while a quarter of Latinos (24.5%), and relatively few Chinese Americans (1.3%) were born in the US. With respect to employment status, Chinese Americans were most likely to be working (53.0%), followed by Latinos (48.7%), and then African Americans (44.7%). However, the reverse pattern was observed for having health insurance coverage; African Americans were most likely to be insured (89.7%), then Latinos (64.9%), and Chinese Americans were the least likely to be insured (53.0%). Latinos were more likely to report depression (9.8%) than Chinese and African Americans (6.0–6.6%). Chinese Americans were most likely to have visual impairment (8.9%), followed by Latinos (6.2%), and African Americans were least likely to have visual impairment (4.4%). Among the MOCCaS analytic cohort (n = 14,570), 12,469 (85.6%) of participants had complete cases overall. Less than 5% of participants were missing data for most covariates: 182 (1.2%) for education, 10 (0.1%) for sex, 118 (0.8%) for US born, 182 (1.2%) for employment, 306 (2.1%) for insurance, 12 (0.1%) for depression, and 2 ( 0.0%) for visual acuity loss. Income was by far the largest incomplete covariable, at 1,868 (12.8%) missing. 71 Table 11: Sociodemographic and Clinical Characteristics of Participants in MOCCaS (n = 14,570)*,† Cohort MOCCaS LALES CHES AFEDS Race/ Ethnicity Latino 5,248 (36.0%) 5,248 (100.0%) Chinese American 4,151 (28.5%) 4,151 (100.0%) African American 5,171 (35.5%) 5,171 (100.0%) Age 59.0 (10.7) 55.0 (10.7) a 61.2 (8.5) b 61.2 (11.0) b Education‡ 11.5 (4.5) 8.2 (4.4) a 12.2 (3.8) b 14.3 (2.4) c Number§ of Comorbidities 1.6 (1.5) 1.5 (1.5) a 1.2 (1.4) b 2.0 (1.5) c Sex (Female) 8903 (61.1%) 3069 (58.5%) a 2599 (62.6%) b 3235 (62.6%) b US Born 6048 (41.5%) 1284 (24.5%) a 56 (1.3%) b 4708 (91.0%) c Employment (Working) 7066 (48.5%) 2555 (48.7%) a 2202 (53.0%) b 2309 (44.7%) c Annual Income ≥ $20,000 6997 (48.0%) 2285 (43.5%) a 1716 (41.3%) b 2996 (57.9%) c Health Insurance 10243 (70.3%) 3407 (64.9%) a 2199 (53.0%) b 4637 (89.7%) c Depression|| 1424 (9.8%) 840 (16.0%) a 274 (6.6%) b 310 (6.0%) b Visual Impairment¶ 905 (6.2%) 310 (5.9%) a 369 (8.9%) b 226 (4.4%) c MOCCaS = Multiethnic Ophthalmology Cohorts of California Study; LALES = Los Angeles Latino Eye Study; CHES = Chinese American Eye Study; AFEDS = African American Eye Disease Study; SD = Standard Deviation; VFL = Visual Field Loss *Data are presented as mean (SD) for continuous variables (age, education, comorbidities); categorical variables are presented as frequency counts with percentages (%) of participants for each category of visual field loss (VFL) severity; percentages exclude participants with missing responses); the number missing is 182 (1.2%) for education; 10 (0.1%) for sex, 118 (0.8%) for US born; 182 (1.2%) for employment; 1,868 (12.8%) for income; 306 (2.1%) for insurance; 12 (0.1%); and 2 ( 0.0%) for visual acuity loss. †All P-values were < 0.001 for differences by race/ethnicity. P-values for continuous variables are from the test for equal means in a one-way layout, and for categorical variables are from the Pearson’s χ2 test. Superscripts indicate Bonferroni-adjusted pairwise comparisons. ‡Highest educational grade completed §Number of self-reported comorbidities (diabetes, arthritis, stroke/brain hemorrhage, high blood pressure, angina, heart attack, heart failure, asthma, skin cancer, other cancer, back problems, hearing problems and other major health problems). ||Depression was scored using the SF-12 item “Have you felt downhearted or blue a good bit of the time or more during the past 4 weeks?” Participants were considered depressed if they reported “A good bit of the time”, “Most of the time”, or “All of the time”. ¶Visual Impairment was defined as binocular presenting visual acuity 20/40 or worse. 72 NEI-VFQ-25 Outcome Data Many participants reported the maximum possible quality of life response for NEI-VFQ-25 items (Figure 11) resulting in a skewed response distribution with ceiling effects.91 The ceiling effect was more pronounced for the vision related task, rather than the wellbeing domain. The average IRT score for the task composite was 83.1 out of 100 points maximum (standard deviation [SD] = 16.7), and for the well-being composite was 71.8 points (SD = 16.6). Test information curves revealed the NEI-VFQ-25 was most informative for VSQOL ranging from the mean to 3–4 standard deviations below the mean score in the MOCCaS population. High measures of internal consistency and unidimensionality were observed for the IRT graded response models for both the task and well-being composites. Cronbach’s alpha was 0.888 for task (12 items) and 0.872 for well-being (12 items) composites, indicating high inter-item correlation for each composite score. Factor analysis suggested IRT task and well-being model latent traits were unidimensional, as evidenced by the Scree plots and Chi-Square tests for unidimensionality (P < 0.001); a single factor explained 68.3% of the observed variance for the task composite and 64.7% for the well-being composite. 73 Figure 11: Item response theory diagnostics and factor analysis Descriptive and Fit Statistics Task Composite Well-Being Composite Mean (Standard Deviation) [0–100 points] 83.1 (16.7) 71.8 (16.6) Log Likelihood -88216 -110646 Cronbach’s Alpha Internal Consistency 0.888 0.872 Root Mean Square Error of Residuals 0.08 0.08 Proportion Variance Explained by Single Domain 68.3% 64.7% Empirical Chi Square Test for Unidimensionality < 0.001 < 0.001 Histogram Distributions Test Information Curves (Standardized VSQOL Ability) Scree Plots Assess the Unidimensionality of Item Response Theory Domains 74 Main Effects Models After adjusting for covariates, we found statistically significant, inverse associations between VSQOL and worse VFL in the BSE for both IRT task and well-being domains. All sociodemographic and clinical predictors had beta coefficients in expected directions, except health insurance, which was not associated with the IRT task composite only. The adjusted R- squared was 0.355 for task and 0.317 for well-being. Assumptions of linear regression— linearity, normality, and homoscedasticity—were satisfied. We found no evidence of multicollinearity—the variance inflation factor was below 2—for all predictors in the models. For the IRT task model, a 1 dB lower value of MD for VF was associated with 0.659 (95% CI: 0.585, 0.733) lower VSQOL score; to scale this change to a 5-point clinically important difference in the NEI-VFQ-25, we could say that 7.6 dBs of VF loss is necessary to observe a clinically meaningful lower task-related VSQOL score in the multiethnic cohort. There were also statistically significant differences in VSQOL by race/ethnicity. Chinese Americans had an associated 12.9 points higher, and African Americans 13.3 points higher, compared to VSQOL in Latinos as the reference group. For the IRT well-being model, a 1 dB lower value of MD for VFL was associated with 0.554 (95% CI: 0.478, 0.630) lower VSQOL score; therefore 9.0 dBs of VF loss is necessary to observe a clinically meaningful lower well-being-related VSQOL score in MOCCaS. Chinese Americans had an associated 6.6 points higher and African Americans 11.8 points higher VSQOL compared to Latinos. 75 Race/Ethnicity Interaction Models We identified the relationship between VSQOL and VFL varied within each racial/ethnic cohort using a nested interaction model. Model comparisons were made using the multivariate Wald tests, which showed a significant difference (P < 0.01) between the models for both IRT task and well-being domains, as well as for all CTT subscales. Furthermore, the interaction term was significant for at least one cohort in all interaction models. Item Response Theory Domains of Vision-Specific Quality of Life Similar to the main effect models for IRT composite outcomes, all sociodemographic and clinical predictors had significant beta coefficients that were in the expected directions, except for health insurance for the IRT task outcome. The adjusted R-squared values were marginally higher, 0.356 for task and 0.317 for well-being. Assumptions of linear regression were also satisfied and there was no evidence of multicollinearity for IRT composite interaction models. The associations of VSQOL on VFL in the BSE varied by race for both IRT task and well- being domains (Table 12). For both IRT outcomes, Latinos had the strongest associations with VFL, and African Americans had the smallest associations of VSQOL with VFL. In Chinese Americans, associations of task and well-being domains with VFL were not statistically different from those in Latinos; for the task domain only, associations in Chinese Americans were significantly larger than associations in African Americans. Within each cohort, the associations with VFL were stronger for the task—rather than well-being—VSQOL domain. LOWESS plots of predicted IRT task and well-being scores on VFL in the BSE illustrate the associations by racial/ethnic group, which appear linear from 0 dB to -30 dB (Figure 12). Also, the intercept at 0 dB of VFL is larger for the task domain for all three racial/ethnic groups, suggesting that among participants with no VFL, their task specific VSQOL is higher than their socioemotional well- being VSQOL. 76 For the IRT task model, a 1 dB lower value of MD for VFL was associated with 0.750 (95% CI: 0.646, 0.854) lower VSQOL score for Latinos, 0.719 (95% CI: 0.567, 0.872) for Chinese Americans, and 0.496 (95% CI: 0.374, 0.617) for African Americans; assuming a 5-point difference in the NEI-VFQ-25 is important clinically, this may be interpreted as a clinically meaningful loss in task-related VSQOL was associated with 6.7 dB, 7.0 dB, and 10.1 dB of VFL in the BSE, respectively. The association was significantly stronger in Latinos compared to African Americans (P = 0.001), and in Chinese Americans compared to African Americans (P = 0.021). However, there was no statistical difference in the association between task related VSQOL and VFL for Latinos and Chinese Americans (P = 0.742). For the IRT well-being model, a 1 dB lower value of MD for VFL was associated with 0.666 (0.560, 0.772) lower VSQOL score for Latinos, 0.575 (95% CI: 0.419, 0.731) for Chinese Americans, and 0.387 (0.263, 0.511) for African Americans; therefore a clinically meaningful loss in well-being-related VSQOL was associated with 7.5 dB, 8.7 dB, and 12.9 dB of VFL in the BSE, respectively. The difference in associations between Latinos compared to African Americans reached statistical significance (P = 0.001). However, there was no statistically significant difference in the association between well-being-related VSQOL and VFL for Latinos and Chinese Americans (P = 0.330) or for Chinese Americans and African Americans (P = 0.060). 77 Table 12: Linear Regression for VSQOL IRT Composite Scores on VFL in the BSE in the MOCCaS (n = 14,570)* Vision-Specific Quality of Life, Item Response Theory ß (95% CI) Clinically Meaningful Difference in MD (dB)† P-Value‡ Latinos Chinese Americans African Americans Task Composite§ Latinos 0.750 (0.646, 0.854) 6.7 < 0.001 0.742 0.001 Chinese Americans 0.719 (0.567, 0.872) 7.0 < 0.001 0.021 African Americans 0.496 (0.374, 0.617) 10.1 < 0.001 Well-Being Composite|| Latinos 0.666 (0.560, 0.772) 7.5 < 0.001 0.330 0.001 Chinese Americans 0.575 (0.419, 0.731) 8.7 < 0.001 0.060 African Americans 0.387 (0.263, 0.511) 12.9 < 0.001 VSQOL = Vision-Specific Quality of Life; VFL = Visual Field Loss; BSE = Better Seeing Eye; MOCCaS = Multiethnic Ophthalmology Cohorts of California Study; 95% CI = 95% confidence interval; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; MD = Mean Deviation; IRT = Item Response Theory *VFL is presented as mean deviation score in decibels; VSQOL is assessed by IRT analysis of the NEI-VFQ-25. Linear regression models of VSQOL on VFL were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. †Regression coefficients were transformed per 5-point difference in HRQOL score, a clinically significant difference in VSQOL score. ‡P-values are from tests of significance for linear regression beta coefficients; P-values on the diagonal are for VFL main effects and those off the diagonal are for interactions between VFL and race/ethnicity. §IRT Task Composite was calculated from a graded response theory model of 12 items from near vision, distance vision, driving, color vision, peripheral vision, and role difficulties subscales. ||IRT Well-Being Composite was calculated from a graded response model of 12 items from general vision, dependency on others, mental health, ocular pain, and social functioning subscales. 78 Figure 12: LOWESS plot of predicted NEI-VFQ-25 IRT composite scores from linear regression on VFL in the BSE LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; IRT = Item Response Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye The LOWESS smoothing parameter is 0.6. Gray bars represent 95% confidence limits of the predicted NEI-VFQ-25 IRT composite scores. Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. 79 Classical Test Theory Subscales of Vision-Specific Quality of Life VFL had a statistically significant interaction with race/ethnicity in the association with VSQOL for all 11 CTT outcomes. Beta coefficients for VSQOL on VFL were obtained for Latinos (Table 13), Chinese Americans (Table 14), and African Americans (Table 15) by separately reassigning the baseline to each racial/ethnic group and rerunning the models, similar to the lincom statement in Stata software.137 CTT subscales with the strongest association with VFL were ranked in descending order. The top four subscales, excluding general health and vision, are reported for each cohort. CTT subscales most strongly associated with VFL overall included vision-related driving difficulties, role function, and mental health (Figure 13). Driving difficulties was most impacted by VFL in the BSE of all CTT subscales; it was the strongest association in Latinos and African Americans such that 1 dB lower value of MD for VFL was associated with 1.71 and 0.884 lower points of the VSQOL score, respectively, but was the third largest CTT subscale association in Chinese Americans (0.806). The association was statistically significantly larger in Latinos than Chinese and African Americans, but there was no significant difference in associations between the two latter groups. However, the linearity assumption of the linear regression model was violated, as participants had much lower observed driving difficulties scores than predicted at worse levels of VFL; this may suggest an exponential decline in driving related VSQOL with VFL. 80 Table 13: Linear Regression for the Association of VSQOL on VFL in the BSE among Latinos in MOCCaS Vision-Specific Quality of Life ß* (95% CI)* Clinically Meaningful Difference in MD (dB)** P-Value Main Effects Latinos P-Value Interaction VFL and Race/Ethnicity Chinese Americans African Americans Item Response Theory Composites Task Composite, IRT† 0.750 (0.646, 0.854) 6.7 < 0.001 0.742 0.001 Well-Being Composite, IRT‡ 0.666 (0.560, 0.772) 7.5 < 0.001 0.330 0.001 Classical Test Theory Subscales Overall Composite, CTT§ 0.970 (0.895, 1.045) 5.2 < 0.001 < 0.001 < 0.001 Driving Difficulties|| 1.712 (1.538, 1.886) 2.9 < 0.001 < 0.001 < 0.001 Vision-Related Dependency 1.382 (1.263, 1.502) 3.6 < 0.001 < 0.001 < 0.001 Peripheral Vision 1.151 (1.046, 1.257) 4.3 < 0.001 < 0.001 < 0.001 Vision-Related Mental Health 1.142 (1.000, 1.283) 4.4 < 0.001 0.037 < 0.001 Distance Vision 1.101 (1.004, 1.199) 4.5 < 0.001 < 0.001 < 0.001 Vision-Related Role Function 1.071 (0.932, 1.210) 4.7 < 0.001 0.795 < 0.001 Near Vision 0.920 (0.810, 1.030) 5.4 < 0.001 < 0.001 < 0.001 Vision-Related Social Function 0.911 (0.839, 0.984) 5.5 < 0.001 < 0.001 < 0.001 Color Vision 0.856 (0.780, 0.931) 5.8 < 0.001 < 0.001 < 0.001 Ocular Pain 0.721 (0.588, 0.853) 6.9 < 0.001 0.037 < 0.001 General Vision 0.287 (0.173, 0.400) 17.4 < 0.001 0.051 0.001 General Health Item General Health 0.337 (0.170, 0.504) 14.8 < 0.001 0.068 0.071 12-Item Short Form Composites Physical Composite 0.305 (0.250, 0.361) 16.4 < 0.001 0.122 < 0.001 Mental Composite 0.078 (0.023, 0.134) 63.8 0.005 0.013 0.158 n = 5,248; VSQOL = Vision-Specific Quality of Life; VFL = Visual Field Loss; BSE = Better Seeing Eye; MOCCaS = Multiethnic Ophthalmology Cohorts of California Study; 95% CI = 95% confidence interval; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; MD = Mean Deviation; IRT = Item Response Theory; CTT = Classical Test Theory; SF-12 = 12- Item Short-Form Health Survey *VFL is presented as mean deviation score in decibels; VSQOL is assessed by the NEI-VFQ-25; and health-related quality of life is assessed by the SF-12. Data are presented as coefficient (95% CI). The SF-12 and NEI-VFQ-25 scores are adjusted for age, gender, education, employment status, income, acculturation, co-morbidities, health insurance, vision insurance, and visual acuity impairment. There was an interaction term for race/ethnicity and VFL. **Regression coefficients were transformed per 5-point difference in HRQOL score, a clinically significant difference in VSQOL score. †IRT Task Composite was calculated from a graded response theory model of 12 items from near vision, distance vision, driving, color vision, peripheral vision, and role difficulties subscales. ‡IRT Well-Being Composite was calculated from a graded response model of 12 items from general vision, dependency on others, mental health, ocular pain, and social functioning subscales. §Composite score is an un-weighted mean of the 12 subscale scores (excluding general health). ||Scores could be generated for only 3,816 Latinos who reported that they were currently driving or had driven in the past. 81 Table 14: Linear Regression for the Association of VSQOL on VFL in the BSE among Chinese Americans in MOCCaS Vision-Specific Quality of Life ß* (95% CI)* Clinically Meaningful Difference in MD (dB)** P-Value Main Effects Chinese Americans P-Value Interaction VFL and Race/Ethnicity Latinos African Americans Item Response Theory Composites Task Composite, IRT† 0.719 (0.567, 0.872) 7.0 < 0.001 0.742 0.021 Well-Being Composite, IRT‡ 0.575 (0.419, 0.731) 8.7 < 0.001 0.330 0.060 Classical Test Theory Subscales Overall Composite, CTT§ 0.498 (0.388, 0.608) 10.0 < 0.001 < 0.001 0.001 Vision-Related Role Function 1.039 (0.835, 1.243) 4.8 < 0.001 0.795 < 0.001 Vision-Related Mental Health 0.881 (0.673, 1.088) 5.7 < 0.001 0.037 < 0.001 Driving Difficulties|| 0.806 (0.583, 1.029) 6.2 < 0.001 < 0.001 0.570 Near Vision 0.565 (0.404, 0.726) 8.9 < 0.001 < 0.001 0.103 Vision-Related Dependency 0.516 (0.341, 0.691) 9.7 < 0.001 < 0.001 < 0.001 Distance Vision 0.489 (0.347, 0.631) 10.2 < 0.001 < 0.001 0.001 General Vision 0.481 (0.316, 0.647) 10.4 < 0.001 0.051 0.445 Ocular Pain 0.478 (0.283, 0.672) 10.5 < 0.001 0.037 0.007 Peripheral Vision 0.196 (0.042, 0.351) 25.5 0.013 < 0.001 0.236 Vision-Related Social Function 0.15 (0.044, 0.256) 33.3 0.006 < 0.001 0.746 Color Vision 0.051 (-0.06, 0.162) 0.372 < 0.001 0.360 General Health Item General Health 0.068 (-0.18, 0.313) 0.583 0.068 0.001 12-Item Short Form Composites Physical Composite 0.23 (0.148, 0.311) 21.8 < 0.001 0.122 < 0.001 Mental Composite -0.04 (-0.12, 0.038) 0.296 0.013 < 0.001 n = 4,151; VSQOL = Vision-Specific Quality of Life; VFL = Visual Field Loss; BSE = Better Seeing Eye; MOCCaS = Multiethnic Ophthalmology Cohorts of California Study; 95% CI = 95% confidence interval; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; MD = Mean Deviation; IRT = Item Response Theory; CTT = Classical Test Theory; SF-12 = 12- Item Short-Form Health Survey *VFL is presented as mean deviation score in decibels; VSQOL is assessed by the NEI-VFQ-25; and health-related quality of life is assessed by the SF-12. Data are presented as coefficient (95% CI). The SF-12 and NEI-VFQ-25 scores are adjusted for age, gender, education, employment status, income, acculturation, co-morbidities, health insurance, vision insurance, and visual acuity impairment. There was an interaction term for race/ethnicity and VFL. **Regression coefficients were transformed per 5-point difference in HRQOL score, a clinically significant difference in VSQOL score. †IRT Task Composite was calculated from a graded response theory model of 12 items from near vision, distance vision, driving, color vision, peripheral vision, and role difficulties subscales. ‡IRT Well-Being Composite was calculated from a graded response model of 12 items from general vision, dependency on others, mental health, ocular pain, and social functioning subscales. §Composite score is an un-weighted mean of the 12 subscale scores (excluding general health). ||Scores could be generated for only 3,047 Chinese Americans who reported that they were currently driving or had driven in the past. 82 Table 15: Linear Regression for the Association of VSQOL on VFL in the BSE among African Americans in MOCCaS Vision-Specific Quality of Life ß* (95% CI)* Clinically Meaningful Difference in MD (dB)** P-Value Main Effects African Americans P-Value Interaction VFL and Race/Ethnicity Latinos Chinese Americans Item Response Theory Composites Task Composite, IRT† 0.496 (0.374, 0.617) 10.1 < 0.001 0.001 0.021 Well-Being Composite, IRT‡ 0.387 (0.263, 0.511) 12.9 < 0.001 0.001 0.060 Classical Test Theory Subscales Overall Composite, CTT§ 0.261 (0.174, 0.349) 19.1 < 0.001 < 0.001 0.001 Driving Difficulties|| 0.884 (0.725, 1.044) 5.7 < 0.001 < 0.001 0.570 General Vision 0.562 (0.430, 0.694) 8.9 < 0.001 0.001 0.445 Near Vision 0.397 (0.269, 0.525) 12.6 < 0.001 < 0.001 0.103 Peripheral Vision 0.313 (0.190, 0.436) 16.0 < 0.001 < 0.001 0.236 Vision-Related Mental Health 0.263 (0.098, 0.428) 19.0 0.002 < 0.001 < 0.001 Vision-Related Role Function 0.191 (0.029, 0.353) 26.2 0.021 < 0.001 < 0.001 Distance Vision 0.181 (0.068, 0.294) 27.6 0.002 < 0.001 0.001 Ocular Pain 0.143 (-0.01, 0.297) 34.9 0.069 < 0.001 0.007 Vision-Related Social Function 0.128 (0.044, 0.212) 39.0 0.003 < 0.001 0.746 Vision-Related Dependency 0.112 (-0.03, 0.251) 44.5 0.113 < 0.001 < 0.001 Color Vision -0.01 (-0.10, 0.074) 0.753 < 0.001 0.360 General Health Item General Health 0.564 (0.370, 0.759) 8.9 < 0.001 0.071 0.001 12-Item Short Form Composites Mental Composite 0.137 (0.072, 0.202) 36.4 < 0.001 0.158 < 0.001 Physical Composite -0.18 (-0.24, -0.109) < 0.001 < 0.001 < 0.001 n = 5,171; VSQOL = Vision-Specific Quality of Life; VFL = Visual Field Loss; BSE = Better Seeing Eye; MOCCaS = Multiethnic Ophthalmology Cohorts of California Study; 95% CI = 95% confidence interval; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; MD = Mean Deviation; IRT = Item Response Theory; CTT = Classical Test Theory; SF-12 = 12- Item Short-Form Health Survey *VFL is presented as mean deviation score in decibels; VSQOL is assessed by the NEI-VFQ-25; and health-related quality of life is assessed by the SF-12. Data are presented as coefficient (95% CI). The SF-12 and NEI-VFQ-25 scores are adjusted for age, gender, education, employment status, income, acculturation, co-morbidities, health insurance, vision insurance, and visual acuity impairment. There was an interaction term for race/ethnicity and VFL. **Regression coefficients were transformed per 5-point difference in HRQOL score, a clinically significant difference in VSQOL score. †IRT Task Composite was calculated from a graded response theory model of 12 items from near vision, distance vision, driving, color vision, peripheral vision, and role difficulties subscales. ‡IRT Well-Being Composite was calculated from a graded response model of 12 items from general vision, dependency on others, mental health, ocular pain, and social functioning subscales. §Composite score is an un-weighted mean of the 12 subscale scores (excluding general health). ||Scores could be generated for only 4,610 African Americans who reported that they were currently driving or had driven in the past. 83 Figure 13: Linear regression beta coefficients of NEI-VFQ-25 IRT and CTT on VFL in the BSE by cohort LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; IRT = Item Response Theory; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. 84 Vision-related role function had the next strongest association with VFL (Figure 14), which was most pronounced in Chinese Americans with a 1 dB lower value of MD for VFL associated with 1.039 (95% CI: 0.835, 1.243) lower points of VSQOL. Role function had the fifth strongest association with VFL in both Latinos at 1.071 (95% CI: 0.932, 1.210), and African Americans at 0.191 (95% CI: 0.029, 0.353). The associations were significantly larger in Chinese Americans and Latinos compared to African Americans, but there was no significant difference between the former groups. Vision-related mental health was the third largest association between VSQOL and VFL overall (Figure 15); it was ranked the fourth largest association in Latinos at 1.142 (95% CI: 1.000, 1.283), the second largest in Chinese Americans at 0.881 (95% CI: 0.673, 1.088), and fourth largest in African Americans at 0.263 (95% CI: 0.098, 0.428); there were statistically significant differences in the associations between vision-related mental health and VFL for all three cohorts. 85 Figure 14: LOWESS plot of predicted NEI-VFQ-25 CTT role function scores from regression on VFL in the BSE LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye The LOWESS smoothing parameter is 0.6. Gray bars represent 95% confidence limits of the predicted NEI-VFQ-25 IRT composite scores. Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. LOWESS curves are shown stratified by age < 65 and ≥ 65 to illustrate effect modification of VSQOL on VFL by age. 86 Figure 15: LOWESS plot of predicted NEI-VFQ-25 CTT mental health scores from regression on VFL in the BSE LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye The LOWESS smoothing parameter is 0.6. Gray bars represent 95% confidence limits of the predicted NEI-VFQ-25 IRT composite scores. Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. LOWESS curves are shown stratified by age < 65 and ≥ 65 to illustrate effect modification of VSQOL on VFL by age. 87 Several CTT subscales were most strongly associated with VFL in certain racial/ethnic differences, but not others. For example, role function was uniquely important in Chinese Americans. In addition, vision-related dependency (Figure 16) was the second strongest association in Latinos at 1.382 (95% CI: 1.263, 1.502), but the fifth in Chinese Americans 0.516 (95% CI: 0.341, 0.691), and was not significantly associated with VFL in African Americans 0.112 (95% CI: -0.027, 0.251). Furthermore, near vision (Figure 17) was the second largest association with VFL in African Americans at 0.397 (95% CI: 0.269, 0.525) and fourth in Chinese Americans at 0.565 (95% CI: 0.404, 0.726), but seventh in Latinos at 0.920 (95% CI: 0.810, 1.030). Finally, peripheral vision (Figure 18) was the third strongest association for Latinos at 1.151 (95% CI: 1.046, 1.257) and African Americans at 0.313 (95% CI: 0.190, 0.436), but the ninth strongest association in Chinese Americans at 0.196 (95% CI: 0.042, 0.351). We only described the CTT subscales of VSQOL that had the largest ranked associations with VFL within and across all three cohorts, although other subscales were significantly associated with VFL in the BSE (Figure 19). 88 Figure 16: LOWESS plot of predicted NEI-VFQ-25 CTT dependency scores from regression on VFL in the BSE LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye The LOWESS smoothing parameter is 0.6. Gray bars represent 95% confidence limits of the predicted NEI-VFQ-25 IRT composite scores. Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. LOWESS curves are shown stratified by age < 65 and ≥ 65 to illustrate effect modification of VSQOL on VFL by age. 89 Figure 17: LOWESS plot of predicted NEI-VFQ-25 CTT near vision scores from regression on VFL in the BSE LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye The LOWESS smoothing parameter is 0.6. Gray bars represent 95% confidence limits of the predicted NEI-VFQ-25 IRT composite scores. Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. LOWESS curves are shown stratified by age < 65 and ≥ 65 to illustrate effect modification of VSQOL on VFL by age. 90 Figure 18: LOWESS plot of predicted NEI-VFQ-25 CTT peripheral vision scores from regression on VFL in the BSE LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye The LOWESS smoothing parameter is 0.6. Gray bars represent 95% confidence limits of the predicted NEI-VFQ-25 IRT composite scores. Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. LOWESS curves are shown stratified by age < 65 and ≥ 65 to illustrate effect modification of VSQOL on VFL by age. 91 Figure 19: Linear regression beta coefficients of all QOL scales on VFL in the BSE by cohort LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; IRT = Item Response Theory; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. 92 Age Interaction Models A larger model with VFL interacting with both race/ethnicity and age—dichotomized at 65 years—is shown in Table 16. Using the multivariate Wald tests, we found evidence of statistical interaction with age (P < 0.01) for both IRT task and well-being domains, as well as for all CTT subscales. However, the VFL-age interaction terms only reached statistical significance for the CTT driving outcome of VSQOL regressed on VFL in the BSE (P < 0.01), not for the IRT domains or the 10 other CTT subscale outcomes. These data suggest that only the CTT driving difficulties subscale association with VFL was modified by both age and race/ethnicity. For each 1 dB of MD in VFL, there was an associated 0.487 point (95% CI: 0.279, 0.695) greater loss in driving difficulties VSQOL for those 65 years and older (P for interaction < 0.001) (Table 16). Predicted VSQOL stratified by race/ethnicity and age illustrates a larger association with VFL in participants more than 65 years old compared to those younger (Figure 20) across all three racial/ethnic groups. For both young and old, the associations of VSQOL and VFL were not statistically different for African Americans and Chinese Americans but were greater in Latinos (Figure 21). 93 Table 16: Linear Regression for the Association Between CTT Driving Difficulties and VFL in the BSE in MOCCaS* Driving Difficulties† (Classical Test Theory) ß (95% CI) Clinically Meaningful Difference in MD (dB)‡ P-Value§ Interaction by Age Age ≥ 65 Years 0.487 (0.279, 0.695) 10.3 < 0.001 Interaction by Age and Race/Ethnicity Latinos Chinese Americans African Americans Age < 65 Years Latinos 1.502 (1.306, 1.698) 3.3 < 0.001 < 0.001 < 0.001 Chinese Americans 0.654 (0.422, 0.887) 7.6 < 0.001 0.865 African Americans 0.631 (0.435, 0.827) 7.9 < 0.001 Age ≥ 65 Years Latinos 1.989 (1.779, 2.198) 2.5 < 0.001 < 0.001 < 0.001 Chinese Americans 1.141 (0.878, 1.404) 4.4 < 0.001 0.865 African Americans 1.117 (0.932, 1.303) 4.5 < 0.001 n = 11,473; VSQOL = Vision-Specific Quality of Life; VFL = Visual Field Loss; BSE = Better Seeing Eye; MOCCaS = Multiethnic Ophthalmology Cohorts of California Study; 95% CI = 95% confidence interval; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-Item; MD = Mean Deviation; IRT = Item Response Theory; CTT = Classical Test Theory; SF-12 = 12- Item Short- Form Health Survey *VFL is presented as mean deviation score in decibels; VSQOL is assessed by the NEI-VFQ-25. Data are presented as coefficient (95% CI). The CTT Driving Difficulties subscale of the NEI-VFQ- 25 scores are adjusted for age, gender, education, employment status, income, acculturation, co- morbidities, health insurance, vision insurance, and visual acuity impairment. There was an interaction term for race/ethnicity and VFL, and for age and VFL. †Scores could be generated for only 11,473 of participants who reported that they were currently driving or had driven in the past. ‡Regression coefficients were transformed per 5-point difference in HRQOL score, a clinically significant difference in VSQOL score. §P-values are from tests of significance for linear regression beta coefficients; P-values on the diagonal are for VFL main effects and those off the diagonal are for interactions between VFL and race/ethnicity. 94 Figure 20: LOWESS plot of predicted NEI-VFQ-25 CTT driving difficulties scores from regression on VFL in the BSE LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye The LOWESS smoothing parameter is 0.6. Gray bars represent 95% confidence limits of the predicted NEI-VFQ-25 IRT composite scores. Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. LOWESS curves are shown stratified by age < 65 and ≥ 65 to illustrate effect modification of VSQOL on VFL by age. 95 Figure 21: Linear regression of NEI-VFQ-25 CTT driving difficulties on VFL in the BSE by age group LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. 96 Sensitivity analyses Sensitivity analyses for VSQOL IRT outcomes in subgroups with no VI (n = 13,663) and no depression (n = 13,134) were conducted using the model used to test the interaction of race/ethnicity and VFL only. The analysis with no VI found similar results as that of the full analytic cohort, except the differences by race were marginally magnified (Figure 22). There was, however, no change in the statistical significance of results by racial/ethnic group; the association was still the strongest in Latinos, and there was no difference in the relatively smaller associations between Chinese and African Americans. The only change from the full model was the association in Chinese Americans, which were slightly weaker for IRT outcomes. For IRT task on VFL, regression coefficients were reduced from 0.719 (95% CI: 0.567, 0.872) to 0.533 (95% CI: 0.361, 0.705), and for IRT well-being on VFL, regression coefficients were reduced from 0.575 (95% CI: 0.419, 0.731) to 0.423 (95% CI: 0.247, 0.599); this constitutes a 25% decrease for both IRT outcomes. The analysis with no depression produced beta coefficients that were nearly identical to those in the full analytic cohort for both IRT task and well-being outcomes (Figure 23). 97 Figure 22: Regression of VSQOL IRT composite domains on VFL in the BSE excluding visual impairment* n = 13,663; LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye *The sensitivity analysis excluded all participants with visual impairment, defined as visual acuity of 20/40 or worse (n excluded = 907). The confidence intervals that are faded in color represent the associations and confidence intervals of the overall analytic cohort (n = 14,570). Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. 98 Figure 23: Regression of VSQOL IRT composite domains on VFL in the BSE excluding self- reported depression* n = 13,134; LOWESS = Locally Weighted Scatterplot Smoothing; NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire 25-ltem; CTT = Classical Test Theory; VFL = Visual Field Loss; MD = Mean Deviation; dB = Decibels; BSE = Better-Seeing Eye *The sensitivity analysis excluded all participants with self-reported depression (n excluded = 1,436). The confidence intervals that are faded in color represent the associations and confidence intervals of the overall analytic cohort (n = 14,570). Linear regression models were adjusted for race/ethnicity, age, number of comorbidities, sex (female), born in USA (yes), education highest grade obtained), working status (unemployed), income (≤ $20,000), has health insurance (yes), presenting binocular visual acuity (LogMAR score), depression (a good bit of the time or more in the last 4 weeks), and an interaction between VFL and race/ethnicity. 99 Discussion Participation of racial/ethnic minorities in clinical research is important for identifying social determinants of health, achieving health equity, and generalizing medical findings to an increasingly diverse population.138 Multiethnic population-based epidemiologic studies of chronic diseases can elucidate health disparities. Racial/ethnic differences in health outcomes have largely been explained by differences in socioeconomic status (SES).139 For example, the Salisbury Eye Study identified disparities in visual impairment among African Americans and non-Hispanic Whites and found these differences were largely explained by cataracts, a surgically treatable condition for those with sufficient resources to access care.140 More generally, SES is a primary determinant of racial/ethnic disparities in overall mortality from specific health conditions.141,142 However, it should be noted that race and ethnicity do not have universally agreed upon definitions, are social constructs that depend largely upon phenotypic traits, and there is overall more genetic variation between racial/ethnic groups than within them. Personalized medicine may one day allow patients’ genotypes to drive precise clinical decision making and render racial/ethnic studies of health outcomes obsolete. But genetically tailored medical treatments are still in their nascency and far from standard, particularly in healthcare for eye disease. In this study, we did not assess the genetic makeup of participants. However, even after controlling for SES-related variables, we observed statistically and clinically meaningful differences across self-identified racial/ethnic groups in the association of VFL—an objective measure of VI—and how participants subjectively perceive their QOL related to their vision. 100 Summary of Findings We hypothesized that VFL variably impacts VSQOL among the most populous racial/ethnic groups in the United States, which we assessed using data from MOCCaS, a pooled analysis of eye disease from three large, population-based cohort studies. We found racial/ethnic differences in the associations between VFL and IRT scores for vision tasks and socioemotional well-being. The impact of VFL on participants ability to complete vision-specific tasks was greater for Latino and Chinese Americans compared to African Americans, after adjusting for covariates. We also found the impact of VFL on self-reported socioemotional well-being scores was greater among Latino participants compared to African American participants after controlling for covariates; however, effects in Chinese American participants were intermediate and not statistically significantly different from either group. We used the Hodapp-Parrish- Anderson method to classify participants’ VFL by severity, which categorizes VFL as mild (> -6 dB), moderate (-6 dB to -12 dB), and severe VFL (< -12 dB).143 After adjusting for visual acuity and sociodemographic factors, we found clinically important losses in vision-specific task scores were found at moderate VFL for all racial/ethnic groups. For socioemotional well-being scores, African Americans reported clinically meaningful losses only when reaching severe VFL, but a decline in socioemotional well-being scores for Latinos and Chinese Americans were found as early as mild VFL. We hypothesized a priori, that VFL would have the greatest impact on patients’ ability to drive for all racial/ethnic groups, due to the consistent and strong associations reported in prior literature. Previous studies have found that the driving difficulties VSQOL scale was inversely associated with VFL, independent of central VA loss.20,21,23 In the MOCCaS, VFL had the strongest association with driving difficulties of all VSQOL subscales among Latinos and African Americans, and represented the third largest association among Chinese Americans. Furthermore, a clinically meaningful 5-point lower driving difficulty score was found for mild 101 levels of VFL in Latinos (2.9 dB) and African Americans (5.7 dB), but not observed until moderate VFL in Chinese Americans (6.2 dB). There was evidence, however, that the linear models underestimated the potential impact of VFL on driving related VSQOL among those with the most extreme VFL ; this may suggest an exponential decrease in driving difficulties and VFL. Finally, we hypothesized that VFL would more greatly impact VSQOL in participants aged 65 years and older compared to their younger counterparts due to compounding comorbidities. However, we found no evidence that VFL impacted task or well-being VSQOL differentially by age group. For individual CTT subscales of VSQOL we exclusively found evidence of effect modification by age for vision-related driving difficulties. In older participants, driving difficulties VSQOL scores were lower by about half a point more for every 1 dB of VFL. Across all racial/ethnic groups, a clinically meaningful decrement in driving VSQOL was observed at approximately 1 dB to 3.5 dB less VFL in older participants. Perhaps aging is associated with other functional losses that otherwise protect younger adults with VFL from losing their ability to drive; cognitive function and hearing are several of many factors that decline with age that may explain the greater association of VFL on driving-related VSQOL.115 Shared Findings Across Racial/Ethnic Groups We observed similarities across all racial/ethnic groups. The VSQOL task, driving difficulties, and vision-related mental health subscales had associations with VFL of similar magnitudes for Latinos, Chinese Americans and African Americans. Perhaps VFL impacts these aspects of VSQOL in people of all races and ethnicities by a common mechanism. We did find a statistically significant interaction for VFL and the IRT vision-related task composite for Latinos compared to African Americans, however clinically, the presence of moderate VFL was required among all racial/ethnic groups to detect a clinically important change in vision-related task scores. This suggests that losses in self-reported ability to complete vision-tasks related to VFL 102 are effectively similar across racial/ethnic groups. More specifically, driving is a complex activity that requires intact vision, cognition, and reaction time, among other abilities; the ability to drive during the day and at night may be directly related to visual impairment with less influence by racial/ethnic differences. Although vision-related driving difficulties were highly associated with VFL for all racial/ethnic groups, Latinos were most strongly impacted among those aged both younger and older than 65 years. With respect to the IRT socioemotional well-being scores, vision-related mental health also was equally affected by VFL across racial/ethnic groups. The mental health subscale is related to worry, frustration, loss of control and embarrassment due to vision loss, which may also be closely related to VFL without modification by racial/ethnic variations. While we identified some statistically significant differences in scales (vision-related task and driving), we must acknowledge that similarities were more common, and many statistically significant differences may not have been clinically large. Distinct Findings in Each Racial/Ethnic Group Racial/ethnic variations in the impact of VFL on VSQOL were observed for the IRT well- being domain and several CTT subscales. A clinically meaningful difference in vision-related well-being was associated with moderate VFL in Latinos and Chinese Americans, but with severe VFL in African Americans. While our study interviewers, doctors, and staff developed a working relationship of trust in the AFEDS community, it is possible that African Americans may be less comfortable reporting complaints about their vision-related well-being because of lasting sentiments of mistrust of medical services. African Americans may mistrust the healthcare system due to historical events including the Tuskegee Study and institutionalized discriminatory health practices historically,138,144 and in fact early participants in the study brought up concern about participation related to these historical events. However, we did have high participation in each of the 3 communities and successful recruitment happened in part 103 because of positive experiences communicated by participants to their neighborhood friends and family members. Further, we would expect Latinos and Chinese Americans, who are largely first-generation immigrants, to potentially have similar reservations about disclosure of personal information, yet we saw variation in Likert scale responses between these groups. Without data for non-Hispanic Whites, we were not able to assess whether visual impairment and VSQOL in these racial/ethnic groups differed compared to non-Hispanic whites in the US population. In Chinese Americans and Latinos there were no racial/ethnic differences in the IRT based task and well-being domains of VSQOL and VFL, but there were racial/ethnic differences when using a narrower focus on CTT subscales. Exclusive to Chinese Americans, we found a strong association between VFL and vision-related role function. Role function includes questions on how long people are able to work and how much they are able to accomplish due to their vision (Table 10). This association was even stronger than that of driving difficulties among Chinese American participants. Perhaps multigenerational family structures may better provide for Chinese Americans with respect to VFL and vision-related tasks in the home; a relative who is able to drive may be able to transport them, thus protecting them from perceived losses to vision-related driving limitations. This difference may reflect cultural differences in Chinese Americans compared to Latinos and African Americans. This remains only a point of discussion here as we did not collect data on family structure and support beyond marital status. Focus groups may help elucidate why decrements in role function were singularly related to VFL in Chinese Americans. Latinos reported greater dependency on others due to their vision compared to African and Chinese Americans. These uniquely large associations between VFL and dependency may be explained by familial or cultural differences, which focus groups or follow-up interviews may be better able to elucidate in this study. 104 Clinical and Public Health Impact of VFL on VSQOL In the present study, we observed race and ethnicity were important modifiers of the relationship between VFL and VSQOL for some domains, but not for others. Providers caring for patients of all races and ethnicities with VFL might advise patients that they may be most affected in their ability to drive and their mental health—worry, frustration, loss of control, embarrassment—related to their vision. Furthermore, patients over 65 years of age may be particularly sensitive to losses in visual field related to their driving abilities. Participants in all racial/ethnic groups reported lower satisfaction for items related to vision-related well-being compared to vision-related tasks, even among individuals reporting mild or no VFL. However, in those with moderate and severe VFL, losses in task and well-being VSQOL were similarly low. The incremental relationship between VFL and vision-related task scores was stronger than for vision-related well-being. Of note, individual variation in VSQOL was stronger than differences by race/ethnicity. The NEI-VFQ-25 should be administered to each patient in order to assess individual patient’s subjective experiences related to their visual health. The association between VFL and IRT well- being scores were weaker for African American patients than Chinese American and Latino patients with the same amount of VFL. However, participants of all racial/ethnic groups appear to report similar levels of VFL with IRT vision-related task scores . Clinical trials have used health-related QOL as an outcome to assess the utility of medical treatments. QOL is an especially important outcome in clinical trials when the intent of the intervention is to facilitate communication between patients and providers, to relieve symptoms, or to palliate rather than cure disease.120 QOL is also useful for differentiating between multiple treatment options that have all been shown to be safe and efficacious. Furthermore, interventions may have no biological effect for patient health but may considerably improve QOL. Therefore QOL has and will continue to be an important outcome in deciding which 105 treatment options are available to the public. VSQOL is no less important to assessing treatments for chronic eye diseases. VSQOL has been useful as a measure of function to indicate at what level of severity patients perceive a change in their ability to complete daily tasks or be socially and emotionally well due to changes in their eye health and vision. Measures of VSQOL in clinical trials continues to shape ophthalmic practice. A meta- analysis comparing multifocal and monofocal lenses for treating cataracts reported 5 of 8 studies used VSQOL or non-validated PRO for vision as an outcome.121 The variation in instruments used to measure VSQOL reflects the need for standardized methods for assessing patient experiences of vision. More recent clinical trials have used the NEI-VFQ-25 to measure outcomes for glaucoma treatments. One international trial showed clear lens extraction was the superior intervention for primary angle closure glaucoma using VSQOL as one of the outcomes;98 the improved subjective patient experience mirrored in the intervention arm mirrored the similar increase in visual function compared to those receiving standard care. Participants of this study were of Chinese (61%) and non-Chinese origin. A large meta-analysis assessed the effect of vision rehabilitation programs on QOL, which included a diverse pooled patient population from developed countries, mostly suffering from age-related macular degeneration.122 VSQOL—but not health-related QOL—had small improvements in rehabilitation programs that employed psychological therapies and taught how to use vision- improving devices like magnifying glasses. Without measures of VSQOL, it would have appeared that there were no improvements in QOL, and vision rehabilitation programs may have been disregarded. With the overarching impact of VSQOL and health-related QOL in medical care, it is essential that large racial/ethnic groups are represented in study patient populations to ensure findings are externally valid for the diverse general population. 106 Limitations There are several limitations of this work despite the many strengths of population-based cohort studies in assessing representative associations of exposure and chronic disease. Foremost, there were differences in excluded data across the cohorts, which may bias the observed findings for each racial/ethnic group (Table 10). 14.6% of Latinos were excluded from the LALES data, primarily due to incomplete VSQOL data. 18.5% of African Americans were excluded from the AFEDS analysis, but due to incomplete or unreliable VFL data. However, the NEI-VFQ-25 was only implemented in the LALES after a year of the study, therefore there is no reason this might select for participants with different VFL or VSQOL. Furthermore, the AFEDS cohort was older, which may explain why they had less complete VFL measurements; the VF analyzer requires sustained attention for at least 20 minutes per eye, and there were multiple measurements per eye, which may have been more difficult for older participants with greater comorbidities to complete. Chinese Americans were the least likely to be excluded at 9.4%, which was mostly due to incomplete VFL measurements. But the CHES only included participants aged 50 years and older; a large difference from the other two cohorts, which included those aged 40 years and older. However, all models were adjusted for age. The period of data collection across the three population-based cohort studies that comprise the MOCCaS may influence perception of importance of visual field loss. The LALES was completed in the early 2000s, the CHES in the early 2010s, and the AFEDS in the mid 2010s. Temporal differences could contribute to the variation that was observed across the cohorts rather than racial/ethnic differences. Similarly, the cohorts were performed in three distinct cities, therefore geographical and SES differences may also partially explain variation attributed to race/ethnicity. However we believe geographical differences are minimal as all cohorts are within Los Angeles County and values were adjusted for measures of SES. 107 Future work may involve contacting participants from MOCCaS to conduct a follow up survey or focus group to understand why we observed the racial/ethnic differences in each cohort. In addition, longitudinal cohorts might assess how changes to VI impacts VSQOL differentially by race and ethnicity. The LALES includes longitudinal data, but this data remains to be collected for both CHES and AFEDS. External Validity The baseline methods papers of all three cohorts in the MOCCaS analysis have demonstrated that the study populations are representative of Latino, Chinese American, and African American populations in the US overall.51,56,57 We believe that the findings presented here stands upon previous claims and are externally valid for racial/ethnic groups in the State of California and the US overall. 108 Chapter 4: The Association of Traffic-Related Air Pollution with Reduced Blood Perfusion through Peripapillary Capillaries of the Retina in African Americans Abstract Importance Traffic related air pollution (TRAP) has been associated with greater self-reported glaucoma and related structural changes in the retina. Of the limited studies of air pollution and progressive glaucomatous disease, none have assessed the association between air pollution and the radial peripapillary capillaries (RPCs) that supply the retinal nerve fiber layer. Objective To determine whether acute and chronic TRAP is associated with lower perfusion of the RPC in a population-based sample of African Americans, a group which endures a higher prevalence of open angle glaucoma and exposure to pollution. Study Design The African American Eye Disease Study (AFEDS) is a cross-sectional, population-based cohort study of eye disease in Southern California that was conducted from 2014-2018. Participants completed in-home interviews to ascertain sociodemographic information, as well as detailed clinical eye exams, including optical coherence tomography angiography (OCTA) of the retina. Hierarchical linear regression modeling was used to assess the relationship between VAD and TRAP. 109 Participants 6,347 self-reported African Americans aged 40 years or older residing in 32 US census tracts of Inglewood, California. Participants were excluded with incomplete or poor quality OCTA images, glaucoma, or diabetic retinopathy. Exposures Daily maximum nitrogen dioxide (NO2 [ppb]) and daily mean particulate matter with aerodynamic diameter less than 2.5 µm (PM2.5 [µg/m3]). Estimates of TRAP exposure were produced from generalized additive models (GAMs) created using the Environmental Protection Agency’s Air Quality System data for Los Angeles and Orange County from 2014 through 2018. Main Outcome Measure Vessel area density (VAD) calculated from 6x6 mm images of the peripapillary, optic nerve head (ONH) on spectral-domain OCTA with complex signal-based optical microangiography. VAD represents the percentage of white pixels on the binarized image, which indicate motion- contrast of red blood cells compared to stationary neurosensory tissues. The hypothesis was generated after data collection had been completed. Results Participants who met exclusion criteria (n = 1,009; 15.9%) were on average 58.3 years old, 64.2% female, completed 14.3 years of education, were half employed, and 61% had annual household income ≥ $40,000. Adjusting for sociodemographic and clinical outcomes, lower VAD was statistically significantly associated (P < 0.05) with acute NO2 exposure the week before, and chronic PM2.5 exposure the 6 months before the eye exam. None of the associations in the fully adjusted model were significant after including signal strength—a quality control measure of OCTA imaging. Yet acute NO2 and chronic PM2.5 associations with VAD were in the expected directions. In the fully adjusted model, the percentage of VAD was -0.160 (95% CI: -0.377, 110 0.056) lower for every 10-ppb increase in NO2 over the week before the eye exam. The percentage of VAD was -0.271 (95% CI: -0.588, 0.0453) lower for every 10 µg/m3 increase in PM2.5. These differences in VAD equated to 1.58 (95% CI: -0.730, 2.98) years older age for NO2, and 2.68 (95% CI: -0.593, 4.66) years for PM2.5. Conclusions and Relevance We found evidence of a small, inverse association between NO2 and PM2.5 with VAD of retinal RPCs after adjusting for sociodemographics and clinical covariates. Although a similar pattern persisted for acute and chronic pollution exposures, the relationship between TRAP and VAD was not significant in the fully adjusted model. Non-significant findings may be due to a small true effect size, reduced power from a high exclusion rate in the analytic cohort, and limited predictive value for spatiotemporal estimates of air pollution exposure. This work fits within an emerging body of evidence that TRAP may be related to microvascular changes of the retina and could partially explain disparities in open angle glaucoma for African Americans. 111 Introduction Traffic related air pollution (TRAP) is a complex mixture of gases and particles that has been linked with many adverse health outcomes. Vascular diseases associated with TRAP include cardiovascular disease145–149, ischemic and hemorrhagic stroke,150 and vascular Alzheimer’s Disease.151 Both toxicologic and epidemiologic studies have demonstrated that components of TRAP— nitrogen dioxide (NO2) and fine particulate matter with aerodynamic diameter less than 2.5 micrometers (PM2.5)—are associated with endothelial dysfunction.152,153 Although the eye is a highly vascular organ, a limited number of studies have considered the association of environmental air pollution and the microvasculature of the eye.154,155 The retina is the interior structure of the posterior eye containing photochemical cells—rods and cones—that transduce light into electrochemical signals interpreted by the brain as vision. The retina contains an expansive network of blood vessels that provides nutrients and oxygen and clears metabolic waste necessary for the physiology of vision. Population-based studies have found among medium-large retinal vessels (60-300 µm), narrower arterial and wider venular diameters are associated with increased cardiovascular disease and mortality.156–158 Two areas of the retina with the greatest density of smaller retinal capillaries include the macula and the peripapillary region surrounding the head of the optic nerve are (Figure 24A). Vessel density of small (< 32 µm) radial peripapillary capillaries (RPCs) has been demonstrated to be a useful indicator for disease progression in glaucoma.159,160 Furthermore, glaucomatous structural changes in retinal nervous—but not vascular—tissues have been observed in patients living in areas of greater NO2 and PM2.5 exposure.161 African Americans are both more likely to have a greater prevalence of open angle glaucoma, and live in areas of higher air pollution exposure.162–164 The prevalence of primary open-angle glaucoma (POAG) is more than twice as high in African Americans older than 75 years compared to non-Hispanic Whites.4 Globally, patients 112 with POAG are 2.80 (OR; 95% CI, 1.83-4.06) times as likely to have African compared to European ancestry.165 Genetic studies have not sufficiently explained this difference, although few studies have been conducted.166 Racial/ethnic variations in air pollution exposures may in part explain the observed racial/ethnic differences in POAG prevalence, which has previously been associated with disparities in cardiovascular disease,163 developmental disease,167,168 childhood asthma,169,170 and cancer.171 TRAP is of particular importance to Los Angeles County, CA where impoverished communities and African Americans often live near congested freeways enduring disproportionately higher exposure to air pollution.172,173 However no studies have assessed how air pollution may affect the RPCs that are diminished during the disease process of glaucoma. Optical coherence tomography angiography (OCTA) is an imaging modality that measures movement of red blood cells flowing through retinal vessels and capillaries.174 Vessel area density (VAD) is a parameter that represents the overall blood flow in an OCTA image, which is an important quantitative measure of blood perfusion through the peripapillary vessel for monitoring glaucoma. Our study is the first cross-sectional, population-based study that uses OCTA to assess whether traffic-related air pollution is associated with microvascular differences in small RPCs of the retina among healthy African Americans. Furthermore, AFEDS is the largest dedicated eye-study of African Americans, a population that is disproportionately impacted by both environmental air pollution exposure and glaucomatous disease. We hypothesized that acute and chronic exposure to NO2 and PM2.5 is inversely associated with retinal blood perfusion of the radial peripapillary microvasculature (< 32 µm) as measured in African Americans using OCTA. 113 Figure 24: Images of the Retina and Retinal Blood Vessels A: Anatomy of the Eye B: Ophthalmoscope C: Optical Coherence Tomography 114 Methods The African American Eye Disease Study (AFEDS) is a population-based, cross-sectional cohort of self-identified African Americans aged 40 years and older living in Inglewood, California of Los Angeles County, conducted from 2014-2018. Further details of the study are published elsewhere.25–30 In brief, sociodemographic and medical histories were obtained by interviews in participants’ homes, including data on education, employment, income, smoking status, health insurance, vision insurance, and disease status. On a later date, comprehensive eye exams were conducted at the clinic to obtain data on systemic biomarkers, systemic clinical diagnoses, ophthalmic clinical diagnoses, and OCTA imaging of the peripapillary region. OCTA was introduced in February 2016—partway through the AFEDS—and measurements were completed for a subset of the overall cohort. Eyes of participants were excluded for OCTA measurements with poor image quality or signal strength less than 8 as well as for diagnoses of either glaucoma or diabetic retinopathy. One eye was randomly selected from participants with binocular measurements. The University of Southern California Health Sciences Review Board granted institutional review board/ethics committee approval. The study adheres to the Declaration of Helsinki. All participants gave written and informed consent before participating. Retinal Blood Flow Outcome The protocol for measuring retinal blood flow using OCTA in the AFEDS has been reported previously.175 In brief, perfusion of the radial peripapillary capillaries (RPC) was measured using a spectral domain OCTA (SD-OCTA) device with complex signal-based optical microangiography (CIRRUS HD-OCT 5000 with AngioPlex OCT; Zeiss, Dublin, CA, USA). Images were 6x6 mm and centered on the optic nerve head (ONH). Commercial and developmental algorithms were used to analyze the RPC images in the AFEDS, which have been described previously.175 Motion signal was produced from intensity and phase information between sequential B-scans. An automated segmentation software (CIRRUS 11.0; ZEISS) was 115 used to measure the gradient of OCT signals on cross section, which identified the perpendicular limits of the retinal nerve fiber layer (RNFL)—from the inner limiting membrane to the superficial outer boundary. A custom quantification software was used to quantify the microvasculature of the peripapillary region (MATLAB R2017a; MathWorks, Natick, MA).176 Excess noise was removed during global thresholding by selecting the avascular center of the ONH as the baseline signal. Large vessels greater than 32 µm in diameter were excluded. Grayscale images were converted to dichotomous, black-and-white maps of retinal vessels through a Hessian filtration and adaptive thresholding. White pixels in the binary image indicate red blood cells were detected. Vessel area density (VAD) was calculated from the binarized image as the percentage of all pixels that were white, which provides an estimate of blood moving through retinal capillaries and small vessels. 116 Air Pollution Exposure Outdoor air pollution exposures at participant residencies were estimated using the generalized additive model (GAM). Spatiotemporal GAM models were fit separately for NO2 and PM2.5 with Air Quality System data from the Environmental Protection Agency.177 A network of 19 NO2 monitors and 16 PM2.5 monitors located throughout Los Angeles and Orange County recorded air pollution levels for years 2014 through 2018 (Figure 25). The GAM is similar to linear regression, but with spline functions that allow nonlinear smoothing of associations between dependent and independent variables.178 Daily NO2 exposure was predicted by the model (Equation 1) where [NO2]s,t,p is the daily maximum NO2 concentration (parts per billion); α is the intercept, fs(x,y) is the thin-plate spline for geographical coordinates; ft(t) is the cubic regression spline for time (days), fp(p) is the cubic regression spline for periodic trends for the day of the week, month of the year, and year of the total 5-year period; and εs,t,p is the standard normal residual error. A similar model was built for the daily mean PM2.5 (µg/m3 at local conditions). Models were validated by 5-fold cross validation (Equation 2) Daily air pollution exposures were predicted for each AFEDS participant from their home address and the date of their clinical eye exam. Addresses were geocoded to longitudes and latitudes,179 which were transformed to x- and y-coordinates using the California Albers projection. The daily estimates of exposure were aggregated by taking the mean over time intervals of increasing length prior to the date of the exam, including the day, week, month, 3 months, 6 months, year, 1.5 years, and 2 years. This allowed assessment of air pollution exposure in the short term and over longer periods of cumulative exposure. 117 Figure 25: Map of air pollution monitors for NO2 and PM2.5 and participants from the AFEDS NO2 = nitrogen dioxide; PM2.5 = fine particulate matter with aerodynamic diameter less than 2.5 micrometers; AFEDS = African American Eye Disease Study; AFEDS participants are in blue; NO2 monitors are in dark green; PM2.5 monitors are in coral. Two PM 2.5 monitors are not shown; one in the northwest of Los Angeles County (Lebec, CA) and the other in Orange County (Mission Viejo, CA) 118 Statistical Analysis Linear regression modeling was conducted to assess the cross-sectional relationships between retinal perfusion and air pollution for both acute and chronic exposure. Sociodemographic, clinical, and air pollution variables were summarized as means and standard deviations or percentages. These variables were also summarized by quartiles of VAD and compared for those in the analytic cohort and excluded AFEDS participants by tests for equal means in a one-way layout and Pearson’s χ2 tests. Individual relationships between VAD and all variables were assessed using univariate linear regression; beta coefficients estimating the associations were reported in a forest plot for each covariate and another for each air pollution exposure. Associations of VAD and air pollution were assessed separately for NO2 and PM2.5, adjusting for covariables. Conceptual diagrams were created from a literature review of the effects of air pollution on chronic health outcomes and represented as directed acyclic graphs (DAGs). Four hierarchical, multivariable linear regression models were built from these associations, each model nested within the next. The first model was minimally adjusted for sex and age. The second model added socioeconomic measures: employment status, income, and education. Model three incorporated clinical variables with known associations with air pollution and microvasculature: body mass index (BMI), glycated hemoglobin (HbA1c), diabetes duration, systolic blood pressure (SBP), hypertension, ever smoking at least 100 cigarettes during lifetime, and health insurance status. The fourth and largest model also included ophthalmic measures of vision insurance, axial length of the eye, and OCTA signal strength. The fit of each model was reported as the adjusted R2. The effect of recent versus cumulative air pollution exposure on retinal blood flow was reported for each exposure window as estimated beta coefficients in forest plots for each model. Multiple imputation with chained equations (MICE) produces unbiased and accurate regression coefficients from data that is missing at random; mice was used to analyze 10 multiply imputed data sets.134 Locally weighted scatterplot 119 smoothing (LOWESS) plots with 95% confidence limits were created from predicted VAD (Model 4) to illustrate the potential nonlinear effects of air pollution on retinal blood flow adjusted for sociodemographic, clinical, and ophthalmic covariables. All analysis was conducted using R version 3.6.1.135 All statistical tests were performed using a type 1 error rate of 0.05. 120 Results Retinal OCTA measurements were obtained for 2,127 of the AFEDS cohort (n = 6,347). The analytic cohort was composed of 1,009 participants meeting the exclusion criteria of high-quality images and no glaucomatous or diabetic retinopathy disease (Figure 26). Participants excluded compared to those in the analysis tended to be similar except for marginally worse SES and health outcomes (Table 17). There were no statistically significant differences for sex, education, income, BMI, HbA1c, SBP, or hypertension diagnosis. Those excluded (n = 5,338) were on average 3.8 years older, 4.8% less likely to be working, 0.8 years longer diabetes, 0.8% more likely to ever have smoked cigarettes, 0.6% less likely to have health insurance, 2.0% less likely to have vision insurance, and 0.15 mm shorter mean axial length for both eyes (P < 0.01). However, the standardize mean differences (SMD) were considered negligible (SMD < 0.20) for all comparisons,113 except for a small difference in age (SMD = 0.35). 121 Figure 26: Flow Diagram for OCTA imaging in AFEDS Eligible n n = 7,957 Excluded Participants • Not completing home questionnaire 842 • Not completing comprehensive eye examination 1,610 • Not completing clinic questionnaire 1,927 AFEDS Cohort n = 6,347 Excluded Participants • Not completing OCTA exam 4,220 Complete OCTA Exam n = 2,127 Excluded Eyes • Signal Strength < 7 294 • Poor Image Quality 2285 OCTA High Quality n = 1,101 Excluded Eyes • Glaucoma 85 • Diabetic Retinopathy • Severe, Non-Proliferative DR 3 • Proliferative DR 9 Participants with Healthy Eyes n = 1033 Excluded Eyes • Signal Strength < 8 49 Analytic Cohort n = 1,009 AFEDS = African American Eye Disease Study; OCTA = Optical Coherence Tomography Angiography, DR = Diabetic Retinopathy 122 Table 17: Comparing Sociodemographics of Included and Excluded Adults ≥ 40 Years in the AFEDS Analytic Cohort (n = 1,009) Excluded (n = 5,338) P Value* Mean (SD) % Mean (SD) % Demographics Age (Years) 58.3 (9.83) 62.1 (11.5) < 0.001 Sex (Female) 64.2 62.8 0.403 Socioeconomic Status Highest Grade Completed 14.3 (2.42) 14.3 (2.47) 0.868 Working Status (No) 54.8 50.0 0.010 Household Income < $40,000 38.9 40.8 0.337 Clinical Outcomes Body Mass Index (kg/m^2) 30.2 (7.02) 30.4 (6.76) 0.416 Glycated Hemoglobin (%) 5.96 (1.14) 5.97 (1.23) 0.828 Duration of Diabetes (Years) 1.60 (5.35) 2.35 (7.24) 0.002 Systolic Blood Pressure (mmHg) 130 (20.4) 130 (19.9) 0.488 Hypertension Diagnosis (Yes) 59.0 62.1 0.062 Smoked More than 100 Cigarettes in Entire Life? (Yes) 41.9 42.7 0.009 Health Insurance Status (No) 7.9 8.5 0.012 Ophthalmic Outcomes Vision Insurance Status (No) 32.0 30.0 0.008 Average Axial Length (mm) 23.6 (0.967) 23.7 (1.16) < 0.001 Participants in the analytic cohort (n = 1,009) had missing data for education 39 (3.9%), working status 39 (3.9%), income 274 (27.2%), BMI 8 (0.8%), glycated hemoglobin 67 (6.6%), diabetes duration 49 (4.9%), SBP 3 (0.3%), smoking history 39 (3.9%), health insurance 35 (3.5%), vision insurance 72 (7.1%) and axial length 29 (2.9%). Comparisons were made for continuous variables using the T-test and for categorical variables the Chi-squared test 123 Sociodemographic and clinical descriptive statistics for the analytic cohort are presented in Table 18. Participants had a mean age of 58.3 years (standard deviation [SD] 9.83) and were 64.2% female. The average highest grade of education achieved was 14.3 years (SD 2.42) or some college, 50.0% were working, and 39.9% had an annual household income less than $40,000. The average BMI was at the lower limit of obese at 30.2 (SD 7.02). The average HbA1c was in the prediabetic range at 5.96 (SD 1.14). The average duration of diabetes was 1.60 (SD 5.35) years but was 11.16 years (SD 9.65) among those who had ever been diagnosed with diabetes mellitus (n = 138; 13.7%). The mean SBP was 130 mmHg (SD 20.4), and 59.0% of participants had been diagnosed with hypertension. 7.9% were ever-smokers, and 32% had health insurance. For ophthalmic covariates, 32.0% had vision coverage, and the mean axial length of the eye was 9.38 mm (SD 0.722). Participants with missing covariate data included 39 (3.9%) for education, 39 (3.9%) for working status, 274 (27.2%) for income, 8 (0.8%) for BMI, 67 (6.6%) glycated hemoglobin, 49 (4.9%) for diabetes duration, 3 (0.3%) for SBP, 39 (3.9%) for smoking history, 35 (3.5%) for health insurance, 72 (7.1%) for vision insurance, and 29 (2.9%) for axial length. 124 Table 18: Sociodemographic and Clinical Variables of 1,009 Adults ≥ 40 Years in the AFEDS Mean (SD) Range % Retinal Blood Flow Vessel Area Density (%) 34.8 (4.31) 17.9 – 45.1 Quantiles of Vessel Area Density 2.50 (1.12) 1 – 4 Demographics Age (Years) 58.3 (9.83) 40.1 – 91.5 Sex (Female) 64.2 Socioeconomic Status Highest Grade Completed 14.3 (2.42) 2 – 20 Working Status (No) 50.0 Household Income (< $40,000) 38.9 Clinical Outcomes Body Mass Index (kg/m2) 30.2 (7.02) 16.7 – 65.8 Glycated Hemoglobin (%) 5.96 (1.14) 3.9 – 14.0 Duration of Diabetes (Years) 1.60 (5.35) 0 – 44 Systolic Blood Pressure (mmHg) 130 (20.4) 81 – 203 Hypertension Diagnosis (Yes) 59.0 Smoked More than 100 Cigarettes in Entire Life? (Yes) 41.9 Health Insurance Status (No) 7.9 Ophthalmic Outcomes Vision Insurance Status (No) 32.0 Axial Length (mm) 9.38 (0.722) 8 – 10 Signal Strength (8 – 10) 23.6 (0.962) 20.8 – 26.7 Air Pollution: Average NO2 (10 ppb) Day 3.07 (0.972) 0.818 – 5.55 Week 3.07 (0.988) 0.915 – 5.70 Month 3.06 (1.10) 0.391 – 5.99 3 Months 2.99 (1.37) -0.487 – 5.69 6 Months 2.82 (1.38) -1.28 – 5.09 Year 2.35 (0.952) -0.382 – 4.23 1.5 Years 2.13 (0.976) -1.03 – 3.88 2 Years 1.91 (1.06) -1.01 – 3.87 Air Pollution: Average PM2.5 (10 µg/m3) Day 1.41 (0.351) 0.666 – 3.43 Week 1.43 (0.359) 0.673 – 3.49 Month 1.50 (0.436) 0.625 – 3.57 3 Months 1.68 (0.662) 0.052 – 3.59 6 Months 1.69 (0.672) -0.040 – 2.93 Year 1.69 (0.825) 0.165 – 3.07 1.5 Years 1.80 (0.966) 0.040 – 3.29 2 Years 1.92 (0.912) 0.281 – 3.49 Participants in the analytic cohort (n = 1,009) had missing data for education 39 (3.9%), working status 39 (3.9%), income 274 (27.2%), BMI 8 (0.8%), glycated hemoglobin 67 (6.6%), diabetes duration 49 (4.9%), SBP 3 (0.3%), smoking history 39 (3.9%), health insurance 35 (3.5%), vision insurance 72 (7.1%) and axial length 29 (2.9%). 125 Vessel Area Density Vessel area density had a mean of 34.8% (range 17.9% to 45.1%) of white pixels, indicating the presence of red blood cells and perfusion in the retina microvasculature of the peripapillary area. Univariate linear regression of covariables (Figure 27A) demonstrated significant associations between greater VAD and increasing signal strength and female sex as well as lower VAD with being unemployed, greater axial length, having hypertension, higher HbA1c, older age, longer duration of diabetes, and higher SBP. There were no significant univariate associations with health or vision insurance, education, BMI, income, or smoking status. Figure 28 illustrates LOWESS and boxplots of continuous and categorical variables to assess for linearity of univariate associations with VAD. 126 Figure 27: Univariate linear regression for the association of vessel area density and predictors A B Beta estimates and 95% confidence intervals of univariate linear regression for the association of vessel area density on (A) sociodemographic, clinical, and ophthalmic covariates; (B) air pollution mean exposure (10 ppb NO2, or 10 µg/m3 for PM2.5) over time windows of increasing duration leading up to the date of the eye exam. Figure 28: LOWESS Plots of Covariates with Significant Univariate Associations with Vessel Area Density 127 128 Air Pollution Exposure Daily air pollution levels at each monitor location were averaged over the entire 5-year period from 2014 through 2018 (Figure 29). Mean daily NO2 levels ranged from 22 ppb in the more rural areas of Lancaster and Santa Clarita to 38 ppb in more urban areas of Downtown Los Angeles and Long Beach. Mean daily PM2.5 followed a similar pattern ranging from 4 to 18 µm/m3, but there was less variation observed at each monitor site. Annual trends in air pollution show higher daily maximum NO2 in the winter months as temperatures cool; the annual trend for daily mean PM2.5 is much more subtle (Figure 30). Predicted air pollution exposures are illustrated as 3-dimensional GAMs over Los Angeles County (Figure 31). Predicted NO2 has local maximums in Downtown Los Angeles and Long Beach and reaches a global maximum inland towards Riverside County. Predicted PM2.5 is smoother and has a maximum over LAX and the coast before decreasing inland to the east. Observed and predicted daily air pollution are shown for 2014 to 2018 in Figure 32. The R2 for 5-fold cross validation was 0.577 for NO2 and 0.391 for PM2.5. 129 Figure 29: Daily NO2 and PM2.5 averaged from 2014–2018 in Los Angeles County A B Figure 30: Temporal Trends in daily maximum NO2 and daily mean PM2.5 for Downtown Los Angeles Monitor A B C D 130 131 Figure 31: Generalized Additive Model of NO2 and PM2.5 from 2014–2019 in Los Angeles County A R2 = 0.577 B R2 = 0.391 Figure 32: Observed and Predicted Air Pollution Exposures from 2014–2018 at Monitors in Los Angeles County A B C D Red lines represent generalized additive models for time with the same parameters for smoothing as in the predictive models 132 133 Association of VAD and Air Pollution Hierarchical linear regression modeling was used to assess the association between retinal blood flow and air pollution. Nested models of increasing complexity were created using multivariable linear regression and were based on published direct and indirect relationships between covariates and either the outcome or the exposure variables. DAGs are shown for each conceptual model (Figure 33). Although the conceptual models depict the relationship between air pollution and retinal blood flow with potential confounders, mediation, and indirect and direct effects, all analyses were performed using multivariable linear regression only. No interaction terms, variable transformations, or path analyses were performed. Figure 33: Conceptual models of the relationship of RBF and TRAP for hierarchical linear models A B C D 134 135 Model 0: Univariate Linear Regression The univariate associations of VAD with NO2 and PM2.5 demonstrate opposing trends for acute and chronic air pollution exposures (Figure 27B). For every 10 ppb greater NO2 exposure the day of the eye exam there was -0.442 (95% CI: -0.714, -0.170) lower percentage of VAD (P < 0.01). This association diminished in magnitude as the time window was increased to the previous week (-0.436 [95% CI: -0.704, -0.168]), month (-0.373 [95% CI: -0.613, -0.133]), 3 months (-0.304 [95% CI: -0.497, -0.111]), and 6 months (-0.239 [95% CI: -0.432, -0.047]). The univariate, inverse association of VAD with NO2 was not statistically significant for chronic NO2 exposures the year, 1.5 years, and 2 years prior to the eye exam. In contrast, univariate linear regression of VAD with PM2.5 exposures were not statistically significant for acute exposures the day, week, or month leading up to measuring the eye. However, for every 10 µg/m3 greater PM2.5 during the 3 months before the exam there was an association of -0.599 (95% CI: -1.000, -0.199) lower percentage of VAD (P < 0.01). Statistically significant inverse associations between VAD and PM2.5 were of similar magnitude for the 6 months (-0.638 [-1.032, -0.245]), year (-0.424 [-0.745, -0.102]), 1.5 years (-0.469 [-0.743, -0.195]), and 2 years (-0.514 [-0.804, - 0.224)] leading up to the exam. 136 Model 1: Linear Regression Minimally Adjusted The first linear model was minimally adjusted for age and sex to account for demographic differences among study participants (Figure 33A). Age is a potential confounder, as it may influence where participants live in the study area, thus affecting air pollution exposure. The univariate analysis demonstrated greater age was statistically significantly associated with reduced VAD in the radial peripapillary area, which supplies the retinal nerve fiber layer and is known to thin with age.180 Furthermore, sex may influence air pollution exposure, as one gender may be more likely to be indoors than the other and thus impact their exposure to outdoor air pollution. Males are known to have lower VAD,181 which was observed in the univariate analysis. Beta coefficients of the association of VAD on air pollution in Model 1 were nearer to the null compared to the univariate associations, but all relationships were in the same directions, did not change in statistical significance, and showed similar negative associations with VAD over lengthening time windows for both recent NO2 and prolonged PM2.5 pollution exposure (Figure 34A). Figure 34: Multivariable hierarchical linear regression for VAD on TRAP of increasing exposure duration A B C D Beta estimates and 95% confidence intervals of multivariable linear regression for the association of vessel area density on air pollution mean exposure (10 ppb NO2, or 10 µg/m3 for PM2.5) over time windows of increasing duration leading up to the date of the eye exam. 137 138 Model 2: Linear Regression Adjusted for Socioeconomic Status In addition to sex and age, the second model was adjusted for socioeconomic status (SES) covariables (Figure 33B). Working status, annual household income (yes/no < $40,000), and education (highest grade completed) compose the latent factor SES, and have been shown to explain racial/ethnic disparities in many diseases,139 including cataract lens opacities.140 SES is a possible confounder of the association of interest because poverty and lower education have been associated with living in areas exposed to greater environmental air pollution,172 and is related to improved health outcomes including glaucoma.182,183 Beta coefficients of the association of VAD on air pollution in Model 2 were nearly identical to those in Model 1 (Figure 34B). Model 3: Linear Regression Adjusted for Clinical Outcomes The third model was adjusted for the previous sociodemographics as well as for clinical biomarkers and disease outcomes related to both air pollution and vascular disease (Figure 33C). The association of obesity and air pollution is an active area of research,184 but obesity has been shown to be an important modifier of the relationship between greater air pollution and increased lipid levels in adolescent children.185 Hypertension is well known to be related to both air pollution and vascular disease.186 Similarly diabetes mellitus is also associated with both air pollution and vascular changes in the eye.187 Smoking status may also be related to location of residence and therefore air pollution, but is likely mediated by education; smoking has also been related to larger diameter venules and narrower arterioles of the retina.188 Finally, health insurance is thought to be related to vascular disease and air pollution, but is also likely mediated through SES. Beta coefficients of the association of VAD on air pollution in Model 3 were again near identical to the first two models (Figure 34C). 139 Model 4: Linear Regression Adjusted for Ophthalmic Outcomes The fourth model was further adjusted for ophthalmic measures related to vessel area density (Figure 33D). Vision insurance was not significantly associated with VAD in the univariate analysis but was included in the model to control for access to eye care. Axial length and signal strength exhibited the strongest univariate associations with VAD. Axial length has been associated with hypertensive primary open angle glaucoma, as increased intraocular pressure may elongate the eye over time;189 however patients with glaucoma were excluded to eliminate these effects. Myopic elongation of the eye has also been demonstrated in school children with greater exposure to TRAP,190 suggesting it may be an important confounder of the primary association of interest. Finally, signal strength has been positively associated with VAD in numerous studies.181,191 None of the beta coefficients of the association of VAD on air pollution in Model 4 were significant. However, all directions of effect were similar to the previous models, and the overall trends persisted for increasing time windows of air pollution exposures (Figure 34D). Individual time windows were selected for NO2 and PM2.5 to demonstrate the associations in the fully adjusted, fourth model of VAD on air pollution. Mean NO2 exposure was selected during the week leading up to the eye exam because all models—0 through 4—revealed earlier and more acute windows of exposure to NO2 were most strongly related to VAD. Mean PM2.5 exposure was chosen over the 6 months prior to the exam because all models had greater associations for longer, chronic exposure to PM2.5 and VAD. LOWESS plots of predicted VAD from Model 4 on air pollution exposures are illustrated for NO2 exposure the week leading up to the eye exam (Figure 35A) and for PM2.5 the 6 months before the exam (Figure 35B). Both plots reveal the negative, non-statistically significant associations of VAD with air pollution. Figure 35: LOWESS plot of predicted VAD from regression of TRAP exposure in the AFEDS A B NO2 = nitrogen dioxide; PM2.5 = fine particulate matter with aerodynamic diameter less than 2.5 micrometers Linear regression models are adjusted for age, sex, education, working status, household income, body mass index, glycated hemoglobin, duration of diabetes, systolic blood pressure, hypertension diagnosis, smoking more than 100 cigarettes over lifetime, health insurance, OCTA signal strength, axial length, and vision insurance NO2 and PM2.5 exposures were predicted from spatiotemporal generalized additive models created from daily maximum NO2 and daily mean PM2.5 reported by the Environmental Protection Agency of Outdoor Air Quality Data monitors as well as participants’ addresses and dates of eye exams 140 141 Table 19 enumerates linear regression parameters from all four hierarchical multivariable models describing the association of VAD on mean NO2 exposure the week prior to the eye exam. In the fully adjusted model, the linear regression beta coefficient for the percentage of VAD was -0.160 (95% CI: -0.377, 0.056) lower for every 10-ppb increase in NO2. This association was not statistically significant; however the same relationship was significant for the previous three models that did not adjust for axial length, signal strength, or vision insurance. Age was statistically significantly and negatively associated with VAD in all four models; the association in the fourth model was -0.102 (95% CI: -0.126, -0.077). Females were statistically significantly more likely to have greater VAD in all models except there was no association in the final model: -0.018 (-0.487, 0.45). VAD was 0.153 (95% CI: 0.0488, 0.256) higher for every year of education achieved, and was statistically significant in the fully adjusted model only. Unemployment was negatively associated with VAD for all models; VAD was -0.605 (95% CI: -1.10, -0.114) lower among those unemployed in the fully adjusted model. VAD was not significantly associated with household income dichotomized at $40,000 for any model. Of note, no clinical outcomes or biomarkers were statistically significantly associated with VAD in either model. Furthermore, vision insurance was also not associated with VAD. Signal strength was statistically significantly and positively associated with increased VAD at 2.55 (95% CI: 2.24, 2.86). VAD was significantly associated with -0.823 (95% CI: -1.06, -0.585) mm shorter axial length of the eye. The R2 was 0.376 for the fully adjusted model of VAD on NO2 exposure the week before the exam. Table 19: Regression of VAD on NO2 Exposure the Week before the Exam for 1,009 Adults ≥ 40 Years in the AFEDS Model 1 R2 = 0.136 Model 2 R2 = 0.139 Model 3 R2 = 0.141 Model 4 R2 = 0.376 β 95% CI β 95% CI β 95% CI β 95% CI Air Pollution Average NO2 (10 ppb) -0.288* -0.539, -0.0372 -0.279* -0.532, -0.0266 -0.273* -0.526, -0.0197 -0.160 -0.377, 0.056 Demographics Age (Years) -0.155* -0.181, -0.13 -0.145* -0.172, -0.118 -0.144* -0.172, -0.115 -0.102* -0.126, -0.0767 Sex (Female) 0.859* 0.342, 1.38 0.842* 0.322, 1.36 0.818* 0.286, 1.35 -0.018 -0.487, 0.45 Socioeconomic Status Highest Grade Completed 0.0547 -0.0648, 0.174 0.0525 -0.068, 0.173 0.153* 0.0488, 0.256 Working Status (No) -0.577* -1.14, -0.0108 -0.588* -1.16, -0.0158 -0.605* -1.1, -0.114 Household Income (< $40,000) 0.294 -0.38, 0.968 0.216 -0.471, 0.903 0.0142 -0.586, 0.615 Clinical Outcomes Body Mass Index (kg/m2) 0.0124 -0.0258, 0.0506 0.0174 -0.0155, 0.0503 Glycated Hemoglobin (%) -0.261 -0.522, 0.00111 -0.083 -0.297, 0.131 Duration of Diabetes (Years) -0.0011 -0.0534, 0.0511 -0.0194 -0.0635, 0.0247 Systolic Blood Pressure (mmHg) -0.0097 -0.0243, 0.00487 -0.0109 -0.0234, 0.00155 Hypertension (Yes) 0.114 -0.524, 0.753 0.0795 -0.465, 0.624 Smoked More than 100 Cigarettes in Entire Life? 0.281 -0.26, 0.823 0.118 -0.349, 0.586 Health Insurance Status (No) -0.161 -1.11, 0.784 -0.368 -1.21, 0.478 Ophthalmic Outcomes Vision Insurance Status (No) 0.0794 -0.439, 0.598 Signal Strength (8 – 10) 2.55* 2.24, 2.86 Axial Length (mm) -0.823* -1.06, -0.585 Intercept 44.2* 42.6, 45.9 42.9* 40.5, 45.4 45.2* 41.7, 48.7 36.2* 28.5, 43.9 NO2 = nitrogen dioxide NO2 exposure was predicted from spatiotemporal generalized additive models created from daily maximum NO2 reported by the Environmental Protection Agency of Outdoor Air Quality Data monitors as well as participants’ addresses and dates of eye exams *Coefficients are statistically different from zero (P < 0.05) 142 143 Table 20 lists linear regression parameters from all four hierarchical multivariable models characterizing the association of VAD on mean PM2.5 exposure the 6 months prior to the eye exam. In the fully adjusted model, the linear regression beta coefficient for the percentage of VAD was -0.271 (95% CI: -0.588, 0.0453) lower for every 10 µg/m3 increase in PM2.5. Associations for all covariates were similar in the model of VAD on the 6-month exposure to PM2.5 model as they were for the 1-week exposure with NO2. Age was significantly inversely associated with VAD in all four PM2.5 models. Females had greater associated VAD than males, except in the final model where there was no relationship. Education was positively and significantly associated with VAD in the fourth model alone. Unemployment was significantly associated with lower VAD in all multivariable models. In all models, annual household income and all clinical biomarkers and outcomes were not significantly associated with VAD. In the fully adjusted model for PM2.5, similar significant associations were observed for VAD and signal strength (2.54 [95% CI: 2.23, 2.85]) and mm of axial length (-0.817 [95% CI: -1.05, -0.579]). The R2 was 0.387 for the fully adjusted model of VAD on PM2.5 exposure the 6 months before the exam. Table 20: Regression of VAD on PM2.5 Exposure 6 Months before the Exam for 1,009 Adults ≥ 40 Years in the AFEDS Model 1 R2 = 0.139 Model 2 R2 = 0.142 Model 3 R2 = 0.144 Model 4 R2 = 0.387 β 95% CI β 95% CI β 95% CI β 95% CI Air Pollution -0.539* -0.906, -0.172 -0.555* -0.922, -0.188 -0.542* -0.911, -0.172 -0.271 -0.588, 0.0453 Average PM2.5 (10 µg/m3) Demographics -0.157* -0.182, -0.131 -0.145* -0.172, -0.117 -0.142* -0.171, -0.114 -0.101* -0.126, -0.0764 Age (Years) 0.874* 0.358, 1.39 0.858* 0.339, 1.38 0.840* 0.309, 1.37 -0.001 -0.47, 0.466 Sex (Female) Socioeconomic Status 0.0523 -0.0666, 0.171 0.049 -0.0709, 0.169 0.149* 0.0458, 0.253 Highest Grade Completed -0.635* -1.2, -0.0714 -0.64* -1.21, -0.0699 -0.635* -1.13, -0.144 Working Status (No) 0.228 -0.439, 0.895 0.157 -0.523, 0.838 -0.014 -0.613, 0.585 Household Income (< $40,000) Clinical Outcomes 0.0103 -0.0278, 0.0484 0.0162 -0.0167, 0.049 Body Mass Index (kg/m2) -0.238 -0.499, 0.0229 -0.0723 -0.286, 0.142 Glycated Hemoglobin (%) -0.0009 -0.0529, 0.0511 -0.0192 -0.0633, 0.0248 Duration of Diabetes (Years) -0.0106 -0.025, 0.00394 -0.0114 -0.024, 0.00112 Systolic Blood Pressure (mmHg) 0.0562 -0.581, 0.693 0.046 -0.498, 0.59 Hypertension (Yes) 0.275 -0.266, 0.816 0.114 -0.353, 0.582 Smoked More than 100 Cigarettes in Entire Life? -0.203 -1.15, 0.739 -0.382 -1.23, 0.464 Health Insurance Status (No) Ophthalmic Outcomes 0.0587 -0.46, 0.577 Vision Insurance Status (No) 2.54* 2.23, 2.85 Signal Strength (8 – 10) -0.817* -1.05, -0.579 Axial Length (mm) 44.3* 42.7, 45.9 43.1* 40.6, 45.6 45.4* 41.9, 48.9 36.2* 28.5, 43.9 Intercept -0.539* -0.906, -0.172 -0.555* -0.922, -0.188 -0.542* -0.911, -0.172 -0.271 -0.588, 0.0453 PM2.5 = fine particulate matter with aerodynamic diameter less than 2.5 micrometers PM2.5 exposure was predicted from spatiotemporal generalized additive models created from daily mean PM2.5 reported by the Environmental Protection Agency of Outdoor Air Quality Data monitors as well as participants’ addresses and dates of eye exams *Coefficients are statistically different from zero (P < 0.05) 144 145 Discussion This population-based study assessed how blood perfusion through the RPCs of the retina was associated with TRAP exposure in African Americans living in Los Angeles County. We found a modest inverse association between recent NO2 exposure—in the previous day, week, month, and 3 months—was associated with lower VAD of the retina circumscribing the optic nerve head measured by OCTA. We also found a similar association between VAD and PM2.5 exposure, but for longer, chronic exposures—previous 3 months, 6 months, year, 1.5 years, and 2 years. After adjusting for signal strength, all associations maintained directions of effect and similar patterns across increasing exposure windows. However, none of the associations remained significant. Even so, the results from multivariable linear regression models can be interpreted by standardizing the association between VAD and air pollution on those of VAD and age. Therefore, in the fully adjusted model a 10 ppb greater exposure in NO2 over the previous week can be equated to the same difference in VAD as 1.578 (95% CI: -0.730, 2.98) years older age. And a 10 µg/m3 greater exposure to PM2.5 over the previous 6 months can be equated to 2.68 (95% CI: -0.593, 4.66) years. A recent National Health Service study of community-dwelling British participants aged 40 to 69 years old found a 1.06 (95% CI: 1.01–1.12) greater odds of glaucoma for every interquartile range increase in air pollution.192 The authors recently followed up using OCT measurements to assess how structural differences in nervous tissue of the retinal might be related to NO2 and PM2.5 exposures; they found greater pollution was associated with increased thickness of the retinal nerve fiber layer and a thinner ganglion cell-inner plexiform layer. They noted these structural differences related to TRAP did not clearly fit within the pathophysiology of glaucoma. But observed differences may have been relate to other glaucomatous pathologic differences, perhaps of the retinal vasculature. Previous studies found TRAP was related to variations in the larger vessels of the retina (60-300 µm); arteries were smaller in diameter and venules 146 somewhat wider among those with greater exposure to PM2.5.155 However, they did not investigate the RPC of the retina. This was the first study to elucidate how the smaller (< 32 µm) capillaries surrounding the optic nerve head—a key structure in glaucoma—might be associated with air pollution. The previous study was limited to large vessels because the retinal imaging was standardized digital photography. More recent advances in higher-resolution OCTA allows a more detailed assessment of the microvasculature involved in glaucomatous disease. The AFEDS is the largest population-based cohort study of eye disease for African Americans (n = 6,347). However, in this analysis our outcome was VAD measured by OCTA, which was not implement into the ophthalmology clinic until several thousand participants had completed data collection. Selection bias may exist if healthier participants with greater VAD were more likely to participate early in the study and also lived at addresses with lower air pollution; this would have likely biased estimates towards the null. Comparing sociodemographics of participants (Table 17) suggest that those with worse SES were excluded from the final analysis, providing evidence that findings may be attenuated from the truth. The follow-up study for the AFEDS should include an aim to study the effects of air pollution exposure and retinal blood flow measured by OCTA, especially because the prevalence of glaucoma and the exposure to air pollution are increased in this population. The air pollution models could also be improved to predict spatiotemporal variations in air pollution for AFEDS participants. Only 58% of the observed variation for NO2 and 39% for PM2.5 was explained by spatiotemporal GAM models. Previous models of pollution exposure have reached 71%.193 Traffic noise exposure (dB) was included as a potential source of pollution that may have been associated with the retinal microvasculature, as it has been implicated in health outcomes independent of air pollution, including children’s’ lung function.194 However, the prediction models provided estimate for 4 km tiles, which allowed limited variation in our study area which covered about 10 kilometers. Distance from major roadway could be added to the 147 GAM models to improve their predictive abilities. Increasing the estimated air pollution exposure for participants may explain more of the observed variation and might allow more precise estimates of the association between VAD and pollution. Furthermore, daily mean PM2.5 demonstrated less spatial variation over the same geographical region of Inglewood, CA compared to daily maximum NO2 (Figure 31). The reduced variation and the limited study area may have contributed to the lack of association between acute PM2.5 exposures and VAD observed in all linear regression models. Finally, signal strength is a quality control variable that depicts the reliability of OCTA images. Previous studies have demonstrated signal strength is an important confounder that should be adjusted for OCTA studies of eye disease,175,181,191 In the present study, it elicited the strongest univariate association with VAD of all other predictors. However, participants with more severe disease also have lower signal strength and VAD. Therefore, controlling for this variable may explain away some of the observed variation. The hierarchical design of the regression modeling in this study demonstrated that VAD was significantly associated with acute NO2 exposure and chronic PM2.5 exposures after adjusting for demographic, socioeconomic, and clinical covariates, albeit the proportion of the variation explained was limited (R2 < 0.20). Signal strength may be masking the relatively small association of air pollution and VAD. This may be working in concert with the incomplete sample of the population-study and the limited predictive power of the air pollution exposure to reduce the observed true effect. However, despite this we consistently observed inverse relationships between retinal perfusion with acute NO2 and chronic PM2.5 air pollution exposures. To the best of our knowledge, these results are unique in that TRAP has not been demonstrated to be related to blood perfusion of the radial peripapillary capillaries in African Americans. However, they do contribute to recent work that aims to identify the toxicologic role of pollution in structural changes of the retinal nervous, epithelial, and epithelial tissues. Health 148 care providers should be aware of the developing evidence that together SES and environmental air pollution may play in vascular diseases of the eye. Future population-based studies should aim to study a diverse patient population over and area with high spatial variation in TRAP chemical pollutants. The question remains to what extent air pollution may explain racial/ethnic disparities in the greater prevalence of open angle glaucoma among African Americans. 149 References 1. Cost of Vision Problems: The Economic Burden of Vision Loss and Eye Disorders in the United States. University of Chicago; 2013. 2. Varma R, Vajaranant TS, Burkemper B, et al. Visual Impairment and Blindness in Adults in the United States: Demographic and Geographic Variations from 2015 to 2050. JAMA Ophthalmol. 2016;134(7):802-809. doi:10.1001/jamaophthalmol.2016.1284 3. Klein R, Klein BEK. The prevalence of age-related eye diseases and visual impairment in aging: current estimates. Invest Ophthalmol Vis Sci. 2013;54(14):ORSF5-ORSF13. doi:10.1167/iovs.13-12789 4. Friedman DS, Jampel HD, Muñoz B, West SK. The prevalence of open-angle glaucoma among blacks and whites 73 years and older: the Salisbury Eye Evaluation Glaucoma Study. Arch Ophthalmol. 2006;124(11):1625-1630. doi:10.1001/archopht.124.11.1625 5. Stein JD, Kim DS, Niziol LM, et al. Differences in rates of glaucoma among Asian Americans and other racial groups, and among various Asian ethnic groups. Ophthalmology. 2011;118(6):1031-1037. doi:10.1016/j.ophtha.2010.10.024 6. Varma R, Wu J, Chong K, Azen SP, Hays RD, Los Angeles Latino Eye Study G. Impact of severity and bilaterality of visual impairment on health-related quality of life. Ophthalmology. 2006;113(10):1846-1853. doi:10.1016/j.ophtha.2006.04.028 7. McKean-Cowdin R, Varma R, Hays RD, et al. Longitudinal changes in visual acuity and health-related quality of life: the Los Angeles Latino Eye study. Ophthalmology. 2010;117(10):1900-1907, 1907 e1. doi:10.1016/j.ophtha.2010.01.059 8. West SK, Munoz B, Rubin GS, et al. Function and visual impairment in a population-based study of older adults. The SEE project. Salisbury Eye Evaluation. Invest Ophthalmol Vis Sci. 1997;38(1):72-82. 9. Broman AT, Munoz B, Rodriguez J, et al. The impact of visual impairment and eye disease on vision-related quality of life in a Mexican-American population: proyecto VER. Invest Ophthalmol Vis Sci. 2002;43(11):3393-3398. 10. Globe DR, Wu J, Azen SP, Varma R, Los Angeles Latino Eye Study G. The impact of visual impairment on self-reported visual functioning in Latinos: The Los Angeles Latino Eye Study. Ophthalmology. 2004;111(6):1141-1149. doi:10.1016/j.ophtha.2004.02.003 11. Freeman EE, Munoz B, Rubin G, West SK. Visual field loss increases the risk of falls in older adults: the Salisbury eye evaluation. Invest Ophthalmol Vis Sci. 2007;48(10):4445- 4450. doi:10.1167/iovs.07-0326 12. Patino CM, McKean-Cowdin R, Azen SP, Allison JC, Choudhury F, Varma R. Central and peripheral visual impairment and the risk of falls and falls with injury. Ophthalmology. 2010;117(2):199-206.e1. doi:10.1016/j.ophtha.2009.06.063 150 13. Ivers RQ, Cumming RG, Mitchell P, Attebo K. Visual impairment and falls in older adults: the Blue Mountains Eye Study. J Am Geriatr Soc. 1998;46(1):58-64. doi:10.1111/j.1532- 5415.1998.tb01014.x 14. Ivers RQ, Cumming RG, Mitchell P, Simpson JM, Peduto AJ. Visual risk factors for hip fracture in older people. J Am Geriatr Soc. 2003;51(3):356-363. 15. Ivers RQ, Cumming RG, Mitchell P, Peduto AJ. Risk factors for fractures of the wrist, shoulder and ankle: the Blue Mountains Eye Study. Osteoporos Int. 2002;13(6):513-518. doi:10.1007/s001980200063 16. Freeman EE, Munoz B, Turano KA, West SK. Measures of visual function and their association with driving modification in older adults. Investigative ophthalmology & visual science. 2006;47(2):514-520. doi:10.1167/iovs.05-0934 17. Segal-Gidan F, Varma R, Salazar X, Mack WJ. Factors influencing driving status in an older Latino population. J Aging Health. 2010;22(3):332-347. doi:10.1177/0898264309358763 18. Gilhotra JS, Mitchell P, Ivers R, Cumming RG. Impaired vision and other factors associated with driving cessation in the elderly: the Blue Mountains Eye Study. Clin Exp Ophthalmol. 2001;29(3):104-107. 19. Ramulu PY, West SK, Munoz B, Jampel HD, Friedman DS. Driving Cessation and Driving Limitation in Glaucoma: The Salisbury Eye Evaluation Project. Ophthalmology. 2009;116(10):1846-1853. doi:10.1016/j.ophtha.2009.03.033 20. Qiu M, Wang SY, Singh K, Lin SC. Association between visual field defects and quality of life in the United States. Ophthalmology. 2014;121(3):733-740. doi:10.1016/j.ophtha.2013.09.043 21. McKean-Cowdin R, Varma R, Wu J, Hays RD, Azen SP. Severity of visual field loss and health-related quality of life. American journal of ophthalmology. 2007;143(6):1013-1023. doi:10.1016/j.ajo.2007.02.022 22. Patino CM, Varma R, Azen SP, et al. The impact of change in visual field on health-related quality of life the los angeles latino eye study. Ophthalmology. 2011;118(7):1310-1317. doi:10.1016/j.ophtha.2010.12.018 23. McKean-Cowdin R, Wang Y, Wu J, Azen SP, Varma R. Impact of visual field loss on health- related quality of life in glaucoma: the Los Angeles Latino Eye Study. Ophthalmology. 2008;115(6):941-948.e1. doi:10.1016/j.ophtha.2007.08.037 24. Freeman EE, Munoz B, West SK, Jampel HD, Friedman DS. Glaucoma and quality of life: the Salisbury Eye Evaluation. Ophthalmology. 2008;115(2):233-238. doi:10.1016/j.ophtha.2007.04.050 25. Rubin GS, Bandeen-Roche K, Huang GH, et al. The association of multiple visual impairments with self-reported visual disability: SEE project. Invest Ophthalmol Vis Sci. 2001;42(1):64-72. 151 26. Kaleem MA, Munoz BE, Munro CA, Gower EW, West SK. Visual characteristics of elderly night drivers in the Salisbury Eye Evaluation Driving Study. Invest Ophthalmol Vis Sci. 2012;53(9):5161-5167. doi:10.1167/iovs.12-9866 27. Johnson CA, Keltner JL. Incidence of visual field loss in 20,000 eyes and its relationship to driving performance. Arch Ophthalmol. 1983;101(3):371-375. 28. van Landingham SW, Willis JR, Vitale S, Ramulu PY. Visual field loss and accelerometer- measured physical activity in the United States. Ophthalmology. 2012;119(12):2486-2492. doi:10.1016/j.ophtha.2012.06.034 29. Ramulu PY, Hochberg C, Maul EA, Chan ES, Ferrucci L, Friedman DS. Glaucomatous visual field loss associated with less travel from home. Optometry and vision science : official publication of the American Academy of Optometry. 2014;91(2):187-193. doi:10.1097/opx.0000000000000139 30. Curriero FC, Pinchoff J, van Landingham SW, Ferrucci L, Friedman DS, Ramulu PY. Alteration of travel patterns with vision loss from glaucoma and macular degeneration. JAMA Ophthalmology. 2013;131(11):1420-1426. doi:10.1001/jamaophthalmol.2013.4471 31. Ramulu PY, Mihailovic A, West SK, Friedman DS, Gitlin LN. What Is a Falls Risk Factor? Factors Associated with Falls per Time or per Step in Individuals with Glaucoma. J Am Geriatr Soc. 2019;67(1):87-92. doi:10.1111/jgs.15609 32. Ramulu PY, Mihailovic A, West SK, Gitlin LN, Friedman DS. Predictors of Falls per Step and Falls per Year At and Away From Home in Glaucoma. Am J Ophthalmol. 2019;200:169-178. doi:10.1016/j.ajo.2018.12.021 33. Coleman AL, Cummings SR, Ensrud KE, et al. Visual field loss and risk of fractures in older women. J Am Geriatr Soc. 2009;57(10):1825-1832. doi:10.1111/j.1532-5415.2009.02432.x 34. Medeiros FA, Gracitelli CP, Boer ER, Weinreb RN, Zangwill LM, Rosen PN. Longitudinal changes in quality of life and rates of progressive visual field loss in glaucoma patients. Ophthalmology. 2015;122(2):293-301. doi:10.1016/j.ophtha.2014.08.014 35. Gutierrez P, Wilson MR, Johnson C, et al. Influence of glaucomatous visual field loss on health-related quality of life. Archives of ophthalmology (Chicago, Ill : 1960). 1997;115(6):777-784. 36. Parrish RK, Gedde SJ, Scott IU, et al. Visual function and quality of life among patients with glaucoma. Arch Ophthalmol. 1997;115(11):1447-1455. 37. Ringsdorf L, McGwin G, Owsley C. Visual field defects and vision-specific health-related quality of life in African Americans and whites with glaucoma. Journal of glaucoma. 2006;15(5):414-418. doi:10.1097/01.ijg.0000212252.72207.c2 38. van Gestel A, Webers CA, Beckers HJ, et al. The relationship between visual field loss in glaucoma and health-related quality-of-life. Eye (Lond). 2010;24(12):1759-1769. doi:10.1038/eye.2010.133 152 39. Abe RY, Diniz-Filho A, Costa VP, Gracitelli CP, Baig S, Medeiros FA. The Impact of Location of Progressive Visual Field Loss on Longitudinal Changes in Quality of Life of Patients with Glaucoma. Ophthalmology. 2016;123(3):552-557. doi:10.1016/j.ophtha.2015.10.046 40. Spaeth G, Walt J, Keener J. Evaluation of quality of life for patients with glaucoma. Am J Ophthalmol. 2006;141(1 Suppl):S3-14. doi:10.1016/j.ajo.2005.07.075 41. Sawada H, Fukuchi T, Abe H. Evaluation of the relationship between quality of vision and visual function in Japanese glaucoma patients. Clinical ophthalmology (Auckland, NZ). 2011;5:259-267. doi:10.2147/opth.S16989 42. Pesudovs K, Gothwal VK, Wright T, Lamoureux EL. Remediating serious flaws in the National Eye Institute Visual Function Questionnaire. Journal of cataract and refractive surgery. 2010;36(5):718-732. doi:10.1016/j.jcrs.2009.11.019 43. Valbuena M, Bandeen-Roche K, Rubin GS, Munoz B, West SK. Self-reported assessment of visual function in a population-based study: the SEE project. Salisbury Eye Evaluation. Invest Ophthalmol Vis Sci. 1999;40(2):280-288. 44. Broman AT, Munoz B, West SK, et al. Psychometric Properties of the 25-Item NEI-VFQ in a Hispanic Population: Proyecto VER. Invest Ophthalmol Vis Sci. 2001;42(3):606-613. 45. Margolis MK, Coyne K, Kennedy-Martin T, Baker T, Schein O, Revicki DA. Vision-specific instruments for the assessment of health-related quality of life and visual functioning: a literature review. PharmacoEconomics. 2002;20(12):791-812. doi:10.2165/00019053- 200220120-00001 46. Kuo YS, Liu CJL, Cheng HC, Chen MJ, Chen WT, Ko YC. Impact of socioeconomic status on vision-related quality of life in primary open-angle glaucoma. In: Eye (Lond). Vol 31. ; 2017:1480-1487. 47. Paz SH, Globe DR, Wu J, Azen SP, Varma R. Relationship between self-reported depression and self-reported visual function in Latinos. Arch Ophthalmol. 2003;121(7):1021- 1027. doi:10.1001/archopht.121.7.1021 48. Warrian KJ, Spaeth GL, Lankaranian D, Lopes JF, Steinmann WC. The effect of personality on measures of quality of life related to vision in glaucoma patients. Br J Ophthalmol. 2009;93(3):310-315. doi:10.1136/bjo.2008.139592 49. Skenduli-Bala E, de Voogd S, Wolfs RC, et al. Causes of incident visual field loss in a general elderly population: the Rotterdam study. Arch Ophthalmol. 2005;123(2):233-238. doi:10.1001/archopht.123.2.233 50. Wang YX, Xu L, Sun XY, Zou Y, Zhang HT, Jonas JB. Five year incidence of visual field loss in adult Chinese. The Beijing Eye Study. Plos One. 2012;7(5):e37232. doi:10.1371/journal.pone.0037232 51. McKean-Cowdin R, Fairbrother-Crisp A, Torres M, et al. The African American Eye Disease Study: Design and Methods. Ophthalmic epidemiology. 2018;25(4):306-314. doi:10.1080/09286586.2018.1454965 153 52. Azur MJ, Stuart EA, Frangakis C, Leaf PJ. Multiple Imputation by Chained Equations: What is it and how does it work? Int J Methods Psychiatr Res. 2011;20(1):40-49. doi:10.1002/mpr.329 53. Yohannan J, Wang J, Brown J, et al. Evidence-based Criteria for Assessment of Visual Field Reliability. Ophthalmology. 2017;124(11):1612-1620. doi:10.1016/j.ophtha.2017.04.035 54. Petrillo J, Bressler NM, Lamoureux E, Ferreira A, Cano S. Development of a new Rasch- based scoring algorithm for the National Eye Institute Visual Functioning Questionnaire to improve its interpretability. Health Qual Life Outcomes. 2017;15(1):157. doi:10.1186/s12955-017-0726-5 55. Netuveli G, Blane D. Quality of life in older ages. Br Med Bull. 2008;85(1):113-126. doi:10.1093/bmb/ldn003 56. Varma R, Hsu C, Wang D, Torres M, Azen SP. The Chinese American Eye Study: Design and Methods. Ophthalmic epidemiology. 2013;20(6):335-347. doi:10.3109/09286586.2013.823505 57. Varma R, Paz SH, Azen SP, et al. The Los Angeles Latino Eye Study: design, methods, and baseline data. Ophthalmology. 2004;111(6):1121-1131. doi:10.1016/j.ophtha.2004.02.001 58. Brody BL, Gamst AC, Williams RA, et al. Depression, visual acuity, comorbidity, and disability associated with age-related macular degeneration. Ophthalmology. 2001;108(10):1893-1900; discussion 1900-1901. doi:10.1016/s0161-6420(01)00754-0 59. Linn BS, Linn MW, Gurel L. Cumulative illness rating scale. J Am Geriatr Soc. 1968;16(5):622-626. doi:10.1111/j.1532-5415.1968.tb02103.x 60. Globe DR, Varma R, Torres M, et al. Self-reported comorbidities and visual function in a population-based study: the Los Angeles Latino Eye Study. Arch Ophthalmol. 2005;123(6):815-821. doi:10.1001/archopht.123.6.815 61. Lee S, Chae DH, Jung MY, Chen L, Juon HS. Health Examination Is Not a Priority for Less Acculturated Asian Americans. J Racial Ethn Health Disparities. Published online October 31, 2016. doi:10.1007/s40615-016-0306-0 62. Suinn RM, Khoo G, Ahuna C. The Suinn ‐Lew Asian Self ‐Identity Acculturation Scale: Cross ‐ Cultural Information - Suinn - 1995 - Journal of Multicultural Counseling and Development - Wiley Online Library. Journal of Multicultural Counseling and Development. 1995;23:139– 148. doi:10.1002/j.2161-1912.1995.tb00269.x 63. Cuellar I, Harris LC, Jasso R. An acculturation scale for Mexican American normal and clinical populations. Hispanic Journal of Behavioral Sciences. 1980;2(3):199-217. 64. Daga FB, Gracitelli CPB, Diniz-Filho A, Medeiros FA. Is vision-related quality of life impaired in patients with preperimetric glaucoma? Br J Ophthalmol. 2019;103(7):955-959. doi:10.1136/bjophthalmol-2018-312357 154 65. Siantar RG, Cheng CY, Gemmy Cheung CM, et al. Impact of Visual Impairment and Eye diseases on Mortality: the Singapore Malay Eye Study (SiMES). Scientific reports. 2015;5:16304. doi:10.1038/srep16304 66. Kuper H, Mathenge W, Macleod D, et al. Mortality during 6 years of follow-up in relation to visual impairment and eye disease: results from a population-based cohort study of people aged 50 years and above in Nakuru, Kenya. BMJ Open. 2019;9(6):e029700. doi:10.1136/bmjopen-2019-029700 67. Akbari M, Akbari S, Pasquale LR. The association of primary open-angle glaucoma with mortality: a meta-analysis of observational studies. Arch Ophthalmol. 2009;127(2):204-210. doi:10.1001/archophthalmol.2008.571 68. Lee MJ, Varadaraj V, Ramulu PY, Whitson HE, Deal JA, Swenor BK. Memory and Confusion Complaints in Visually Impaired Older Adults: An Understudied Aspect of Well- Being. Gerontol Geriatr Med. 2019;5:2333721418818944. doi:10.1177/2333721418818944 69. Ferris FL, Bailey I. Standardizing the measurement of visual acuity for clinical research studies: Guidelines from the Eye Care Technology Forum. Ophthalmology. 1996;103(1):181-182. 70. Ferris FL 3rd, Kassoff A, Bresnick GH, Bailey I. New visual acuity charts for clinical research. American journal of ophthalmology. 1982;94(1):91-96. 71. Johnson CA. Standardizing the measurement of visual fields for clinical research: Guidelines from the Eye Care Technology Forum. Ophthalmology. 1996;103(1):186-189. doi:10.1016/s0161-6420(96)30740-9 72. Ernest PJ, Schouten JS, Beckers HJ, Hendrikse F, Prins MH, Webers CA. The evidence base to select a method for assessing glaucomatous visual field progression. Acta ophthalmologica. 2012;90(2):101-108. doi:10.1111/j.1755-3768.2011.02206.x 73. Heijl A. The Field Analyzer Primer: Effective Perimetry. Carl Zeiss Meditec Incorporated; 2012. https://books.google.com/books/about/The_Field_Analyzer_Primer.html?id=Yf_yoAEACAAJ 74. Cello KE, Nelson-Quigg JM, Johnson CA. Frequency doubling technology perimetry for detection of glaucomatous visual field loss. Am J Ophthalmol. 2000;129(3):314-322. 75. Chun YS, Lee DI, Kwon J, Park IK. Comparison of Impact of Monocular and Integrated Binocular Visual Fields on Vision-related Quality of Life. Journal of glaucoma. 2017;26(3):283-291. doi:10.1097/ijg.0000000000000623 76. Arora KS, Boland MV, Friedman DS, Jefferys JL, West SK, Ramulu PY. The Relationship between Better-Eye and Integrated Visual Field Mean Deviation and Visual Disability. Ophthalmology. 2013;120(12). doi:10.1016/j.ophtha.2013.07.020 77. Heijl A, Krakau CE. An automatic static perimeter, design and pilot study. Acta Ophthalmol (Copenh). 1975;53(3):293-310. doi:10.1111/j.1755-3768.1975.tb01161.x 155 78. Springelkamp H, Wolfs RC, Ramdas WD, et al. Incidence of glaucomatous visual field loss after two decades of follow-up: the Rotterdam Study. European journal of epidemiology. 2017;32(8):691-699. doi:10.1007/s10654-017-0270-y 79. Jammal AA, Ogata NG, Daga FB, Abe RY, Costa VP, Medeiros FA. What Is the Amount of Visual Field Loss Associated With Disability in Glaucoma? Am J Ophthalmol. 2019;197:45- 52. doi:10.1016/j.ajo.2018.09.002 80. Czudowska MA, Ramdas WD, Wolfs RC, et al. Incidence of glaucomatous visual field loss: a ten-year follow-up from the Rotterdam Study. Ophthalmology. 2010;117(9):1705-1712. doi:10.1016/j.ophtha.2010.01.034 81. Tielsch JM, Sommer A, Katz J, Royall RM, Quigley HA, Javitt J. Racial variations in the prevalence of primary open-angle glaucoma. The Baltimore Eye Survey. Jama. 1991;266(3):369-374. 82. Foster PJ, Buhrmann R, Quigley HA, Johnson GJ. The definition and classification of glaucoma in prevalence surveys. In: Br J Ophthalmol. Vol 86. ; 2002:238-242. 83. Jung KI, Park CK. Mental Health Status and Quality of Life in Undiagnosed Glaucoma Patients: A Nationwide Population-Based Study. Medicine. 2016;95(19):e3523. doi:10.1097/md.0000000000003523 84. Wu S-Y, Hennis A, Nemesure B, Leske MC, Barbados Eye Studies Group. Impact of glaucoma, lens opacities, and cataract surgery on visual functioning and related quality of life: the Barbados Eye Studies. Invest Ophthalmol Vis Sci. 2008;49(4):1333-1338. doi:10.1167/iovs.07-1252 85. Cheng HC, Guo CY, Chen YJ, et al. Patient-Reported Vision-Related Quality-of-Life Differences between Primary Angle-Closure Glaucoma and Primary Open-Angle Glaucoma. Plos One. 2016;11(9):e0163123. doi:10.1371/journal.pone.0163123 86. Porta M; G. A Dictionary of Epidemiology. 6ed ed. Oxford University Press; 2014. 87. Ware J, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity. Med Care. 1996;34(3):220-233. 88. Loriaut P, Boyer P, Massin P, Cochereau I. Visual impairment and hip fractures: a case- control study in elderly patients. Ophthalmic research. 2014;52(4):212-216. doi:10.1159/000362881 89. Mangione CM, Lee PP, Pitts J, Gutierrez P, Berry S, Hays RD. Psychometric properties of the National Eye Institute Visual Function Questionnaire (NEI-VFQ). NEI-VFQ Field Test Investigators. Archives of ophthalmology (Chicago, Ill : 1960). 1998;116(11):1496-1504. 90. Mangione CM, Lee PP, Gutierrez PR, Spritzer K, Berry S, Hays RD. Development of the 25- item National Eye Institute Visual Function Questionnaire. Archives of ophthalmology (Chicago, Ill : 1960). 2001;119(7):1050-1058. 91. Petrillo J, Cano SJ, McLeod LD, Coon CD. Using classical test theory, item response theory, and Rasch measurement theory to evaluate patient-reported outcome measures: a 156 comparison of worked examples. Value in health : the journal of the International Society for Pharmacoeconomics and Outcomes Research. 2015;18(1):25-34. doi:10.1016/j.jval.2014.10.005 92. Skalicky SE, Lamoureux EL, Crabb DP, Ramulu PY. Patient-reported Outcomes, Functional Assessment, and Utility Values in Glaucoma. J Glaucoma. 2019;28(2):89-96. doi:10.1097/IJG.0000000000001155 93. Elliott DB, Pesudovs K, Mallinson T. Vision-related quality of life. Optom Vis Sci. 2007;84(8):656-658. doi:10.1097/OPX.0b013e31814db01e 94. Javitt JC, Steinberg EP, Sharkey P, et al. Cataract surgery in one eye or both. A billion dollar per year issue. Ophthalmology. 1995;102(11):1583-1592; discussion 1592-1593. doi:10.1016/s0161-6420(95)30824-x 95. Laidlaw DA, Harrad RA, Hopper CD, et al. Randomised trial of effectiveness of second eye cataract surgery. Lancet. 1998;352(9132):925-929. doi:10.1016/s0140-6736(97)12536-3 96. Elliott DB, Patla AE, Furniss M, Adkin A. Improvements in clinical and functional vision and quality of life after second eye cataract surgery. Optom Vis Sci. 2000;77(1):13-24. doi:10.1097/00006324-200001000-00009 97. Busbee BG, Brown MM, Brown GC, Sharma S. Cost-utility analysis of cataract surgery in the second eye. Ophthalmology. 2003;110(12):2310-2317. doi:10.1016/S0161- 6420(03)00796-6 98. Azuara-Blanco A, Burr J, Ramsay C, et al. Effectiveness of early lens extraction for the treatment of primary angle-closure glaucoma (EAGLE): a randomised controlled trial. Lancet. 2016;388(10052):1389-1397. doi:10.1016/S0140-6736(16)30956-4 99. EuroQol Group. EuroQol--a new facility for the measurement of health-related quality of life. Health Policy. 1990;16(3):199-208. doi:10.1016/0168-8510(90)90421-9 100. Dolan P, Gudex C, Kind P, Williams A. A social tariff for EuroQol: results from a UK general population survey. In: ; 1995. 101. Burr JM, Kilonzo M, Vale L, Ryan M. Developing a preference-based Glaucoma Utility Index using a discrete choice experiment. Optometry and vision science : official publication of the American Academy of Optometry. 2007;84(8):797-808. doi:10.1097/OPX.0b013e3181339f30 102. Baker RS, Bazargan M, Calderón JL, Hays RD. Psychometric performance of the National Eye Institute visual function questionnaire in Latinos and non-Latinos. Ophthalmology. 2006;113(8):1363-1371. doi:10.1016/j.ophtha.2006.01.073 103. Clemons TE, Chew EY, Bressler SB, McBee W. National Eye Institute Visual Function Questionnaire in the Age-Related Eye Disease Study (AREDS): AREDS Report No. 10. Arch Ophthalmol. 2003;121(2):211-217. 157 104. Owen CG, Rudnicka AR, Smeeth L, Evans JR, Wormald RP, Fletcher AE. Is the NEI- VFQ-25 a useful tool in identifying visual impairment in an elderly population? BMC ophthalmology. 2006;6:24. doi:10.1186/1471-2415-6-24 105. Hays RD, Morales LS, Reise SP. Item Response Theory and Health Outcomes Measurement in the 21st Century. Med Care. 2000;38(9 Suppl):II28-42. 106. Andrich D. Rating scales and Rasch measurement. http://dx.doi.org/101586/erp1159. Published online January 9, 2014. doi:10.1586/erp.11.59 107. Streiner DL, Norman GR, Cairney J. Health Measurement Scales: A Practical Guide to Their Development and Use. 5th ed. Oxford University Press; 2015. 108. Marella M, Pesudovs K, Keeffe JE, O’Connor PM, Rees G, Lamoureux EL. The psychometric validity of the NEI VFQ-25 for use in a low-vision population. Investigative Ophthalmology & Visual Science. 2010;51(6):2878-2884. doi:10.1167/iovs.09-4494 109. Jiang S, Wang C, Weiss DJ. Sample Size Requirements for Estimation of Item Parameters in the Multidimensional Graded Response Model. Front Psychol. 2016;7:109. doi:10.3389/fpsyg.2016.00109 110. Edelen MO, Reeve BB. Applying item response theory (IRT) modeling to questionnaire development, evaluation, and refinement. Quality of life research : an international journal of quality of life aspects of treatment, care and rehabilitation. 2007;16 Suppl 1:5-18. doi:10.1007/s11136-007-9198-0 111. Cole K. Rasch Model Calibrations with SAS PROC IRT and WINSTEPS. J Appl Meas. 2019;20(1):27-45. 112. Lisboa R, Chun YS, Zangwill LM, et al. Association between rates of binocular visual field loss and vision-related quality of life in patients with glaucoma. JAMA Ophthalmol. 2013;131(4):486-494. doi:10.1001/jamaophthalmol.2013.2602 113. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed.; 1988. https://books.google.com/books/about/Statistical_Power_Analysis_for_the_Behav.html?id=c IJH0lR33bgC 114. Rao P, Munoz B, Turano K, Munro C, West SK. The decline in attentional visual fields over time among older participants in the Salisbury Eye Evaluation Driving Study. Invest Ophthalmol Vis Sci. 2013;54(3):1839-1844. doi:10.1167/iovs.11-8874 115. Schubert CR, Cruickshanks KJ, Fischer ME, et al. Sensory Impairments and Cognitive Function in Middle-Aged Adults. J Gerontol A Biol Sci Med Sci. 2017;72(8):1087-1090. doi:10.1093/gerona/glx067 116. Fisher DE, Shrager S, Shea SJ, et al. Visual Impairment in White, Chinese, Black, and Hispanic Participants from the Multi-Ethnic Study of Atherosclerosis Cohort. Ophthalmic Epidemiol. 2015;22(5):321-332. doi:10.3109/09286586.2015.1066395 158 117. Varma R, Torres M, Pena F, Klein R, Azen SP, Los Angeles Latino Eye Study G. Prevalence of diabetic retinopathy in adult Latinos: the Los Angeles Latino eye study. Ophthalmology. 2004;111(7):1298-1306. doi:10.1016/j.ophtha.2004.03.002 118. Gupta P, Zhao D, Guallar E, Ko F, Boland MV, Friedman DS. Prevalence of Glaucoma in the United States: The 2005-2008 National Health and Nutrition Examination Survey. Invest Ophthalmol Vis Sci. 2016;57(6):2905–2913. doi:10.1167/iovs.15-18469 119. Choudhury F, Meuer SM, Klein R, et al. Prevalence and Characteristics of Myopic Degeneration in an Adult Chinese American Population: The Chinese American Eye Study. American journal of ophthalmology. 2018;187:34-42. doi:10.1016/j.ajo.2017.12.010 120. Fayers P, Machin D. Quality of Life: The Assessment, Analysis and Reporting of Patient- Reported Outcomes. Third. John Wiley & Sons, Ltd; 2016. https://www.wiley.com/en- us/Quality+of+Life%3A+The+Assessment%2C+Analysis+and+Reporting+of+Patient+report ed+Outcomes%2C+3rd+Edition-p-9781444337952 121. Leyland M, Zinicola E. Multifocal versus monofocal intraocular lenses in cataract surgery: a systematic review. Ophthalmology. 2003;110(9):1789-1798. doi:10.1016/S0161- 6420(03)00722-X 122. van Nispen RM, Virgili G, Hoeben M, et al. Low vision rehabilitation for better quality of life in visually impaired adults. Cochrane Database Syst Rev. 2020;1:CD006543. doi:10.1002/14651858.CD006543.pub2 123. Doward LC, Gnanasakthy A, Baker MG. Patient reported outcomes: looking beyond the label claim. Health Qual Life Outcomes. 2010;8:89. doi:10.1186/1477-7525-8-89 124. Eshera N, Itana H, Zhang L, Soon G, Fadiran EO. Demographics of clinical trials participants in pivotal clinical trials for new molecular entity drugs and biologics approved by FDA From 2010 to 2012. Am J Ther. 2015;22(6):435-455. doi:10.1097/mjt.0000000000000177 125. Berlin NL, Momoh AO, Qi J, et al. Racial and ethnic variations in one-year clinical and patient-reported outcomes following breast reconstruction. Am J Surg. 2017;214(2):312- 317. doi:10.1016/j.amjsurg.2017.02.009 126. Tyson MD, Alvarez J, Koyama T, et al. Racial Variation in Patient-Reported Outcomes Following Treatment for Localized Prostate Cancer: Results from the CEASAR Study. Eur Urol. 2017;72(2):307-314. doi:10.1016/j.eururo.2016.10.036 127. Milani CJ, Rundell SD, Jarvik JG, et al. Associations of Race and Ethnicity With Patient- Reported Outcomes and Health Care Utilization Among Older Adults Initiating a New Episode of Care for Back Pain. Spine. 2018;43(14):1007-1017. doi:10.1097/BRS.0000000000002499 128. Grisafe D, Varma R, Bruce B, et al. Impact of Visual Field Loss on Vision-Specific Quality of Life in the African American Eye Disease Study. Pending. Published online Pending. 159 129. Ferris FL 3rd, Bailey I. Standardizing the measurement of visual acuity for clinical research studies: Guidelines from the Eye Care Technology Forum. Ophthalmology. 1996;103(1):181-182. 130. Chun YS, Lee DI, Kwon J, Park IK. Comparison of Impact of Monocular and Integrated Binocular Visual Fields on Vision-related Quality of Life. Journal of glaucoma. 2017;26(3):283-291. doi:10.1097/ijg.0000000000000623 131. Andrich D. Rating scales and Rasch measurement. http://dx.doi.org/101586/erp1159. Published online January 9, 2014. doi:10.1586/erp.11.59 132. Streiner DL, Norman GR, Cairney J. Health Measurement Scales: A Practical Guide to Their Development and Use. 5th ed. Oxford University Press; 2015. https://books.google.com/books/about/Health_Measurement_Scales.html?id=JH3OBAAAQ BAJ 133. Rizopoulos D. Latent Trait Models under Item Response Theory.; 2018. https://github.com/drizopoulos/ltm 134. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained equations in r. Journal of Statistical Software. 2011;45(3):1-67. 135. R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing; 2019. http://www.R-project.org/ 136. Wickham H. Ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York; 2016. https://ggplot2.tidyverse.org 137. lincom - Stata 16 Base Reference Manual. Published online 2019. 138. George S, Duran N, Norris K. A Systematic Review of Barriers and Facilitators to Minority Research Participation Among African Americans, Latinos, Asian Americans, and Pacific Islanders. Am J Public Health. 2014;104(2):e16-e31. doi:10.2105/AJPH.2013.301706 139. Braveman PA, Cubbin C, Egerter S, Williams DR, Pamuk E. Socioeconomic disparities in health in the United States: what the patterns tell us. Am J Public Health. 2010;100 Suppl 1:S186-196. doi:10.2105/AJPH.2009.166082 140. Friedman DS, West SK, Munoz B, et al. Racial variations in causes of vision loss in nursing homes: The Salisbury Eye Evaluation in Nursing Home Groups (SEEING) Study. Arch Ophthalmol. 2004;122(7):1019-1024. doi:10.1001/archopht.122.7.1019 141. Hayward MD, Miles TP, Crimmins EM, Yang Y. The Significance of Socioeconomic Status in Explaining the Racial Gap in Chronic Health Conditions. American Sociological Review. 2000;65(6):910-930. doi:10.2307/2657519 142. Seeman T, Merkin SS, Crimmins E, Koretz B, Charette S, Karlamangla A. Education, income and ethnic differences in cumulative biological risk profiles in a national sample of US adults: NHANES III (1988-1994). Soc Sci Med. 2008;66(1):72-87. doi:10.1016/j.socscimed.2007.08.027 160 143. Hodapp E, Parrish RK, Anderson DR. Clinical Decisions in Glaucoma. Mosby; 1993. 144. Lujan HL, DiCarlo SE. Science reflects history as society influences science: brief history of “race,” “race correction,” and the spirometer. Adv Physiol Educ. 2018;42(2):163-165. doi:10.1152/advan.00196.2017 145. Brook RD, Franklin B, Cascio W, et al. Air pollution and cardiovascular disease: A statement for healthcare professionals from the expert panel on population and prevention science of the American Heart Association. Circulation. 2004;109(21):2655-2671. doi:10.1161/01.CIR.0000128587.30041.C8 146. Brook RD, Rajagopalan S, Pope CA, et al. Particulate Matter Air Pollution and Cardiovascular Disease. Circulation. 2010;121(21):2331-2378. doi:10.1161/CIR.0b013e3181dbece1 147. Roger VL, Go AS, Lloyd-Jones DM, et al. Heart Disease and Stroke Statistics—2012 Update. Circulation. 2011;125(1):e2-e220. doi:10.1161/cir.0b013e31823ac046 148. Lelieveld J, Evans JS, Fnais M, Giannadaki D, Pozzer A. The contribution of outdoor air pollution sources to premature mortality on a global scale. Nature. 2015;525(7569):367-371. doi:10.1038/nature15371 149. Robertson S, Miller MR. Ambient air pollution and thrombosis. Particle and Fibre Toxicology. 2018;15(1):1. doi:10.1186/s12989-017-0237-x 150. Shah ASV, Lee KK, McAllister DA, et al. Short term exposure to air pollution and stroke: systematic review and meta-analysis. BMJ. 2015;350:h1295. doi:10.1136/bmj.h1295 151. Kilian J, Kitazawa M. The emerging risk of exposure to air pollution on cognitive decline and Alzheimer’s disease - Evidence from epidemiological and animal studies. Biomed J. 2018;41(3):141-162. doi:10.1016/j.bj.2018.06.001 152. Münzel T, Gori T, Al-Kindi S, et al. Effects of gaseous and solid constituents of air pollution on endothelial function. Eur Heart J. 2018;39(38):3543-3550. doi:10.1093/eurheartj/ehy481 153. Stieb DM, Zheng C, Salama D, et al. Systematic review and meta-analysis of case- crossover and time-series studies of short term outdoor nitrogen dioxide exposure and ischemic heart disease morbidity. Environ Health. 2020;19(1):47. doi:10.1186/s12940-020- 00601-1 154. Rozanova E, Heilig P, Godnic-Cvar J. The eye--a neglected organ in environmental and occupational medicine: an overview of known environmental and occupational non- traumatic effects on the eyes. Arh Hig Rada Toksikol. 2009;60(2):205-215. doi:10.2478/10004-1254-60-2009-1869 155. Adar SD, Klein R, Klein BEK, et al. Air Pollution and the Microvasculature: A Cross- Sectional Assessment of In Vivo Retinal Images in the Population-Based Multi-Ethnic Study of Atherosclerosis (MESA). PLoS Med. 2010;7(11). doi:10.1371/journal.pmed.1000372 161 156. Wong TY, Klein R, Nieto FJ, et al. Retinal microvascular abnormalities and 10-year cardiovascular mortality: a population-based case-control study. Ophthalmology. 2003;110(5):933-940. doi:10.1016/S0161-6420(03)00084-8 157. Wong TY, Kamineni A, Klein R, et al. Quantitative retinal venular caliber and risk of cardiovascular disease in older persons: the cardiovascular health study. Arch Intern Med. 2006;166(21):2388-2394. doi:10.1001/archinte.166.21.2388 158. McGeechan K, Liew G, Macaskill P, et al. Prediction of incident stroke events based on retinal vessel caliber: a systematic review and individual-participant meta-analysis. Am J Epidemiol. 2009;170(11):1323-1332. doi:10.1093/aje/kwp306 159. Mammo Z, Heisler M, Balaratnasingam C, et al. Quantitative Optical Coherence Tomography Angiography of Radial Peripapillary Capillaries in Glaucoma, Glaucoma Suspect, and Normal Eyes. Am J Ophthalmol. 2016;170:41-49. doi:10.1016/j.ajo.2016.07.015 160. Holló G. Vessel density calculated from OCT angiography in 3 peripapillary sectors in normal, ocular hypertensive, and glaucoma eyes. Eur J Ophthalmol. 2016;26(3):e42-45. doi:10.5301/ejo.5000717 161. Chua SYL, Khawaja AP, Dick AD, et al. Ambient Air Pollution Associations with Retinal Morphology in the UK Biobank. Invest Ophthalmol Vis Sci. 2020;61(5):32. doi:10.1167/iovs.61.5.32 162. Mikati I, Benson AF, Luben TJ, Sacks JD, Richmond-Bryant J. Disparities in Distribution of Particulate Matter Emission Sources by Race and Poverty Status. Am J Public Health. 2018;108(4):480-485. doi:10.2105/AJPH.2017.304297 163. Jones MR, Diez-Roux AV, Hajat A, et al. Race/Ethnicity, Residential Segregation, and Exposure to Ambient Air Pollution: The Multi-Ethnic Study of Atherosclerosis (MESA). Am J Public Health. 2014;104(11):2130-2137. doi:10.2105/AJPH.2014.302135 164. Clark LP, Millet DB, Marshall JD. National Patterns in Environmental Injustice and Inequality: Outdoor NO2 Air Pollution in the United States. PLOS ONE. 2014;9(4):e94431. doi:10.1371/journal.pone.0094431 165. Tham YC, Li X, Wong TY, Quigley HA, Aung T, Cheng CY. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta- analysis. Ophthalmology. 2014;121(11):2081-2090. doi:10.1016/j.ophtha.2014.05.013 166. Restrepo NA, Cooke Bailey JN. Primary Open-Angle Glaucoma Genetics in African Americans. Curr Genet Med Rep. 2017;5(4):167-174. doi:10.1007/s40142-017-0131-8 167. Morello-Frosch Rachel, Shenassa Edmond D. The Environmental “Riskscape” and Social Inequality: Implications for Explaining Maternal and Child Health Disparities. Environmental Health Perspectives. 2006;114(8):1150-1153. doi:10.1289/ehp.8930 168. Geer LA. Identifying exposure disparities in air pollution epidemiology specific to adverse birth outcomes. Environ Res Lett. 2014;9(10):101001. doi:10.1088/1748-9326/9/10/101001 162 169. Lee EY, Oh SS, White MJ, et al. Ambient air pollution, asthma drug response, and telomere length in African American youth. J Allergy Clin Immunol. 2019;144(3):839- 845.e10. doi:10.1016/j.jaci.2019.06.009 170. Douglas JA, Archer RS, Alexander SE. Ecological determinants of respiratory health: Examining associations between asthma emergency department visits, diesel particulate matter, and public parks and open space in Los Angeles, California. Preventive Medicine Reports. 2019;14:100855. doi:10.1016/j.pmedr.2019.100855 171. Morello-Frosch Rachel, Jesdale Bill M. Separate and Unequal: Residential Segregation and Estimated Cancer Risks Associated with Ambient Air Toxics in U.S. Metropolitan Areas. Environmental Health Perspectives. 2006;114(3):386-393. doi:10.1289/ehp.8500 172. Su JG, Morello-Frosch R, Jesdale BM, Kyle AD, Shamasunder B, Jerrett M. An Index for Assessing Demographic Inequalities in Cumulative Environmental Hazards with Application to Los Angeles, California. Environ Sci Technol. 2009;43(20):7626-7634. doi:10.1021/es901041p 173. Houston D, Wu J, Ong P, Winer A. Structural Disparities of Urban Traffic in Southern California: Implications for Vehicle-Related Air Pollution Exposure in Minority and High- Poverty Neighborhoods. Journal of Urban Affairs. 2004;26(5):565-592. doi:10.1111/j.0735- 2166.2004.00215.x 174. Kashani AH, Chen C-L, Gahm JK, et al. Optical coherence tomography angiography: A comprehensive review of current methods and clinical applications. Prog Retin Eye Res. 2017;60:66-100. doi:10.1016/j.preteyeres.2017.07.002 175. Chang R, Nelson AJ, LeTran V, et al. Systemic Determinants of Peripapillary Vessel Density in Healthy African Americans: The African American Eye Disease Study. Am J Ophthalmol. 2019;207:240-247. doi:10.1016/j.ajo.2019.06.014 176. Chu Z, Lin J, Gao C, et al. Quantitative assessment of the retinal microvasculature using optical coherence tomography angiography. J Biomed Opt. 2016;21(6):66008. doi:10.1117/1.JBO.21.6.066008 177. US Environmental Protection Agency. Air Quality System Data Mart. https://www.epa.gov/outdoor-air-quality-data/download-daily-data 178. Wood S. Mgcv: Mixed GAM Computation Vehicle with Automatic Smoothness Estimation. 179. Goldberg D. Geocoding Services. Texas A&M University GeoServices. https://geoservices.tamu.edu/ 180. Budenz DL, Anderson DR, Varma R, et al. Determinants of Normal Retinal Nerve Fiber Layer Thickness Measured by Stratus OCT. Ophthalmology. 2007;114(6):1046-1052. doi:10.1016/j.ophtha.2006.08.046 181. Rao HL, Pradhan ZS, Weinreb RN, et al. Determinants of Peripapillary and Macular Vessel Densities Measured by Optical Coherence Tomography Angiography in Normal Eyes. J Glaucoma. 2017;26(5):491-497. doi:10.1097/IJG.0000000000000655 163 182. Shweikh Y, Ko F, Chan MPY, et al. Measures of socioeconomic status and self-reported glaucoma in the UK Biobank cohort. Eye (Lond). 2015;29(10):1360-1367. doi:10.1038/eye.2015.157 183. Ko F, Boland MV, Gupta P, et al. Diabetes, Triglyceride Levels, and Other Risk Factors for Glaucoma in the National Health and Nutrition Examination Survey 2005–2008. Invest Ophthalmol Vis Sci. 2016;57(4):2152-2157. doi:10.1167/iovs.15-18373 184. An R, Ji M, Yan H, Guan C. Impact of ambient air pollution on obesity: a systematic review. Int J Obes (Lond). 2018;42(6):1112-1126. doi:10.1038/s41366-018-0089-y 185. Kim JS, Chen Z, Alderete TL, et al. Associations of air pollution, obesity and cardiometabolic health in young adults: The Meta-AIR study. Environ Int. 2019;133(Pt A):105180. doi:10.1016/j.envint.2019.105180 186. Yang B-Y, Qian Z, Howard SW, et al. Global association between ambient air pollution and blood pressure: A systematic review and meta-analysis. Environ Pollut. 2018;235:576- 588. doi:10.1016/j.envpol.2018.01.001 187. Eze IC, Hemkens LG, Bucher HC, et al. Association between ambient air pollution and diabetes mellitus in Europe and North America: systematic review and meta-analysis. Environ Health Perspect. 2015;123(5):381-389. doi:10.1289/ehp.1307823 188. Kifley A, Liew G, Wang JJ, et al. Long-term effects of smoking on retinal microvascular caliber. Am J Epidemiol. 2007;166(11):1288-1297. doi:10.1093/aje/kwm255 189. Hsu C-H, Chen RI, Lin SC. Myopia and glaucoma: sorting out the difference. Curr Opin Ophthalmol. 2015;26(2):90-95. doi:10.1097/ICU.0000000000000124 190. Wei C-C, Lin H-J, Lim Y-P, et al. PM2.5 and NOx exposure promote myopia: clinical evidence and experimental proof. Environ Pollut. 2019;254(Pt B):113031. doi:10.1016/j.envpol.2019.113031 191. Venugopal JP, Rao HL, Weinreb RN, et al. Repeatability of vessel density measurements of optical coherence tomography angiography in normal and glaucoma eyes. Br J Ophthalmol. 2018;102(3):352-357. doi:10.1136/bjophthalmol-2017-310637 192. Chua SYL, Khawaja AP, Morgan J, et al. The Relationship Between Ambient Atmospheric Fine Particulate Matter (PM2.5) and Glaucoma in a Large Community Cohort. Invest Ophthalmol Vis Sci. 2019;60(14):4915-4923. doi:10.1167/iovs.19-28346 193. Younan D, Tuvblad C, Franklin M, et al. Longitudinal Analysis of Particulate Air Pollutants and Adolescent Delinquent Behavior in Southern California. J Abnorm Child Psychol. 2018;46(6):1283-1293. doi:10.1007/s10802-017-0367-5 194. Franklin M, Fruin S. The role of traffic noise on the association between air pollution and children’s lung function. Environ Res. 2017;157:153-159. doi:10.1016/j.envres.2017.05.024
Abstract (if available)
Abstract
Chronic diseases of the eye are becoming more prevalent as the average age of the population continues to climb. Most non‐acute eye diseases present in adults aged 40 years or older. Epidemiologic studies over the last several decades have observed that certain eye pathologies are more common among certain racial/ethnic groups. For example, glaucoma is more prevalent in Latinos and African Americans, age‐related macular degeneration is more common in non‐Hispanic Whites, and diabetic retinopathy is more prevalent in Chinese Americans. How people are impacted by chronic eye diseases might also be related to race/ethnicity. Furthermore, degree of visual function—not just prevalence—may prove to be more informative and precise measures for characterizing disparities in eye disease. ❧ A functional perspective of vision focuses on the experiences of individuals suffering from chronic eye diseases, and not on diagnoses of diseases made by panels of expert physicians. The impact of eye pathologies has been shown to be largely mediated through vision loss. For example, a patient with early stage glaucoma who has not suffered loss of vision—visual impairment (VI)—is likely not impaired by their disease. VI may be defined by numerous measures of vision, and all of them should be characterized to understand how chronic eye disease affects patients. ❧ Visual acuity (VA) is a metric of central, high acuity vision that is measured using an illuminated chart the patient reads aloud to the healthcare provider. VA of 20/40 or worse is a standard definition of VI that is commonly used in both the clinic and research. However, there are other, less familiar measures of VI. Visual field loss (VFL) measures the entire visual field—both central and peripheral vision—and can be used to locate specific patterns of vision loss in a patient’s visual field. However, a total score can also be used to provide a global assessment of visual ability separately for each eye. VFL is quantified on a logarithmic scale as the difference in vision from a standard population of healthy adults of the same age. One definition of VI using VFL is -2 decibels (dB) in the better‐seeing eye (BSE) of total mean deviation (MD) from an age‐adjusted standard population. ❧ Racial/ethnic disparities in chronic eye disease should be assessed using both VFL and VA as well as other measures of vision such as contrast sensitivity. Considering continuous measures of visual function improves our understanding of the extent of vision loss for each racial/ethnic group as well as the dimensions of vision that are most affected. For example, population‐based cohort studies have demonstrated that VI defined by VFL is more common in racial/ethnic minorities compared to non‐Hispanic Whites. To see the full picture of ophthalmic illness, however, we must look beyond objective measures of visual impairment. Modern research investigates patients’ perspectives of their quality of life (QOL) to fully understand the impact of disease and to design effective treatments. ❧ Health‐related quality of life (HRQOL) is a latent construct with numerous definitions, but overall is used to measure the value of life for several domains including the ability to perform tasks and socioemotional well‐being. HRQOL has been shown to vary among racial/ethnic groups for cancer patients after adjusting for demographic factors and socioeconomic status (SES). Cultural differences in beliefs of how health is related to illness and death have been identified as explanations for racial/ethnic differences in HRQOL among terminal patients. Race/ethnicity is related to how people find meaning during the end of life. Might similar differences exist in how people find value related to their vision when suffering from chronic eye disease? ❧ Vision‐specific QOL (VSQOL) has been employed by ophthalmologists and investigators for several decades to capture the value of life related to one’s visual ability. VSQOL is another quantitative measure of visual function—like VA and VFL—but focuses on patients’ subjective experiences. Because chronic eye diseases are predominantly disabling only when they cause VI, the association of VI and VSQOL is essential to understanding how chronic eye disease affects patients. VSQOL has been used as an outcome variable in clinical trials to assess whether ophthalmologic interventions are effective in treating chronic eye diseases. In addition, VSQOL has also been used at the population level to assess the perception of vision loss and how it impacts daily life in the community. This information is useful for prioritizing public health resources to improve domains of visual health that are most meaningful for multiethnic groups. Of course, this is only relevant if race/ethnicity modifies the relationship between visual impairment and VSQOL. ❧ Although VSQOL is a subset of QOL, differences have been identified in the literature. Only VSQOL, but not HRQOL, appears to be affected by vision loss. VSQOL is specific to vision but not to health in general, which may explain why broader measures of QOL are not sensitive to VI. Furthermore, VSQOL is measured using instruments that were developed and validated for multiethnic populations. Given that VSQOL is a more specific domain than HRQOL and has demonstrated construct validity in diverse populations, do cultural beliefs and other racial/ethnic factors modify the association between patients’ subjective experiences and quantitative measures of visual impairment? Variation in the prevalence of chronic eye diseases exist by race/ethnicity. Do people experiencing the same quantitative loss in vision have similar magnitudes and patterns of reduced QOL specific to their vision, regardless of their race/ethnicity? ❧ VSQOL has been associated with VA, VFL, and other continuous measures of visual impairment among non‐Hispanic Whites, Latinos, and populations of African descent outside the US. But no studies have assessed these associations in large groups of African Americans or cross‐cultural studies of the most populous racial/ethnic groups in the United States (US). The primary aim of the first two papers in this dissertation are to elucidate how VSQOL may be differentially impacted by VFL for major US racial/ethnic groups. In the first paper we validate modern psychometric measures of VSQOL in the largest, population‐based cohort of chronic eye disease ever conducted in African Americans. In this study, we assess whether clinically meaningful differences in VSQOL are associated with similar magnitudes and patterns of VFL in non‐Hispanic White and Latino populations. The second paper is a pooled analysis concerned with harmonizing methods to assess whether similar associations between VFL and VSQOL exist for Latinos, Chinese Americans, and African Americans. The second study concludes our assessment of whether differences exist in how vision loss impacts VSQOL by race/ethnicity. From here we move on to investigate why differences in chronic eye diseases exist in the first place, and whether air pollution might be an important explanatory exposure. ❧ Recall that Latinos and African Americans have a higher prevalence of open angle glaucoma (OAG), which is the leading cause of irreversible blindness globally. However, the scientific community has not identified why this disparity exists. Several genetic studies have found evidence that a limited number of genetic markers may increase the risk of OAG among participants with African Ancestry. But small effects do not fully explain the relatively large racial/ethnic difference. ❧ In the third paper of this dissertation we identify a modest association between traffic‐related air pollution (TRAP) and reduced blood flow to retinal vessels implicated in OAG in a cohort of African Americans. The pathophysiology of OAG involves the death of retinal ganglion fiber cells, possibly due to deleterious effects of the retinal microvasculature surrounding the optic nerve head (ONH). TRAP has been shown to cause damage to microvasculature in other areas of the body, leading to cardiovascular disease, stroke, vascular Alzheimer’s disease and overall mortality. Also, exposure to TRAP has been shown to be greater for impoverished people and people of color in the US. Optical coherence tomography angiography (OCTA) is a noninvasive imaging methodology that was used to measure retinal blood flow in the vessels surrounding the ONH. These findings complement recent studies of large hospital biobanks that identified TRAP measured at patient addresses was associated with changes in the thickness of retinal nervous tissue associated. We propose that the next population‐based study of eye disease should implement OCTA to more precisely assess TRAP exposure and retinal blood flow in a larger cohort of participants. ❧ This dissertation characterizes multiethnic disparities in domains of chronic eye disease epidemiology including subjective experience related to vision, objective visual function, anatomical measurements of the eye and environmental exposures. Study participants are from the Multiethnic Ophthalmology Cohorts of California Study (MOCCaS), which is composed of population‐based cohorts of Latino, Chinese American, and African American participants living in Los Angeles County. Technical abstracts are available later in this dissertation at the beginning of each chapter. The first paper proposes and assesses the psychometric validity of a VSQOL instrument in a large cohort of African Americans for the first time. Associations are identified between patterns of VFL and VSQOL related to tasks and socioemotional wellbeing. The second paper identifies racial/ethnic differences in VFL and VSQOL across all three racial/ethnic groups using harmonized methods. The final paper investigates a burgeoning hypothesis that TRAP is associated with OAG as measured by blood perfusion to vessels in the eye.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
The multiethnic nature of chronic disease: studies in the multiethnic cohort
PDF
Vision epidemiology and the impact of vision loss on vision-specific quality of life
PDF
Visual acuity outcomes after cataract extraction in Chinese Americans: the Chinese American Eye Study (CHES)
PDF
Diet quality and pancreatic cancer incidence in the multiethnic cohort
PDF
Utility of polygenic risk score with biomarkers and lifestyle factors in the multiethnic cohort study
PDF
Risk factors of pelvic floor disorders in the multiethnic cohort study
PDF
Pharmacogenetic association studies and the impact of population substructure in the women's interagency HIV study
PDF
Identifying genetic, environmental, and lifestyle determinants of ethnic variation in risk of pancreatic cancer
PDF
Prenatal environmental exposures and fetal growth in the MADRES cohort
PDF
Genetic epidemiological approaches in the study of risk factors for hematologic malignancies
PDF
Age related macular degeneration in Latinos: risk factors and impact on quality of life
PDF
Carcinogenic exposures in racial/ethnic groups
PDF
Racial/ethnic differences in colorectal cancer patient experiences, health care utilization and their association with mortality: findings from the SEER-CAHPS data
Asset Metadata
Creator
Grisafe, Dominic Joseph, II
(author)
Core Title
Chronic eye disease epidemiology in the multiethnic ophthalmology cohorts of California study
School
Keck School of Medicine
Degree
Doctor of Philosophy
Degree Program
Epidemiology
Publication Date
06/16/2020
Defense Date
05/29/2020
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
African American Eye Disease Study,Chinese Americans Eye Study,classical test theory,environment,environmental epidemiology,environmental exposure,Epidemiology,generalized additive models,graded response model,health-related quality of life,item response theory,locally weighted scatterplot smoothing,Los Angeles Latino Eye Study,multiethnic health disparities,multiple imputation with chained equations,multivariable linear regression,National Eye Institute Visual Functioning Questionnaire-25,nitrogen dioxide,NO2,OAI-PMH Harvest,open angle glaucoma,Ophthalmology,optic nerve head,optical coherence tomography angiography,particulate matter with aerodynamic diameter less than 2.5 µm,psychometrics,Quality of life,radial peripapillary capillaries,retinal blood flow,traffic related air pollution,vessel area density,vision-specific quality of life,visual acuity,visual angle,visual field,visual field loss,visual function,visual impairment
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
McKean-Cowdin, Roberta (
committee chair
), Baezconde-Garbanati, Lourdes (
committee member
), Mack, Wendy (
committee member
), Patino-Sutton, Cecilia (
committee member
), Xu, Benjamin (
committee member
)
Creator Email
dgrisafe@gmail.com,dgrisafe@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-318788
Unique identifier
UC11663607
Identifier
etd-GrisafeDom-8593.pdf (filename),usctheses-c89-318788 (legacy record id)
Legacy Identifier
etd-GrisafeDom-8593.pdf
Dmrecord
318788
Document Type
Dissertation
Rights
Grisafe, Dominic Joseph, II
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 a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
African American Eye Disease Study
Chinese Americans Eye Study
classical test theory
environment
environmental epidemiology
environmental exposure
generalized additive models
graded response model
health-related quality of life
item response theory
locally weighted scatterplot smoothing
Los Angeles Latino Eye Study
multiethnic health disparities
multiple imputation with chained equations
multivariable linear regression
National Eye Institute Visual Functioning Questionnaire-25
nitrogen dioxide
NO2
open angle glaucoma
optic nerve head
optical coherence tomography angiography
particulate matter with aerodynamic diameter less than 2.5 µm
psychometrics
radial peripapillary capillaries
retinal blood flow
traffic related air pollution
vessel area density
vision-specific quality of life
visual acuity
visual angle
visual field
visual field loss
visual function
visual impairment