University of Southern California Dissertations and Theses

Vision epidemiology and the impact of vision loss on vision-specific quality of life

(USC Thesis Other)

Vision epidemiology and the impact of vision loss on vision-specific quality of life

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Vision Epidemiology and the Impact of Vision Loss on Vision-Specific Quality of Life
By
Malcolm E. Barrett III
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(EPIDEMIOLOGY)
May 2021
Copyright © 2021 Malcolm E. Barrett III
This dissertation is dedicated to the love of my life, Abby Keener; to my parents, Cathy
Barrett and Gene Barrett; and to my grandparents, Judy Zych and Edward Zych.
I also dedicate this work to the health and well-being of those affected by the COVID-19
pandemic that occurred while writing this dissertation. May they heal all their ills.
ii
ACKNOWLEDGEMENTS
Thank you to my committee, Drs. Paul Marjoram, Roberta McKean-Cowdin, Victoria
Cortessis, Chih-Ping Chou, and Grace Richter.
ThankyoutoShugenRoshi, HojinSensei, andallteachersandstudentsoftheMountain
and Rivers Order. Thank you to my mom, dad, and my grandparents. Thank you to Abby,
Suki, Phoebe, Callisto, Tika, and Nanapush.
Thank you to the participants of LALES and CHES, as well as to the researchers and
staff that maintain these studies.
May all your lives go well.
iii
TABLE OF CONTENTS
DEDICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Chapter
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1 Measuring Vision-Specific Quality of Life . . . . . . . . . . . . . . . . . . . . 5
2.2 Specifying dimensions with structural equation models . . . . . . . . . . . . 7
2.3 Measuring Quality of Life with Item Response Theory . . . . . . . . . . . . . 8
2.4 Confounding, DAGs, and sensitivity analyses . . . . . . . . . . . . . . . . . . 11
2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.7 Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3 The Impact of 8-year Change in Visual Acuity on New NEI-VFQ-25 Composites
in the Los Angeles Latino Eye Study (LALES) . . . . . . . . . . . . . . . . . . . . 37
3.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2.1 Sociodemographic and Clinical Data . . . . . . . . . . . . . . . . . . 40
3.2.2 Visual Acuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2.3 Health-Related and Vision-Specific Quality of Life . . . . . . . . . . . 41
3.2.3.1 Medical Outcomes Study 12-Item Short Form Health Survey 41
3.2.3.2 The National Eye Institute Visual Function Questionnaire . 42
3.2.4 Statistical Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3.1 Description of Study Population . . . . . . . . . . . . . . . . . . . . . 45
3.3.2 Overall Changes in Health-Related Quality of Life and Visual Acuity 46
3.3.3 Psychometric Properties of Task and Well-Being Composites . . . . . 47
3.3.4 Changes in Health-Related Quality of Life by Visual Acuity . . . . . 47
3.3.5 PredictedMeanChangeinVS-QOLat2-and4-LinesChangeinVisual
Acuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
iv
3.5 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.7 Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4 Evaluating the Effect of Vision Loss on Quality of Life in the Chinese Eye Study
(CHES) with Item Response Theory . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.2.1 Sociodemographic and Clinical Data . . . . . . . . . . . . . . . . . . 82
4.2.2 Visual Acuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.2.3 Health-Related and Vision-Specific Quality of Life . . . . . . . . . . . 83
4.2.3.1 Medical Outcomes Study 12-Item Short-Form Health Survey
and the National Eye Institute Visual Function Questionnaire 83
4.2.3.2 Item Response Model Estimation of Vision-Specific Quality
of Life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.2.4 Statistical Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.3.1 Description of Study Population . . . . . . . . . . . . . . . . . . . . . 87
4.3.2 Health-Related Quality of Life and Visual Acuity . . . . . . . . . . . 87
4.3.3 Item and Test Information Curves . . . . . . . . . . . . . . . . . . . . 89
4.3.4 Predicted Mean Change in VS-QOL by Visual Acuity . . . . . . . . . 90
4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.5 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.6 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.7 Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5 Summary of Findings and Future Directions . . . . . . . . . . . . . . . . . . . . . 120
5.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.2 Future Directions and Public Health Impacts . . . . . . . . . . . . . . . . . . 121
5.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
5.4 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
A Supplement to Chapter 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
A.1 A primer on causal structural models and directed acyclic graphs . . . . . . 142
B Supplement to Chapter 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
B.0.1 Scoring for new subscales . . . . . . . . . . . . . . . . . . . . . . . . . 146
B.0.2 DAG of causal assumptions for selecting covariates . . . . . . . . . . 149
B.0.3 Contrasts without Inverse Probability of Censoring Weights . . . . . 151
B.0.3.1 VS-QOL by Cause of Visual Impairment . . . . . . . . . . . 152
v
B.0.3.2 Modification Indices . . . . . . . . . . . . . . . . . . . . . . 153
C Supplement to Chapter 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
C.0.1 NEI-VFQ-25 Item Descriptions and Discrimination . . . . . . . . . . 154
C.0.2 CTT-Based Task and Well-Being Composites . . . . . . . . . . . . . 156
C.0.3 DAG of causal assumptions for selecting covariates . . . . . . . . . . 159
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
vi
LIST OF TABLES
2.1 Items in the 25-Item National Eye Institute Visual Function Questionnaire . 29
3.1 Overall Change in Visual Acuity and QOL . . . . . . . . . . . . . . . . . . . 71
3.2 Confirmatory Factor Analysis of NEI-VFQ-25 Scales . . . . . . . . . . . . . 73
3.3 Change in VS-QOL by Change in Visual Acuity (Best Presenting Eye) . . . 74
3.4 Predicted Mean Change of VS-QOL at Levels of Visual Acuity Change . . . 75
4.1 Overall Correlation and Difference in Visual Acuity and QOL . . . . . . . . 113
4.2 Predicted Mean Difference of VS-QOL at Levels of Visual Acuity Difference 118
B.1 Traditional and Improved Scales for Items in the 25-Item National Eye Insti-
tute Visual Function Questionnaire . . . . . . . . . . . . . . . . . . . . . . . 146
B.2 Predicted Mean Change of VS-QOL at Levels of Visual Acuity Change with-
out IPCW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
B.3 VS-QOL among those with incident VI/blindness by cause . . . . . . . . . . 152
B.4 Top 10 Modification Indices for Task and Well-Being CFA . . . . . . . . . . 153
C.1 NEI-VFQ-25 Item Descriptions and Discrimination . . . . . . . . . . . . . . 154
C.2 Predicted Mean Difference of VS-QOL (CTT) at Levels of Visual Acuity Dif-
ference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
vii
LIST OF FIGURES
2.1 A Basic Structural Equation Model . . . . . . . . . . . . . . . . . . . . . . . 32
2.2 Confirmatory Factor Analysis of NEI-VFQ-25: Composite Score . . . . . . . 33
2.3 Confirmatory Factor Analysis of NEI-VFQ-25: Task and Well-being . . . . . 34
2.4 Confirmatory Factor Analysis of NEI-VFQ-25: All Subscales . . . . . . . . . 35
2.5 A Theoretical Item Response Model . . . . . . . . . . . . . . . . . . . . . . . 36
3.1 Change in Vision-Related Quality of Life and Visual Acuity by Subscale, Mul-
tivariably Adjusted . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3.2 G-Formula Calculated Changes in Vision-Related Quality of Life at 2 and 4
Lines Change in Visual Acuity . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.1 Distribution of Vision-Specific Quality of Life and Relationship with Visual
Acuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
4.2 Test and Item Information Functions of the Task and Well-Being Composites 117
4.3 Difference in Vision-Related Quality of Life and Visual Acuity by IRT-based
Composite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
A.1 Confounding as a Directed Acyclic Graph: Blocking Back-Door Paths . . . . 144
A.2 Selection Bias as a Directed Acyclic Graph: Stratifying on a Collider . . . . 145
B.1 Causal Diagram for the Effect of Visual Acuity on Quality of Life . . . . . . 150
B.2 G-Formula Calculated Changes in Vision-Related Quality of Life at 2 and 4
Lines Change in Visual Acuity without IPCW . . . . . . . . . . . . . . . . . 152
C.1 Distribution of CTT-Based Vision-Specific Quality of Life and Relationship
with Visual Acuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
C.2 Difference in Vision-Related Quality of Life and Visual Acuity by CTT-based
Composite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
C.3 Causal Diagram for the Effect of Visual Acuity on Quality of Life . . . . . . 159
viii
ABSTRACT
In this dissertation I investigate the effect that eye diseases and vision loss have on
quality of life (QOL) and functional well-being. With the population aging and prevalence
of visual impairment projected to double in the next 35 years, understanding how visual
acuity affects our quality of life is essential if we seek to appropriately set public health
priorities and help communicate treatment benefits to clinicians and patients. Using data
from two eye disease cohorts, the Los Angeles Latino Eye Study (LALES) and the Chinese
Eye Study (CHES), I attempt to solve several epidemiological and methodological problems
in the study of vision loss and its impact on vision-specific quality of life. In Chapter 4,
"The Impact of 8-year Change in Visual Acuity on New NEI-VFQ-25 Composites in the
Los Angeles Latino Eye Study (LALES)," I investigate the impact that 8-year change in
visual acuity (the sharpness of your vision) has on quality of life in adult Latinos. I also
improve an existing tool that measures vision-specific quality of life. In its original form, the
composite score was multidimensional and thus difficult to interpret. We created two new
subscales that better embody vision-specific QOL: well-being (e.g., social engagement) and
task-related functioning (e.g. being able to read and work). Using the parametric G-formula,
we found that vision loss reduces both well-being and task-related functioning. In Chapter 5,
"Evaluating the Effect of Vision Loss on Quality of Life in the Chinese American Eye Study
(CHES) with Item Response Theory," I estimate the impact of visual acuity on quality of
life in Chinese American participants, the first such study in this population. I use item
response theory rather than the classical measurement for calculating QOL. I also set to
anchor the interpretation of the item response scores to visual acuity, an important clinical
marker, in order to help clinicians and patients better understand changing quality of life
scores. I found that visual acuity loss also decreases well-being and task-specific functioning
in this population.
ix
CHAPTER 1
INTRODUCTION
In this dissertation I investigate the effect that eye diseases and vision loss have on
quality of life and functional well-being. Vision contributes to the ability of people to pro-
cess information around them and to engage in everyday tasks and social situations. Visual
impairment often disrupts these aspects of life. The prevalence of visual impairment will
rise as US population ages. Understanding how visual acuity affects our quality of life is es-
sential to setting public health priorities and communicating treatment benefits to clinicians
and patients. Using data from two eye disease cohorts, the Los Angeles Latino Eye Study
(LALES) and the Chinese Eye Study (CHES), I attempt to solve several epidemiological and
methodological problems in the study of vision loss and its impact on vision-specific quality
of life.
In Chapter 4, “The Impact of 8-year Change in Visual Acuity on New NEI-VFQ-25
Composites in the Los Angeles Latino Eye Study (LALES),” I investigate the impact that
8-year change in visual acuity (the sharpness of your vision) has on quality of life in adult
Latinos. I also improve an existing tool that measures vision-specific quality of life. In
its original form, the composite score of the NEI-VFQ-25 (a single-number summary of
all items in the questionnaire) was multidimensional and thus difficult to interpret. While
there were 12 subscales for this tool, we found it implausible that there were that many
dimensions to vision-specific QOL; those subscales more likely represent clusters of eye and
vision disorders. We created two new subscales that better embody vision-specific QOL:
well-being (e.g., social engagement) and task-related functioning. Using the parametric G-
formula, I found that vision loss reduces both well-being and task-related functioning.
Chapter 5, “Evaluating the Effect of Vision Loss on Quality of Life in the Chinese
1
American Eye Study (CHES) with Item Response Theory,” I estimate the impact of visual
acuity on quality of life in Chinese American participants, the first such study in this pop-
ulation. In it, I am building on my work in Chapter 4 and the work of others to improve
the measurement of vision-specific QOL. I use item response theory rather than the classical
measurement for calculating QOL. Item response theory has some benefits: it may better
measure QOL and increase power in longitudinal studies; it’s also easier to compare item
response scores across scales. Switching to item response theory for measurement has some
complications, though. In particular, it’s difficult to compare scores from past studies that
use standard scoring. Thus I also set to anchor the interpretation of the item response scores
to visual acuity, an important clinical marker, in order to help clinicians and patients better
understand changing quality of life scores. I found that visual acuity loss also decreases
well-being and task-specific functioning in this population and to quantify the relationship
between scores on the item response scale and visual acuity.
Evaluating the impact of vision loss on quality of life is essential for setting clinical
and public health priorities. As the population ages, the impact of vision loss on quality
of life is likely to increase. Improving our tools and analyses to better communicate this
impact–whether its to clinicians and patients considering a treatment course or to policy
makers allocating limited public health resources–is essential to reducing the suffering that
comes along with vision loss.
2
CHAPTER 2
BACKGROUND
Vision is an important aspect of day-to-day life. It contributes to the ability of people
to engage in everyday tasks and take in the information around them. People with visual
impairment often have aspects of their lives interrupted, such as working, reading, driving,
and spending time with others. Vision loss is also associated with chronic health conditions,
depression, isolation, and mortality (1–7). As the US population ages, the prevalence of
visual impairment is projected to double from 3.22 million in 2015 to 6.95 million in 2050
(8). Latinos are particularly at risk: the incidence of visual impairment is higher for US
Latinos than white Americans, particularly for younger Latinos, who are 7 times more likely
to develop VI (8). Our previous work has shown that visual acuity (VA) was associated
with quality of life, as were other eye conditions, such as visual field loss, glaucoma, diabetic
retinopathy, and AMD (9–14). Other clinical and population-based studies have show that
visual impairment worsens quality of life (15–19).
Quality of life is also a core aspect of public health. As life expectancy and treatments
for diseases improve, quality of life is increasingly a guiding compass for paths of health care,
as well as the allocation of our limited resources. We have come to understand that wellness
cannot just be defined as the absence of disease: quality of life addresses the many-faceted
experience of our wellness, including our physical, mental, and social health, often referred
to collectively as health-related quality of life (HRQOL). The Healthy People initiative for
setting public health goals has included HRQOL as core target since Healthy People 2000
(20). Vision-specific quality of life (VSQOL) is the subset of HRQOL concerned with the
effects vision and eye health have on our lives (21).
The primary drivers of changing vision-specific quality of life are the progression or
3
treatment of ophthalmic disorders. Loss of VS-QOL is generally due to age and pathology-
related conditions, while improvement in VS-QOL is due almost entirely to treatment, such
as corrective lenses or cataract surgery (22). In a 2013 meta-analysis, Bourne et al. found
that, in 2010, the leading causes of impairment globally were uncorrected refractive error
(52.9%), cataracts (18.4%), macular degeneration (3.1%), glaucoma (2.2%), and diabetic
retinopathy (1.9%) (23). The leading causes of blindness were similar: cataracts (33.4%),
uncorrected refractive error (20.9%), macular degeneration (6.6%), glaucoma (6.6%), and
diabetic retinopathy (2.6%). In LALES, the four most frequent causes of 4-year incident
VI were refractive error, cataract, diabetic retinopathy, age-related macular degeneration,
and glaucoma (24). In CHES, the most frequent causes were cataract, degenerative myopia,
glaucoma, corneal opacity, macular degeneration, diabetic retinopathy, and other optic nerve
damage (25). Although the distribution of causes depends on the population, the most
important causes were consistent across Latino, Chinese, and other populations (26–36).
In addition to pathology and treatment, psycho-socio-cultural factors also contribute
to the experience of quality of life. For immigrant populations, acculturation to the Anglo-
centric culture of the US also impacts VS-QOL (37). Notably, a primary way acculturation
affects VS-QOL is through treatment: acculturation is a strong predictor of access to care
and pursuit of follow-up care in LALES, CHES, and other studies (38–42). While health
beliefs play a role, language is the primary mediator how acculturation affects access to care;
immigrants who spoke the language of their host country were much more likely to access
care (43). However, acculturation is a complex, multi-faceted phenomenon, and, in the US,
there are several adverse eye-related health associations. For instance, for immigrants to
urban regions, acculturation is also associated with a higher incidence of diabetes (possibly
due to diet changes) and higher rates of myopia (possibly due to more near-work) (44).
Quantifying VS-QOL is also a challenge. How do we measure and analyze VS-QOL,
making the most valid conclusions on the impact of vision change on functioning and well-
being? This question is seated in measurement error, confounding, and selection bias. Cal-
4
culating quality of life requires careful attention to measurement error, as is often the case
in latent variable methodology. There are many ways to improve latent variable models,
but two that are common are to 1) hone the underlying latent variable being measured by
adding or dropping items or dimensions (subscales) and 2) use an alternate mechanism, such
as item response theory (21,45,46). Confounding and selection bias are ubiquitous problems
in observational research, but many tools exist to address these problems that have not been
commonly applied in vision epidemiology. One such tool is the causal directed acyclic graph
(DAG), which can easily solve complicated etiologic problems in modeling (47). A second
set of tools are G-methods, modeling approaches related to DAGs and the counterfactual
framework. In this dissertation, I use two approaches from G-methods to address confound-
ing and selection bias: the parametric G-formula and inverse probability of censoring weights
(to control for loss to follow-up) (48). Applications of these tools may improve interpretabil-
ity of both quality of life scores and the quantitative assesment of how eye diseases may
impact quality of life. Improving interpretability of quality of life scores and accuracy of our
research allows patients and clinicians to better able to make decisions from quality of life
research.
2.1 Measuring Vision-Specific Quality of Life
Using quality of life effectively in the clinic and to make public health decisions requires
that it is measured well. Yet, quality of life can’t be measured directly, relying on question-
naires to approximate scores. In general, these tools ask questions (“items”) related to the
domain (e.g “vision-specific quality of life”) that are then combined systematically to esti-
mate the underlying (“latent”) variable more accurately. Many tools have been developed
to estimate quality of life related to the specific issues that come along with eye diseases
and vision, such as the Activities of Daily Vision Scale and Visual Function Index, but one
of the most commonly used is the National Eye Institute Visual Function Questionnaire
5
(NEI-VFQ-25) (49–51).
In its current form, the NEI-VFQ-25 is a 25-item assessment that measures vision-
specific functioning and well-being (52,53). The original question bank for the NEI-VFQ was
developed using qualitative methods and confirmed with both psychometric and clinically-
relevant criteria. These items were produced using focus groups of participants with a variety
of common eye diseases. They were asked open-ended questions like “How does your vision
problem affect the things you do every day?” “What do you have difficulty doing?” “What
do you do differently than you did in the past?” and “What have you given up doing since
you developed vision problems?” The NEI-VFQ field test team used these focus groups to
create an initial version of the tool with 51 items. After further data collection, the creators
of the NEI-VFQ-51 evaluated the ability of a shorter subset of items to explain the bulk
of the variation in NEI-VFQ-51 scores. Based on these data, the authors chose a subset of
items, dropping 26 of them, to create the NEI-VFQ-25.
The NEI-VFQ-25 has 12 domains that contribute to a larger composite score: general
health, general vision, ocular pain, near vision, distance vision, social functioning, mental
health, role difficulties, dependency, driving, color vision, and peripheral vision (Table 3.1).
The scores of the domains are scaled from 0 to 100 by averaging the items within each
scale and multiplying by 25. The composite score is calculated by taking the mean of the
12 domains (often, the general health domain is omitted, however). Both the development
and the scoring of the NEI-VFQ-25 are based on classical test theory, a school of thought
that assumes a person’s observed obtained score on a test is their true underlying score plus
measurement error (45).
While the NEI-VFQ-25 was carefully developed and validated, some researchers have
raised concerns about the scale (21,54). First, the NEI-VFQ-25, a scale with only 25 items,
has 12 dimensions. Psychometrically and phenomenologically, it is unclear that vision-
specific quality of life has so many facets to it and, if they do truly exist, that such a
6
scale has enough breadth to accurately measure all 12. Instead, the latent factors picked up
by agnostic algorithms used for psychometric development may be detecting other signals,
such as symptom clusters of common eye and vision problems that might hinder quality of
life.
Likewise, there are several potential limitations of the composite score. The most im-
portant is that this score may not be unidimensional (21,54). Unidimensionality is a strong
assumption for some techniques (like item response theory, discussed below) and a weaker as-
sumption for others, such as the classical test theory used to develop the scale (45,46). Even
in the context of classical test theory, multi-dimensional scores are complex; they represent
marginalscoresaveragedacrossdimensions, whichischallengingtointerpretclinically. Asan
alternative, in Chapter 4, I propose two adapted composite scores that represent task-related
quality of life (e.g., being able to read, work, or drive) and well-being-related quality of life
(e.g., being able to leave the home or see people’s reactions during conversations). I discuss
the psychometric properties of these new scales compared to the traditional composite in
Chapter 4, page 47.
2.2 Specifying dimensions with structural equation models
To assess this new way of calculating vision-specific quality of life, we used structural
equation models to compare the traditional composite score, the well-being and task com-
posite scores, and the 12 traditional scales. Classical psychometrics often use a data-driven
approach to dimension reduction, such as exploratory factor analysis (45). Often, however,
researchers have expertise about a domain and how dimensions and items may function
within the latent variable. For these models, we sought to incorporate clinical knowledge
of vision to improve the way the data is reduced. Structural equation models are a flexible
approach to both regression and latent variable analysis that allow one to pre-specify how
7
observed variables relate to latent variables (55). These models commonly have both ob-
served and unobserved components; while estimating latent variables requires some observed
items that contribute to it, latent variables can be used as both dependent and independent
variables. Figure 2.1 shows a basic structural equation model. Here, x1, x2, and x3 are
observed items that contribute to a latent variable, L. L is then used as an independent
variable in a regression model with y as an (observed) outcome.
A structural equation model that estimates a latent variable without an outcome is
called confirmatory factor analysis (55). Instead of the data-driven approach used in ex-
ploratory factor analysis, we are able to specify theorized relationships between variables.
Metrics like goodness-of-fit statistics, comparative fit index, and root mean-squared error of
approximation provide insight into the fit of the latent model, and we can compare nested
models using likelihood ratio tests (56). We can think of the question of dimensions in the
NEI-VFQ-25 as one of nested latent variable models. In Figure 2.2, I show the unidimen-
sionality assumption underlying the traditional composite score: there is a single dimension,
with all 25 items contributing to it. In Figure 2.3, I show the model we propose in Chapter 4,
where the items are divided into two dimensions, task-related quality of life and well-being.
In Figure 2.4, I show the existing 12 subscales of the NEI-VFQ-25. Each of these is nested
in the next and can thus be tested. Confirmatory factor analysis thus allows us to be both
principled in the models we are developing and systematic in the way we compare them.
2.3 Measuring Quality of Life with Item Response Theory
Classical test theory requires several assumptions that may not always be true. The
first is related to the way that scores are calculated: adding Likert-type items (e.g., “strongly
disagree,” “disagree,” “agree,” “strongly agree”) and taking the sum or mean assumes that
the score is best estimated by an equally-spaced, linear combination of the items (45). Each
8
response for a question is treated equally, and each question is equally-weighted in the
composite score. Sometimes this is addressed by differentially weighting items or subscales
(e.g. as done in the FACT-G quality of life scale), but often scales are calculated using equal
weights, which may be an unrealistic assumption (57). In the NEI-VFQ-25, for example,
the item “I stay home most of the time because of my eyesight” is given equal weight to
“Because of your eyesight, how much difficulty do you have picking out and matching your
own clothes?” While both provide information about quality of life, the isolation of not
being able to leave home may be more important to social well-being than matching one’s
clothes.
An alternative way of measuring latent variables is through item response theory, an
approach that I apply and describe in Chapter 5. Item response theory is a set of models that
generalize the Rasch model; the Rasch model is a tool developed in educational assessment
to understand how well items perform at measuring ability across its spectrum and how
well they discriminate between levels of ability (46,58). While there are many ways to fit
these types of models, they share similar properties: items are used to probabilistically
estimate an underlying latent variable, θ (theta, usually estimated as a standard normal
distribution), and each test has a collective ability to accurately estimate θ within a certain
range of θ. Often, researchers fit another parameter, called difficulty, that allows each item
to contribute to θ differently. Each item also has an information function: one item may tell
us about a specific range of θ (for example, a very hard, specific math question tells us a
lot about people with advanced math abilities but little about people with lower abilities),
more broadly across the range of θ (e.g. a question that tells us a little over a wide range of
abilities), or somewhere between.
In Figure 2.5, I present three hypothetical items on a psychometric scale. Each item
gives us different information about θ. Item 1 tells us broadly about people responding the
psychometric scale. Item 2 tells us more specifically about people with lower levels of θ. Item
3 tells us a very limited range about people with higher levels of θ. Ideally, a test will have
9
good coverage for levels of θ within the population one is looking to test, with enough items
that different levels of θ are accurately discriminated from one another. This information
together contributes to the test information function: the best range of θ for which we expect
a tool to function.
A strength of item response models is that they are flexible with respect to the relation-
ship between the model and the scores, but they come with strong assumptions of their own
(21,46,54). Unidimensionality is much more important in item response theory, and scales
that are not unidimensional are not valid. While there are methods for fitting multiple di-
mensions in item response, the usual approach is to remove items until a single dimension
remains. Some type of item response models are more restrictive than others. Rasch models,
for instance, set information functions to be the same shape; only their location across θ can
change. The types of models one chooses to use depends part on measurement philosophy
(some researchers believe Rasch-type models are preferable (21)) and domain knowledge (we
may expect that items do indeed discriminate differently than others based on the domain).
Aswithmuchofstatisticalmodeling, simplermodelscansometimesperformbetterovertime
compared to data-driven approaches, but flexible models combined with expert knowledge
are often superior (59).
Some users of the NEI-VFQ-25 have attempted to improve the tool using item re-
sponse theory to address limitations like multidimensionality and assumptions about item
and response weights, particularly through polytomous Rasch models (46). As discussed in
Chapters 4 and 5, I applied these methods to achieve a better balance between the theory
behind the dimensions discussed above and data-driven decisions for improving models.
10
2.4 Confounding, DAGs, and sensitivity analyses
Many useful tools for improving causal inference have been developed, yet these tools
have not been fully applied in vision epidemiology. A variety of analyses for vision specific
research may benefit from the causal inference toolkit. One simple yet powerful tool is the
structural causal model. Structural causal models are often drawn as a directed acyclic
graphs (DAGs) (47). DAGs may guide decisions in covariate selection, as well as with more
complex issues, like addressing selection bias and dealing with time-varying effects (see the
appendix). These latter benefits are particularly useful in longitudinal data, and I apply
them to the models presented in Chapter 4.
I built each regression model in this dissertation using structural causal models to reflect
thecausalassumptionsbehindeachanalysis, basedondomainknowledgefromtheliterature.
I used DAGs to refine the modeling decision process for building appropriate models out of
available variables using existing knowledge about the relationships between these variables.
Causal DAGs are based on the mathematical relationship between structural causal models
and statistical associations and have the advantage of showing the causal relationships in a
consistent and easily interpretable visualization.
Another way to think about DAGs is as non-parametric structural equation models:
we are explicitly laying out paths between variables, but in the case of a DAG, it doesn’t
matter what form the relationship between two variables takes, only its direction. The rules
underpinning DAGs are consistent whether the relationship is a simple, linear one, or a more
complicated function (48). DAGs allow us to identify what paths in the structural equation
model bias our causal estimates and what sets of variables (called minimal adjustment sets)
are needed to control for to correct for this bias (47,48). Interestingly, identifying a minimal
adjustment set allows us to avoid fitting the entire structural equation model. Because we
don’t have to specify the form of the relationships outside of the minimal adjustment set,
we run a lower risk of biasing our estimate by misspecifying something (inducing residual
11
selection bias or confounding).
A closely related set of tools to DAGs are G-methods (48). G-methods are a collection
of models–Inverse Probability Weighting, G-computation/the parametric G-formula, and G-
estimation–for estimation of counterfactual effects. I use Inverse Probability Weighting and
G-computation with several models in this dissertation. For instance, in Chapter 4, I use
longitudinal data from LALES; while drop-out is normal in population-based studies, it may
be related to visual function and outcomes like quality of life. Inverse Probability Weighting
of models can account for potential selection bias by incorporating knowledge on causes of
drop-out.
2.5 Conclusion
In this dissertation, I evaluate the impact of vision loss on quality of life, an essential
activity for setting clinical and public health priorities. The impact of vision loss on quality
of life is expected to rise as the population ages. More and more, people will need to make
decisions about their vision. Improving our tools and analyses to better communicate what
may happen to quality of life when vision improves or worsens is key to making better clinical
and policy choices at the frontier of a shift in this important public health burden.
12
2.6 References
1. Jones GC, Rovner BW, Crews JE, et al. Effects of depressive symptoms on health
behavior practices among older adults with vision loss. Rehabilitation Psychology.
2009;54(2):164–172.
2. Lamoureux EL, Fenwick E, Moore K, et al. Impact of the Severity of Distance
and Near-Vision Impairment on Depression and Vision-Specific Quality of Life in
Older People Living in Residential Care. Investigative Ophthalmology & Visual Sci-
ence [electronicarticle]. 2009;50(9):4103–4109. (https://iovs.arvojournals.org/article.
aspx?articleid=2185912). (Accessed January 26, 2019)
3. Rovner BW, Ganguli M. Depression and Disability Associated with Impaired Vision:
The MoVIES Project. Journal of the American Geriatrics Society [electronic ar-
ticle]. 1998;46(5):617–619. (http://onlinelibrary.wiley.com/doi/abs/10.1111/j.1532-
5415.1998.tb01080.x). (Accessed January 26, 2019)
4. Zhang X, Bullard KM, Cotch MF, et al. Association Between Depression and Func-
tional Vision Loss in Persons 20 Years of Age or Older in the United States, NHANES
2005-2008. JAMA Ophthalmology [electronic article]. 2013;131(5):573–581. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/1660943). (Accessed Jan-
uary 26, 2019)
5. Crews JE, Campbell VA. Vision impairment and hearing loss among community-
dwelling older Americans: Implications for health and functioning. American Journal
of Public Health. 2004;94(5):823–829.
6. Ivers RQ, Norton R, Cumming RG, et al. Visual impairment and risk of hip fracture.
American Journal of Epidemiology. 2000;152(7):633–639.
7. Lee DJ, Gómez-Marín O, Lam BL, et al. Visual acuity impairment and mortality in
US adults. Archives of Ophthalmology (Chicago, Ill.: 1960). 2002;120(11):1544–1550.
13
8. 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 Ophthalmology [electronic article]. 2016;134(7):802–809. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/2523780). (Accessed Jan-
uary 26, 2019)
9. Choudhury F, Varma R, Klein R, et al. Age-Related Macular Degeneration and
Quality of Life in Latinos: The Los Angeles Latino Eye Study. JAMA Ophthalmol-
ogy [electronic article]. 2016;134(6):683–690. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/2514080). (Accessed March 12, 2018)
10. Globe DR, Wu J, Azen SP, et al. The impact of visual impairment on self-reported vi-
sual functioning in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(6):1141–1149. (http://www.aaojournal.org/article/S0161-
6420(04)00303-3/abstract). (Accessed March 19, 2018)
11. Mazhar K, Varma R, Choudhury F, et al. Severity of Diabetic Retinopathy and
Health-Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology
[electronic article]. 2011;118(4):649–655. (http://www.aaojournal.org/article/S0161-
6420(10)00819-5/abstract). (Accessed March 19, 2018)
12. PatinoCM,VarmaR,AzenSP,etal. TheImpactofChangeinVisualFieldonHealth-
Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2011;118(7):1310–1317. (http://www.aaojournal.org/article/s0161-
6420(10)01344-8/abstract). (Accessed March 19, 2018)
14
13. Varma R, Paz SH, Azen SP, et al. The Los Angeles Latino Eye Study: Design,
methods, and baseline data. Ophthalmology [electronic article]. 2004;111(6):1121–
1131. (http://www.aaojournal.org/article/S0161-6420(04)00291-X/abstract). (Ac-
cessed March 19, 2018)
14. McKean-Cowdin R, Varma R, Hays RD, et al. Longitudinal Changes in Visual Acu-
ity and Health-related Quality of Life: The Los Angeles Latino Eye Study. Ophthal-
mology [electronic article]. 2010;117(10):1900–1907.e1. (http://www.aaojournal.org/
article/S0161-6420(10)00132-6/abstract). (Accessed March 19, 2018)
15. Linder M, Chang TS, Scott IU, et al. Validity of the visual function index (vf-
14) in patients with retinal disease. Archives of Ophthalmology [electronic article].
1999;117(12):1611–1616. (http://dx.doi.org/10.1001/archopht.117.12.1611)
16. Mangione CM, Phillips RS, Lawrence MG, et al. Improved Visual Function
and Attenuation of Declines in Health-Related Quality of Life After Cataract
Extraction. Archives of Ophthalmology [electronic article]. 1994;112(11):1419–
1425. (https://jamanetwork.com/journals/jamaophthalmology/fullarticle/640890).
(Accessed March 19, 2018)
17. Health- and vision-related quality of life among patients with choroidal neovascular-
ization secondary to age-related macular degeneration at enrollment in randomized
trials of submacular surgery: SST report no. 4. American Journal of Ophthalmol-
ogy [electronic article]. 2004;138(1):91–108. (http://www.sciencedirect.com/science/
article/pii/S0002939404001370)
18. Parrish RK, Gedde SJ, Scott IU, et al. Visual function and quality of life
among patients with glaucoma. Archives of Ophthalmology (Chicago, Ill.: 1960).
1997;115(11):1447–1455.
15
19. Scott IU, Smiddy WE, Schiffman J, et al. Quality of life of low-vision patients and the
impact of low-vision services. American Journal of Ophthalmology [electronic article].
128(1):54–62. (http://dx.doi.org/10.1016/S0002-9394(99)00108-7). (Accessed March
20, 2018)
20. Healthy people 2020: vision. Healthy People 2020. (https://www.healthypeople.gov/
2020/topics-objectives/topic/vision)
21. Lamoureux E, Pesudovs K. Vision-Specific Quality-of-Life Research: A Need
to Improve the Quality. American Journal of Ophthalmology [electronic arti-
cle]. 2011;151(2):195–197.e2. (https://doi.org/10.1016/j.ajo.2010.09.020). (Accessed
March 18, 2020)
22. Coleman AL, Yu F, Keeler E, et al. Treatment of Uncorrected Refractive Error
Improves Vision-Specific Quality of Life. Journal of the American Geriatrics Society
[electronic article]. 2006;54(6):883–890. (http://onlinelibrary.wiley.com/doi/abs/10.
1111/j.1532-5415.2006.00817.x). (Accessed March 23, 2020)
23. BourneRR,StevensGA,WhiteRA,etal. Causesofvisionlossworldwide, 1990–2010:
A systematic analysis. The lancet global health. 2013;1(6):e339–e349.
24. Varma R, Chung J, Foong AWP, et al. Four-Year Incidence and Progression of Vi-
sual Impairment in Latinos: The Los Angeles Latino Eye Study (LALES). Amer-
ican journal of ophthalmology [electronic article]. 2010;149(5):713–727. (https:
//www.ncbi.nlm.nih.gov/pmc/articles/PMC2919324/). (Accessed January 26, 2019)
25. Varma R, Kim JS, Burkemper BS, et al. Prevalence and causes of visual impairment
and blindness in chinese american adults: The chinese american eye study. Jama
Ophthalmology. 2016;134(7):785–793.
16
26. Munoz B, West SK, Rubin GS, et al. Causes of blindness and visual impairment
in a population of older americans: The salisbury eye evaluation study. Archives of
ophthalmology. 2000;118(6):819–825.
27. Tang Y, Wang X, Wang J, et al. Prevalence and causes of visual impairment in a
chinese adult population: The taizhou eye study. Ophthalmology. 2015;122(7):1480–
1488.
28. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindness in a rural chinese adult population: The handan eye study. Ophthalmology.
2008;115(11):1965–1972.
29. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in beijing: The beijing eye study. Ophthalmology. 2006;113(7):1134–e1.
30. ZhaoJ,XuX,EllweinLB,etal. Causesofvisualimpairmentandblindnessinthe2006
and 2014 nine-province surveys in rural china. American Journal of Ophthalmology.
2019;197:80–87.
31. AtteboK,MitchellP,SmithW.Visualacuityandthecausesofvisuallossinaustralia:
The blue mountains eye study. Ophthalmology. 1996;103(3):357–364.
32. Hyman L, Wu S-Y, Connell AM, et al. Prevalence and causes of visual impairment
in the barbados eye study. Ophthalmology. 2001;108(10):1751–1756.
33. Klaver CC, Wolfs RC, Vingerling JR, et al. Age-specific prevalence and causes
of blindness and visual impairment in an older population: The rotterdam study.
Archives of ophthalmology. 1998;116(5):653–658.
34. Rodriguez J, Sanchez R, Munoz B, et al. Causes of blindness and visual impairment
in a population-based sample of US hispanics. Ophthalmology. 2002;109(4):737–743.
17
35. CongdonN,O’ColmainB,KlaverC,etal. Causesandprevalenceofvisualimpairment
among adults in the united states. Archives of Ophthalmology (Chicago, Ill.: 1960).
2004;122(4):477–485.
36. VanNewkirk MR, Weih L, McCarty CA, et al. Cause-specific prevalence of bilateral
visual impairment in victoria, australia: The visual impairment project. Ophthalmol-
ogy. 2001;108(5):960–967.
37. 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. Investigative Ophthalmology & Visual Science. 2002;43(11):3393–
3398.
38. Unzueta M, Globe D, Wu J, et al. Compliance with recommendations for follow-up
care in latinos: The los angeles latino eye study. Ethnicity & disease. 2004;14:285–91.
39. Varma R, Mohanty SA, Deneen J, et al. Burden and predictors of undetected eye
disease in mexican-americans. Medical care. 2008;46(5):497–506.
40. Patty LE, Morales LS, Lai M, et al. Self-reported utilization of eye care among latinos
in the los angeles latino eye study (LALES). Investigative Ophthalmology & Visual
Science. 2009;50(13):2502–2502.
41. Varma R, Torres M, McKean-Cowdin R, et al. Prevalence and risk factors for re-
fractive error in adult chinese americans: The chinese american eye study. American
Journal of Ophthalmology. 2017;175:201–212.
42. Jiang X, Varma R, Torres M, et al. Self-reported use of eye care among adult chinese
americans in the chinese american eye study. American journal of ophthalmology.
2017;176:183–193.
18
43. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by hispanics: Findings from HHANES 1982-84. American Journal of
Public Health. 1990;80(Suppl):11–19.
44. Zheng Y, Lamoureux EL, Ikram MK, et al. Impact of migration and acculturation
on prevalence of type 2 diabetes and related eye complications in indians living in a
newly urbanised society. PloS One. 2012;7(4):e34829.
45. Streiner DL, Cairney J, Norman GR. Health measurement scales: A practical guide
to their development and use. Oxford Univ. Press; 2015.
46. Ayala RJde. The theory and practice of item response theory. Guildford Press; 2009.
47. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Medical
Research Methodology [electronic article]. 2008;8:70. (https://www.ncbi.nlm.nih.gov/
pmc/articles/PMC2601045/). (Accessed November 21, 2018)
48. Hernán MA, Robins JM. Causal Inference. Boca Raton: Chapman Hall/CRC; 2019.
49. Steinberg EP, Tielsch JM, Schein OD, et al. The VF-14: An Index of Functional
ImpairmentinPatientsWithCataract. Archives of Ophthalmology [electronicarticle].
1994;112(5):630–638. (https://doi.org/10.1001/archopht.1994.01090170074026)
50. Mangione CM, Phillips RS, Seddon JM, et al. Development of the ’Activities of Daily
Vision Scale’. A measure of visual functional status. Medical Care. 1992;30(12):1111–
1126.
51. Margolis MK, Coyne K, Kennedy-Martin T, et al. Vision-specific instruments for the
assessment of health-related quality of life and visual functioning: A literature review.
PharmacoEconomics. 2002;20(12):791–812.
19
52. Mangione CM, Lee PP, Gutierrez PR, et al. Development of the 25-list-item
National Eye Institute Visual Function Questionnaire. Archives of Ophthalmol-
ogy [electronic article]. 2001;119(7):1050–1058. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/267134). (Accessed March 19, 2018)
53. Mangione CM, Lee PP, Pitts J, et al. Psychometric Properties of the National
Eye Institute Visual Function Questionnaire (NEI-VFQ). Archives of Ophthalmology
[electronic article]. 1998;116(11):1496–1504. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/264225). (Accessed March 19, 2018)
54. Pesudovs K, Gothwal VK, Wright T, et al. Remediating serious flaws in the Na-
tional Eye Institute Visual Function Questionnaire. Journal of Cataract & Refrac-
tive Surgery [electronicarticle]. 2010;36(5):718–732. (http://linkinghub.elsevier.com/
retrieve/pii/S0886335010002531). (Accessed March 12, 2018)
55. Kline RB. Principles and practice of structural equation modeling. The Guilford
Press; 2016.
56. Bentler PM. On tests and indices for evaluating structural models. Personal-
ity and Individual Differences [electronic article]. 2007;42(5):825–829. (http://
www.sciencedirect.com/science/article/pii/S0191886906003898). (Accessed March
18, 2020)
57. Brucker PS, Yost K, Cashy J, et al. General population and cancer pa-
tient norms for the functional assessment of cancer therapy-general (FACT-
g). Evaluation & the Health Professions [electronic article]. 2005;28(2):192–
211. (%20%0A%20%20%20%20%20%20%20%20https://doi.org/10.1177/
0163278705275341%0A%20%20%20%20%0A)
20
58. Wright BD. Solving MEASUREMENT PROBLEMS WITH THE Rasch MODEL.
Journal of Educational Measurement [electronic article]. 1977;14(2):97–116. (https:
//onlinelibrary.wiley.com/doi/abs/10.1111/j.1745-3984.1977.tb00031.x)
59. Harrell F. Regression modeling strategies: With applications to linear models, logistic
and ordinal regression, and survival analysis. springer; 2016.
60. Medeiros FA, Gracitelli CPB, Boer ER, et al. Longitudinal Changes in Quality of
Life and Rates of Progressive Visual Field Loss in Glaucoma Patients. Ophthalmol-
ogy [electronic article]. 122(2):293–301. (http://dx.doi.org/10.1016/j.ophtha.2014.08.
014). (Accessed March 20, 2018)
61. Association WM. World Medical Association Declaration of Helsinki: Ethical Prin-
ciples for Medical Research Involving Human Subjects. JAMA [electronic article].
2013;310(20):2191–2194. (https://doi.org/10.1001/jama.2013.281053)
62. GlobeDR,Schoua-GlusbergA,PazS,etal. Usingfocusgroupstodevelopaculturally
sensitive methodology for epidemiological surveys in a Latino population: Findings
from the Los Angeles Latino Eye Study (LALES). Ethnicity & disease [electronic
article]. 2002;12(2):259–266. (http://europepmc.org/abstract/med/12019936). (Ac-
cessed March 19, 2018)
63. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by Hispanics: Findings from HHANES 1982-84. American Journal
of Public Health [electronic article]. 1990;80(Suppl):11–19. (https://doi.org/10.2105/
AJPH.80.Suppl.11). (Accessed March 20, 2018)
64. Brody BL, Gamst AC, Williams RA, et al. Depression, visual acuity, comorbidity,
and disability associated with age-related macular degeneration. Ophthalmology [elec-
tronicarticle]. 108(10):1893–1900. (http://dx.doi.org/10.1016/S0161-6420(01)00754-
0). (Accessed March 20, 2018)
21
65. Globe DR, Varma R, Torres M, et al. Self-reported comorbidities and visual function
inapopulation-basedstudy: Thelosangeleslatinoeyestudy. Archives of Ophthalmol-
ogy [electronic article]. 2005;123(6):815–821. (http://dx.doi.org/10.1001/archopht.
123.6.815)
66. Linn Bernard S., Linn Margaret W., Gurel Lee. Cumulative illness rating scale.
Journal of the American Geriatrics Society [electronic article]. 2015;16(5):622–626.
(https://onlinelibrary.wiley.com/doi/full/10.1111/j.1532-5415.1968.tb02103.x). (Ac-
cessed March 19, 2018)
67. Azen SP, Varma R, Preston-Martin S, et al. Binocular Visual Acuity Sum-
mation and Inhibition in an Ocular Epidemiological Study: The Los Angeles
Latino Eye Study. Investigative Ophthalmology & Visual Science [electronic ar-
ticle]. 2002;43(6):1742–1748. (http://iovs.arvojournals.org/article.aspx?articleid=
2123581). (Accessed March 19, 2018)
68. Varma R, Fraser-Bell S, Tan S, et al. Prevalence of age-related macular degen-
eration in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(7):1288–1297. (http://www.aaojournal.org/article/S0161-
6420(04)00313-6/abstract). (Accessed March 19, 2018)
69. Varma R, Torres M, Peña F, et al. Prevalence of diabetic retinopathy in adult
Latinos: The Los Angeles Latino Eye Study. Ophthalmology [electronic article].
2004;111(7):1298–1306. (http://www.aaojournal.org/article/S0161-6420(04)00336-
7/abstract). (Accessed March 19, 2018)
22
70. Ware JE, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: Construction
ofScalesandPreliminaryTestsofReliabilityandValidity. Medical Care [electronicar-
ticle]. 1996;34(3):220. (https://journals.lww.com/lww-medicalcare/Abstract/1996/
03000/A_12_Item_Short_Form_Health_Survey__Construction.3.aspx). (Ac-
cessed March 19, 2018)
71. Ware J, A. Kosinski M, D. Keller S. SF-12: How to Score the SF-12 Physical and
Mental Health Summary Scales. 1998;
72. Rosenthal R. Parametric measures of effect size. In: The handbook of research syn-
thesis. New York, NY, US: Russell Sage Foundation; 1994:231–244.
73. Cohen J. A power primer. Psychological bulletin [electronic article]. 1992;112(1):155–
159. (http://libproxy.usc.edu/login?url=https://search-proquest-com.libproxy2.usc.
edu/docview/614317877?accountid=14749)
74. Hays RD, Brodsky M, Johnston MF, et al. Evaluating the Statistical Significance
of Health-Related Quality-Of-Life Change in Individual Patients. Evaluation & the
Health Professions [electronic article]. 2005;28(2):160–171. (https://doi.org/10.1177/
0163278705275339). (Accessed March 19, 2018)
75. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epi-
demiology (Cambridge, Mass.). 1999;10(1):37–48.
76. Greenland S. Quantifying biases in causal models: Classical confounding vs collider-
stratification bias. Epidemiology (Cambridge, Mass.). 2003;14(3):300–306.
77. Buuren S van. Flexible imputation of missing data. CRC Press; 2018.
23
78. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained
equations in r. Journal of Statistical Software [electronic article]. 2011;45(3):1–67.
(https://www.jstatsoft.org/v45/i03/)
79. R Core Team. R: A language and environment for statistical computing. Vienna,
Austria: RFoundationforStatisticalComputing; 2017.(https://www.R-project.org/)
80. Rosseel Y. lavaan: An R package for structural equation modeling. Journal of Sta-
tistical Software [electronic article]. 2012;48(2):1–36. (http://www.jstatsoft.org/v48/
i02/)
81. Epskamp S, Simon Stuber with contributions from. semPlot: Path diagrams and
visual analysis of various SEM packages’ output. 2017.(https://CRAN.R-project.
org/package=semPlot)
82. Harrell Jr FE. Rms: Regression modeling strategies. 2018.(https://CRAN.R-project.
org/package=rms)
83. Textor J, van der Zander B. Dagitty: Graphical analysis of structural causal models.
2016.
84. Barrett M. Ggdag: Analyze and create elegant directed acyclic graphs. 2018.
85. Wickham H. ggplot2: Elegant graphics for data analysis. Springer-Verlag New York;
2016.(http://ggplot2.org)
86. Alqudah A, Mansberger SL, Gardiner SK, et al. Vision-related quality of life in glau-
coma suspect or early glaucoma patients. Journal of glaucoma [electronic article].
2016;25(8):629–633. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4975427/).
(Accessed March 12, 2018)
24
87. Fenwick EK, Man RK, Ong P, et al. Association of changes in visual acuity
with vision-specific functioning in the singapore malay eye study. JAMA Oph-
thalmology [electronic article]. 2016;134(11):1299–1305. (http://dx.doi.org/10.1001/
jamaophthalmol.2016.3692)
88. Responsiveness of the National Eye Institute Visual Function Questionnaire
to Changes in Visual Acuity. Archives of ophthalmology [electronic article].
2003;121(4):531–539. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1434793/).
(Accessed March 19, 2018)
89. Lindblad AS, Clemons TE. Responsiveness of the National Eye Institute Visual Func-
tion Questionnaire to progression to advanced age-related macular degeneration, vi-
sion loss, and lens opacity: AREDS Report no. 14. Archives of Ophthalmology
(Chicago, Ill.: 1960). 2005;123(9):1207–1214.
90. Hays RD, Morales LS, Reise SP. Item Response Theory and Health Outcomes Mea-
surementin the21stCentury. Medical care [electronicarticle]. 2000;38(9Suppl):II28–
II42. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1815384/). (Accessed
March 19, 2018)
91. Wittenborn J, Rein D. The economic burden of vision loss and eye disorders in the
United States. University of Chicago; 2013.(https://www.preventblindness.org/sites/
default/files/national/documents/Economic%20Burden%20of%20Vision%20Final%
20Report_130611.pdf)
92. Wang X, Lamoureux E, Zheng Y, et al. Health Burden Associated with Visual Im-
pairment in Singapore: The Singapore Epidemiology of Eye Disease Study. Ophthal-
mology [electronic article]. 2014;121(9):1837–1842. (http://www.sciencedirect.com/
science/article/pii/S0161642014002413). (Accessed March 23, 2020)
25
93. Tsai S-Y, Chi L-Y, Cheng C-Y, et al. The Impact of Visual Impairment and Use of
Eye Services on Health-Related Quality of Life among the Elderly in Taiwan: The
Shihpai Eye Study. Quality of Life Research [electronic article]. 2004;13(8):1415–
1424. (https://doi.org/10.1023/B:QURE.0000040791.87602.fe). (Accessed March 23,
2020)
94. Vision impairment and major eye diseases reduce vision-specific emotional well-being
in a Chinese population - ProQuest. (https://search-proquest-com.libproxy1.usc.
edu/docview/1894846003?rfr_id=info%3Axri%2Fsid%3Aprimo). (Accessed March
23, 2020)
95. Varma R, Kim JS, Burkemper BS, et al. Prevalence and Causes of Visual Impairment
andBlindnessinChineseAmericanAdults: TheChineseAmericanEyeStudy. JAMA
Ophthalmology [electronic article]. 2016;134(7):785–793. (https://jamanetwork-
com.libproxy1.usc.edu/journals/jamaophthalmology/fullarticle/2523778). (Accessed
March 21, 2020)
96. Hsu W-M, Cheng C-Y, Liu J-H, et al. Prevalence and causes of visual impairment
in an elderly Chinese population in Taiwan: The Shihpai Eye Study. Ophthalmology.
2004;111(1):62–69.
97. Huang S, Zheng Y, Foster PJ, et al. Prevalence and causes of visual impairment in
Chinese adults in urban southern China. Archives of Ophthalmology (Chicago, Ill.:
1960). 2009;127(10):1362–1367.
98. Saw S-M, Foster PJ, Gazzard G, et al. Causes of blindness, low vision, and
questionnaire-assessed poor visual function in Singaporean Chinese adults: The Tan-
jong Pagar Survey. Ophthalmology. 2004;111(6):1161–1168.
26
99. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindnessinaruralchineseadultpopulation: TheHandanEyeStudy. Ophthalmology.
2008;115(11):1965–1972.
100. Zhao J, Ellwein LB, Cui H, et al. Prevalence of vision impairment in older adults in
rural China: The China Nine-Province Survey. Ophthalmology. 2010;117(3):409–416,
416.e1.
101. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in Beijing: The Beijing Eye Study. Ophthalmology. 2006;113(7):1134.e1–
11.
102. Varma R, Hsu C, Wang D, et al. The Chinese American Eye Study: Design and
Methods. Ophthalmic Epidemiology [electronic article]. 2013;20(6):335–347. (http:
//www.tandfonline.com/doi/full/10.3109/09286586.2013.823505). (Accessed March
21, 2020)
103. SuinnRM,AhunaC,KhooG.TheSuinn-LewAsianSelf-IdentityAcculturationScale:
Concurrent and factorial validation. Educational and Psychological Measurement.
1992;52(4):1041–1046.
104. Burkemper B, Torres M, Jiang X, et al. Factors Associated with Visual Impairment in
Chinese American Adults: The Chinese American Eye Study. Ophthalmic Epidemiol-
ogy [electronic article]. 2019;26(5):329–335. (https://doi.org/10.1080/09286586.2019.
1622737). (Accessed March 22, 2020)
105. Linden WJ van der. Handbook of item response theory. Chapman & Hall/CRC;
2019.
106. Wood SN. Generalized additive models: An introduction with r. CRC Press/Taylor
& Francis Group; 2017.
27
107. Rizopoulos D. Ltm: An r package for latent variable modelling and item response
theory analyses. Journal of Statistical Software [electronic article]. 2006;17(5):1–25.
(http://www.jstatsoft.org/v17/i05/)
108. Fenwick EK, Ong PG, Sabanayagam C, et al. Assessment of the psychometric proper-
ties of the Chinese Impact of Vision Impairment questionnaire in a population-based
study: FindingsfromtheSingaporeChineseEyeStudy. Qualityof LifeResearch [elec-
tronic article]. 2016;25(4):871–880. (https://doi.org/10.1007/s11136-015-1141-1).
(Accessed March 23, 2020)
109. Rees G, Tee HW, Marella M, et al. Vision-specific distress and depressive symptoms
in people with vision impairment. Investigative Ophthalmology & Visual Science.
2010;51(6):2891–2896.
110. Morales LS, Varma R, Paz SH, et al. Self-reported use of eye care among Latinos:
The Los Angeles Latino Eye Study. Ophthalmology. 2010;117(2):207–215.e1.
111. Chang W, Cheng J, Allaire J, et al. Shiny: Web application framework for r.
2020.(https://CRAN.R-project.org/package=shiny)
28
2.7 Tables and Figures
Table 2.1: Items in the 25-Item National Eye Institute Visual Function Questionnaire
Scale Item
General Health 5-Level overall health rating
General Vision 5-Level overall vision rating
Ocular Pain How much pain or discomfort have you had in and
around your eyes (for example, burning, itching, or
aching)?
How much does pain or discomfort in or around
your eyes, for example, burning, itching, or aching,
keep you from doing what you’d like to be doing?
Near Activities How much difficulty do you have reading ordinary
print in newspapers?
How much difficulty do you have doing work or
hobbies that require you to see well up close, such
as cooking, sewing, fixing things around the house,
or using hand tools
Because of your eyesight, how much difficulty do
you have finding something on a crowded shelf?
Distance Activities How much difficulty do you have reading street
signs or the names of stores?
Because of your eyesight, how much difficulty do
you have going down steps, stairs, or curbs in dim
light or at night?
Because of your eyesight, how much difficulty do
you have going out to see movies, plays, or sports
events?
29
Scale Item
Social Functioning Because of your eyesight, how much difficulty do
you have seeing how people react to things you say?
Because of your eyesight, how much difficulty do
you have visiting with people in their homes, at
parties, or in restaurants?
Mental Health How much of the time do you worry about your
eyesight?
I feel frustrated a lot of the time because of my
eyesight.
I have much less control over what I do, because of
my eyesight.
I worry about doing things that will embarrass
myself or others, because of my eyesight
Role Difficulties Do you accomplish less than you would like because
of your vision?
Are you limited in how long you can work or do
other activities because of your vision?
Dependency I stay home most of the time because of my
eyesight.
Because of my eyesight, I have to rely too much on
what other people tell me.
I need a lot of help from others because of my
eyesight
Driving How much difficulty do you have driving during the
daytime in familiar places?
How much difficulty do you have driving at night?
30
Scale Item
How much difficulty do you have driving in difficult
conditions, such as in bad weather, during rush
hour, on the freeway, or in city traffic?
Color Vision Because of your eyesight, how much difficulty do you
have picking out and matching your own clothes?
Peripheral Vision Because of your eyesight, how much difficulty do
you have noticing objects off to the side while you
are walking along?
31
L x1
x2
x3
y
Figure 2.1: A Basic Structural Equation Model. x1, x2, and x3 represent observed items that contribute to
a latent variable, L. L is then used as an independent variable in a regression model with y as an (observed)
outcome.
32
cv1
dp1
dp2
dp3
dr1
dr2
dv1
dv2
dv3
gh1
gv1
mh1
mh2
mh3
mh4
nv1
nv2
nv3
op1
op2
pv1
rf1
rf2
sf1
sf2
full
Figure 2.2: Confirmatory Factor Analysis of NEI-VFQ-25: Composite Score. In the traditional composite
score, there is a single latent variable, Vision-Specific Quality of Life (“full”). This latent variable is created
from 25 observed variables related to 12 domains: general health (“gh”), general vision (“gv”), ocular
pain (“op”), near vision (“nv”), distance vision (“dv”), social functioning (“sf), mental health (”mh“), role
difficulties (”rf“), dependency (”dp“), driving (”dr“), color vision (”cv“), and peripheral vision (”pv").
33
cv1
dp1
dp2
dp3
dr1
dr2
dv1
dv2
dv3
gh1
gv1
mh1
mh2
mh3
mh4
nv1
nv2
nv3
op1
op2
pv1
rf1
rf2
sf1
sf2
task
wb
Figure 2.3: Confirmatory Factor Analysis of NEI-VFQ-25: Task and Well-being. In the proposed scoring
approach, there are two latent variables, Task-Related Vision-Specific Quality of Life (“task”) and Well-
Being-Related Vision-Specific Quality of Life (“wb”). The task latent variable is created from 12 observed
variables related to 6 domains: near vision (“nv”), distance vision (“dv”), role difficulties (“rf”), color vision
(“cv”), driving (“dr”), and peripheral vision (“pv”). The well-being latent variable is created from 13
observed variables related to 6 domains: general health (“gh”), general vision (“gv”), ocular pain (“op”),
social functioning (“sf), mental health (”mh“), and dependency (”dp").
34
cv1
dp1
dp2
dp3
dr1
dr2
dv1
dv2
dv3
gh1
gv1
mh1
mh2
mh3
mh4
nv1 nv2
nv3
op1
op2
pv1
rf1
rf2
sf1
sf2
cv
dp
dr
dv
gh
gv
mh
nv
op
pv
rf
sf
Figure 2.4: Confirmatory Factor Analysis of NEI-VFQ-25: All Subscales. In the traditional sub-domain
scores, there are 12 latent variables created from 25 observed items: general health (“gh”), general vision
(“gv”), ocular pain (“op”), near vision (“nv”), distance vision (“dv”), social functioning (“sf), mental health
(”mh“), role difficulties (”rf“), dependency (”dp“), driving (”dr“), color vision (”cv“), and peripheral vision
(”pv"). Arrows between latent variables removed for clarity.
35
1
2
3
−4 −2 0 2 4
latent variable (theta)
information
Figure 2.5: A Theoretical Item Response Model. In three simulated items for an item response model, each
item provides different information about θ. Item 1 provides information broadly about people answering
questions with a wide range of θ. Item 2 provides information more specifically about people with lower
levels of θ. Item 3 provides information for a very limited range about people with higher levels of θ. Items
were simulated using a normal distribution.
36
CHAPTER 3
THE IMPACT OF 8-YEAR CHANGE IN VISUAL ACUITY ON NEW NEI-VFQ-25
COMPOSITES IN THE LOS ANGELES LATINO EYE STUDY (LALES)
3.1 Abstract
We sought to assess the validity of two new composite scales for the NEI-VFQ-25 for
vision-related tasks and well-being and estimate the impact of a change of visual acuity
(VA) of 2 and 4 lines on quality of life (QOL). 2234 Latino adults in a population-based,
longitudinal cohort study in La Puente, CA, underwent complete ophthalmologic exams
including measurements of presenting and best-corrected distance VA. Examinations were
completed at baseline (2000-2003) and 8 years of follow-up (2008-2013). Vision-related QOL
was assessed using the NEI-VFQ-25. Two composite scores were adapted task and well-
being-related questions using confirmatory factor analysis (CFA). Changes in QOL and VA
were assessed between baseline and 8-year follow-up using covariate-adjusted OLS models,
which were then used to predict mean change in VS-QOL at 2 and 4 lines of change in VA
using the parametric G-formula. Compared to the traditional composite score for the NEI-
VFQ-25, introducing separate composites for task and well-being statistically significantly
improved the CFA model (p<.001). For the traditional, Task, and Well-Being composites,
increasing visual acuity by 2 lines improved quality of life by 2.49 points (95% CI 0.95,
3.68), 2.56 points (95% CI 1.01, 3.99), and 2.42 points (95% CI 0.86, 3.62), respectively.
For the traditional, Task, and Well-Being composites, losing 2 lines of visual acuity reduced
quality of life by -1.43 points (95% CI -2.72, -0.27), -1.73 points (95% CI -3.05, -0.46), and
-1.11 points (95% CI -2.44, 0.10), respectively. Using 2 separate composite scores focused on
vision-related daily tasks and well-being may improve interpretability of the effect changing
vision.
37
Vision contributes to the ability of people to process information around them and to
engage in everyday tasks and social situations. People with visual impairment (VI) often
have these aspects of their lives interrupted. While the causes of vision loss vary in their
severity, this can reduce the overall quality of one’s life, sometimes leading to depression and
isolation for those with worse vision loss (1–4); vision loss is also associated with chronic
healthconditionsandmortality(5–7). In2015, theprevalenceofvisualimpairmentintheUS
was 3.22 million (8). As the population ages, projections suggest this prevalence may double
by 2050. Given that US Latinos have higher rates of incidence of visual impairment than
white Americans–as high as 7 times more likely in people under 55 years old–understanding
long-term impacts on quality of life and how it may differ from other ethnic groups is es-
sential (24). In our previous analyses of data from the Los Angeles Latino Eye Study, we
found that visual acuity (VA) affects quality of life, as did other specific eye conditions and
diseases, such as visual field loss, glaucoma, diabetic retinopathy, and age-related macular
degeneration (9–13). We also previously reported on the effect of change in visual acuity
(VA) on vision-specific quality of life in LALES at the 4-year follow-up examination. We
found that a clinically meaningful change in VA (a 2-line change or greater) was associated
with an approximate 5-point change in vision-specific quality of life (VS-QOL) using a single
summary composite measure (14).
The National Eye Institute Visual Function Questionnaire (NEI-VFQ-25) is a 25 item
assessment tool that measures vision-specific functioning and well-being (52,53). Previous
clinicalandpopulation-basedstudieshavefoundthattheNEI-VFQ-25issensitivetochanges
inVS-QOLinthosewithvisionloss(15–19). InLALES,a2-lineorgreaterdifferenceincross-
sectional VA was associated with a 5-point difference in baseline NEI-VFQ-25 scores, and we
found a similar magnitude of association using longitudinal date from baseline to the 4-year
follow-up clinical examination (10,14). Analyzing this association at 8-year follow-up allows
ustoexaminetheconsistencyoftheseestimatesofchangeinVS-QOLbecausemoreextensive
changes in vision may have occurred among participants. Additionally, these findings are
38
based on classical scoring of the NEI-VFQ-25, of which several psychometric limitations
have been discussed (54,60), including that the traditional single-summary (composite) score
may not be unidimensional. Multidimensionality may make these VS-QOL scores harder to
interpret for patients and clinicians.
In this analysis, we address issues of NEI-VFQ composite multidimensionality and in-
terpretability by estimating scores for task-based visual functioning and social well-being
separately (54,60), demonstrating the improvements in psychometric properties of the 2
summary scores over the original composite in LALES. We also discuss the interpretation of
these scores using population-based, longitudinal changes in VA and quantitative association
with VS-QOL. We revisit our analysis of change in VA on VS-QOL with 4 additional years of
data (8-year change) to evaluate how the association varies by severity and direction of VA
change over nearly a decade of follow-up. We present the psychometric properties of the new
composite scores (vision task and well-being) compared to the traditional single composite
score, as well as to the other subscales of the NEI-VFQ-25. We examined the magnitude
of change in VS-QOL (overall and by the direction of vision change) and the consistency of
estimates from baseline, 4-year, and 8-year data.
3.2 Methods
These data were collected as part of the Los Angeles Latino Eye Study (LALES), a
population-based study of eye disease in Latino adults living in La Puente, California. We
described the study design and data collection for LALES in a previous report (13). Briefly,
eligibility for LALES was based on residence in 6 census tracts in La Puente, and eligible par-
ticipantswereself-describedLatinomenandwomenage40orolder. Atbaseline, participants
were provided a written and verbal explanation of the study and were invited to participate
in home and clinical interviews and a clinical examination. Baseline assessment took place
39
between February 2000 and May 2003. Follow-up interviews and examinations occurred
after 4 years and 8 years, from January 2004 through May 2008 and August 2010 through
November 2013, respectively. Participants provided informed consent before completing any
in-home and clinical assessments. All procedures in LALES followed the guidelines for work-
ing with human subjects in research established by the Declaration of Helsinki (61). The
institutional review boards for Los Angles County and the University of Southern California
approved LALES.
3.2.1 Sociodemographic and Clinical Data
Eligible and consenting members of the community participated in home interviews that
included demographic information, history of medical conditions and eye diseases, medical
and vision insurance status, access to health care, and degree of acculturation (62). The
definitions for these variables were adapted from the Hispanic Health and Nutrition Exam-
ination Survey (63). We calculated a composite score, based on a validated approach, for
comorbidities summarizing the history of 12 medical conditions: diabetes mellitus, arthritis,
stroke or brain hemorrhage, high blood pressure, angina, heart attack, heart failure, asthma,
skin cancer, other types of cancers, back problems, and deafness or hearing problems (64–
66). Participants also responded to the Cuellar Acculturation Scale short form, a scale which
measures acculturation from 1 (lowest level of acculturation) to 5 (highest) (63).
3.2.2 Visual Acuity
Visual acuity testing procedures in LALES have been described previously (67–69).
Briefly, the study ophthalmologist measured presenting distance VA, including any present-
ing correction, for each LALES participant at 4 meters using modified Early Treatment
40
Diabetic Retinopathy Study distance charts trans-illuminated with the chart illuminator
(Precision Vision, La Salle, IL). For each participant, presenting VA was the total number of
lines read, scored as the logarithm of the minimum angle of resolution (logMAR). Clinicians
used automated refraction (Humphrey Autorefractor model 509, Carl Zeiss Meditec, Dublin,
CA) for participants that read 55 letters at 4 meters in either eye. Those who subsequently
read 55 letters while viewing through the prescription determined from the autorefractor
also underwent subjective refraction using a standard protocol followed by a measurement of
best-corrected VA. Participants who identified 3 of 5 letters correctly progressed to sequen-
tially smaller lines (logMAR levels) from the top of the chart down, until the participant
identified 2 or fewer letters. Best-corrected VA is the best VA measured at a distance during
subjective refraction based on the person’s better-seeing eye. If a participant was unable
to read 20 letters (20/100 Snellen) at 4 meters, clinicians instead performed VA measure-
ment at 1 meter. The study ophthalmologist additionally assigned a cause of incident visual
impairment at 4 years base on ophthalmic review.
3.2.3 Health-Related and Vision-Specific Quality of Life
3.2.3.1 Medical Outcomes Study 12-Item Short Form Health Survey
To measure general health-related quality of life, we used the Medical Outcomes Study
12-Item Short Form Health Survey version 1 (SF-12) to calculate the Physical Component
Summary (PCS) and the Mental Component Summary (MCS) scores, where higher scores
indicate a higher quality of life (70,71). These scores are calculated using a standard US
norm-based T-Score, where the mean is to 50, and the SD is 10, calibrated using the general
US population.
41
3.2.3.2 The National Eye Institute Visual Function Questionnaire
We used the NEI-VFQ-25 to assess vision-specific quality of life (52,53). The NEI-
VFQ-25 measures visual functioning related to general health and well-being, as well as
areas more specific to daily functioning related to vision. The survey yields 12 scales re-
lated to visual functioning: general health, general vision, near and distance vision, ocular
pain, vision-related social function, vision-related role function, vision-related mental health,
vision-related dependency, driving difficulties, color vision, and peripheral vision. The scales
are composed of different numbers of items, from 1 to 4 questions. We used the standard
scoring method (52). Calculated scales can range from 0 to 100, where higher scores indicate
better visual functioning and well-being. The traditional composite score is calculated by
averaging 11 of the 12 scale scores (excluding the general health rating question). The new
composite scores, adapted from the work of Pesudovs et al. (54), were averages of subsets of
the 11 scales (see appendix). We also used two new scoring approaches to create measures
of task-related quality of life and well-being. The task measure was a composite average of
the near vision, distance vision, driving, color vision, peripheral vision, and vision-related
role function scales, while the well-being measure was a composite average of general vision,
vision-relateddependency, vision-relatedmentalhealth, ocularpain, andvision-relatedsocial
function scales. The NEI-VFQ-25 instrument was developed using classical test theory, the
scoring approach that we used in previous publications; despite recent calls for appliation of
item response theory for VS-QOL, we retained this classical approach for better comparison
to baseline and 4-year reports. We evaluate the use of these scales using item response theory
in a separate publication. Additionally, while our scales are largely consistent with that of
Pesudovs et al., we have two primary differences: 1) Pesudovs et al. dropped pain-related
items from the scale, but we instead chose to retain pain-related items due to the clinical
importance of this measure and consistency with previous analyses in LALES and 2) we
included the General Vision domain in the Task composite rather than Well-Being.
42
3.2.4 Statistical Analyses
We examined the overall difference in VA and VS-QOL from baseline to 8-year follow-up
by calculating the mean difference between examinations and comparing them with paired
t-tests. QOL scores were log-transformed before completing the t-test. For each NEI-VFQ
scale, we calculated the proportion of participants who lost or gained 5 points or more, to
compare the 8-year change to those previously reported at 4-years (14). For each scale,
we calculated the proportion of participants who lost and gained 5 points or higher. We
evaluated mean change in VS-QOL scores by three levels of VA change: 2 lines loss or
greater, 2 lines gain or better, or less than 2 lines change in VA. All VA measurements
shown are for the presenting, best-seeing eye.
Additionally, we calculated the effect sizes (ES) for the magnitude in the change in QOL
by dividing the difference between baseline and follow-up QOL by the standard deviation
of the baseline score (72). Generally, the closer an absolute ES value is to 1, the larger the
magnitude of change is, with .20 and less sometimes being used as an indicator of small to
no effect (73). For each scale, we also calculated the reliable change index (RCI), a z-score
index that also attempts to assess meaningful change at the individual level (rather than in
group means) (74). The RCI is calculated by conducting a z-test of the difference in scores
divided by the SD of the difference, where the SD is multiplied by the square root of 1 -
reliability (in this case, Cohen’s alpha) for the scale. An absolute RCI of 1.96 is a statistically
significant magnitude for that person’s change at P = .05. Finally, we computed the average
change in VS-QOL for participants with incident VI at 4-years by the cause of the incident
VI.
To assess the psychometric validity of the task and well-being composites, we used
confirmatory factor analysis to compare model fit using 1) just the traditional composite 2)
the task and well-being composites and 3) all subscales. To compare the fit of each model,
we calculated the goodness-of-fit chi-square test statistic (GOF), the comparative fit index
43
(CFI) and the root mean square error of approximation (RMSEA), as well as using nested
likelihood tests to calculate P-values (55,56). Generally, the closer a CFI is to 1 and the
closer the RMSEA is to 0, the better the model fits, while smaller P-values in likelihood
tests indicate that the more complex model fits better than the simpler model. Additionally,
we analyzed the modification indices of the task and well-being composites, an estimate of
how a model fit might improve by making modifications to the CFA.
We modeled the impact of changing VA on VS-QOL for each composite (the traditional,
task, and well-being composites) using ordinary least squares models (OLS) of the change
in scores. To address potential non-linearity in the relationship between VA and VS-QOL,
we modeled VA using restricted cubic splines (59). We also fit OLS models using the RCI
transformation of QOL scores. We then calculated predicted mean change in quality of life
at exactly 2 lines and 4 lines of visual acuity gained and lost (compared to no change). We
calculated these differences using the parametric G-formula, a causal inference tool that uses
predictions (which we modeled using the OLS regression) to estimate counterfactual effects
(48). The G-formula is a counterfactual approach that allows us to answer the question:
what if the visual acuity of everyone in the cohort changed by a certain amount compared
to if no one’s visual acuity changed? The parametric G-formula is thus a flexible tool for
answering specific causal questions. We used the bootstrap to compute confidence intervals
for the predicted change in VS-QOL at 2 and 4 lines gained or lost.
Additionally, we addressed potential selection bias due to loss to follow-up by weight-
ing the OLS models with stabilized inverse probability of censoring weights (48). These
weights are estimated by predicting the probability of being lost to follow-up (“censored”)
using a logistic regression model. The probabilities are then inverted and multiplied by the
unconditional probability of censoring (a logistic regression model without predictors). In-
verse probability weights create a psuedopopulation that, if appropriately specified, better
approximates the results one would have had if lost to follow up had not occurred, thus
reducing the impact selection bias if present.
44
We selected potential confounders using a priori knowledge of risk factors for VA and
VS-QOL. To more formally evaluate the minimum set of covariates included in the regression
models, we used directed acyclic graphs and structural causal model theory (see appendix)
(47,75,76). This set of variables included baseline age, sex, cataracts, comorbidities, educa-
tion, employment, acculturation, baseline visual acuity, and insurance status. We imputed
missing covariate values with multiple imputation using all other covariates and visual acu-
ity, a robust alternative to complete case analysis that often improves inference (77,78). In
each model, continuous covariates were modeled using restricted cubic splines. For predic-
tion plots, we set continuous covariates to their mean and categorical covariates to the most
frequent category. Additionally, we conducted sensitivity analyses of these models without
imputation and IPCWs.
We conducted all analyses in R 3.6.3 (79) using the following R packages: lavaan and
semPlot (80,81) (confirmatory factor analysis), rms (82) (restricted cubic splines), mice (78)
(imputation), dagitty and ggdag (83,84) (directed acyclic graphs), and ggplot2 (85) (all data
visualizations).
3.3 Results
3.3.1 Description of Study Population
This analysis includes 2234 LALES participants with complete information for VA and
QOL at both baseline and 8-year follow-up examinations. 6118 participants had complete
VA data at baseline and 3308 at 8-year follow-up. Of those, 280 were missing QOL data at
baseline, 857 at follow-up, and 107 for both. Those who were lost to follow-up had slightly
worse QOL (mean composite NEI-VFQ-25 scores 81.7 vs. 85) and VA (mean logMAR 0.08
vs. 0.03) than those who remained in the study. Those lost to follow-up were also more likely
to be male (54.2% female vs. 61.0% female), to be older (56.8 mean age vs. 53.6), and had
45
more co-morbidities (1.6 mean co-morbidities vs. 1.4).
Of those with complete VA and QOL data, 62.3% were female and had a mean age
of 53.8 at baseline. On average, participants had 1.4 comorbidities. 52.2% reported being
employed at baseline, while 54.1% reported having vision insurance. 64.7% of participants
reported having less than high school education.
3.3.2 Overall Changes in Health-Related Quality of Life and Visual Acuity
In Table 4.1, we summarize the changes in VA and VS-QOL from baseline to 8-year
follow-up. Both presenting and best-corrected VA decreased over the 8-year period. The
mean logMAR values for participants increased modestly, indicating a slight overall worsen-
ing in VA. Many more participants had changes in their vision at 8-year follow-up compared
to 4-year follow-up, with 22.6% with 2 lines or greater reduction in vision at 8 years vs. 1%
at 4 years (6.3% vs. 2.6% 2 lines or greater improved vision). VS-QOL increased on average
for all scales with the exception of the General Health and General Vision Subscales in the
NEI-VFQ-25 and the PCS in the SF-12. The most considerable absolute changes were for
ocular pain (mean± SD: 11.04± 21.46), near vision (9.78± 21.66), and distance vision (6.58
± 18.6). For the composite scores, the Task composite had the most change (5.23 ± 15.1),
followed by the traditional composite (4.25 ± 13.42), and the Well-Being composite (2.33
± 23.26). The Well-Being composite scores were lower overall compared to the Task and
traditional composites; notably, the traditional composite seemed to be an average of the
two new scales. The traditional composite also increased in average difference from 4-year
follow-up to 8-year follow-up (1.9 vs. 4.3). The General Health and PCS scales both had
modest decreases, with 1.96 points and 0.28 points lower at 8 years from baseline, respec-
tively. Both the Mental Health subscale measured by the NEI-VFQ-25 and the MCS scale
measured by the SF-12 increased at follow-up. The effect for the NEI-VFQ-25 subscale,
46
which addresses vision-specific mental health, was small, with less than a 1-point change,
while the MCS was slightly larger, with a 3.6-point change.
3.3.3 Psychometric Properties of Task and Well-Being Composites
In Table 4.2, we present the confirmatory factor analysis of the composite scale, the task
and well-being composites, and the subscales of the NEI-VFQ-25. Nested likelihood tests
suggested that using all subscales fit the data best, while the task and well-being composites
were an improvement over the traditional composite (all P < .001). Likewise, the CFI was
0.94 for all scales, 0.82 for the new composites, and 0.78 for the traditional composite, and
the RMSEA was 0.05 for all scales, 0.08 for the new composites, and 0.09 for the traditional
composite. We also present the modification indices (see appendix) for the Task and Well-
Being model. This analysis suggests that adding the item “Because of your eyesight, how
much difficulty do you have seeing how people react to things you say?” to Task and “Are
you limited in how long you can work or do other activities because of your vision?” to
Well-being may improve model fit. Modeling correlations between several related items may
also improve model fit.
3.3.4 Changes in Health-Related Quality of Life by Visual Acuity
In Table 3.3, we summarize the unadjusted change in QOL stratified by change in VA.
As with Table 4.2, mean changes in vision-specific QOL were largely positive, suggesting
an overall improvement, but gains were greater for those whose VA improved over 8 years.
Notably, effect sizes were all 0.2 or smaller for vision loss, compared to around 0.5 for vision
gain. Similarly, general health measures like the General Health subscale and the PCS
suggested declining health overall, with those with VA loss having larger drops in QOL.
47
Mental health as measured by the MCS improved uniformly across levels of VA change,
while vision-specific mental health improved in a similar pattern to other vision-specific
scales.
Forallthreecompositescales, therewas anincreaseinVS-QOLacrosslevelsofVA,with
the highest increases for those whose VA improved by 2 lines or more. For the traditional
composite, the mean increase in VS-QOL was 8.41 ± 15.84; for the Task composite, the
mean ± SD change in VS-QOL was 9.75 ± 17.38; and for the Well-Being composite, it was
6.94± 16.4. While all scores in those with improvement in VA increased more than 5 points,
on average, it was highest for the Task composite, followed by the traditional composite,
and then the Well-Being composite. A similar pattern was present for those with reduced
VA, with smaller gains, all less than 5 points. Noticeably, the Well-Being composite score
was similar for VA loss and no VA change.
We also present change in VS-QOL by cause of incident VI at 4 years (see appendix).
The four most frequent causes of 4-year incident VI were refractive error, cataract, diabetic
retinopathy, age-related macular degeneration, and glaucoma (24). At four years of follow-
up, 57 participants in LALES with incident VA had complete data on VSQOL. Participants
with incident VA due to diabetic retinopathy were the most impacted. On average, their
composite VS-QOL score decreased by 15.5 points; similarly, their Task-related VS-QOL
decreased 18.4 points, and their Well-being-related VS-QOL decreased 12.5 points. Those
with incident VA due to glaucoma had, on average, a 10-point reduction in composite VS-
QOL. Cataract and other causes had more modest decreases, while those with AMD saw an
increaseinVS-QOL(notably, duetosmallsamplesize, thiswasdrivenbyasingleparticipant
who had near-zero VS-QOL scores at baseline).
In Figure 4.1, we present predictions for change in QOL and the RCI for all three
composites. All the composites showed that visual acuity increased quality of life, as well as
RCI transformation. The adjusted regression lines showed similar results suggested by Table
48
3.3: the slopes for the Task and traditional composite were both steeper at lower levels of
VA change than the Well-Being composite. The less pronounced change in Well-Being with
VA loss was still present after adjustment, with a slightly sharper slope for those with VA
improvement (e.g. VS-QOL improved more in those with VA improvement), suggesting that
the Well-Being composite may function differently when there is vision loss.
3.3.5 Predicted Mean Change in VS-QOL at 2- and 4-Lines Change in Visual Acuity
In Figure 4.2 and Table 3.4, we present the predicted mean change in VS-QOL when
losing or gaining 2 lines of visual acuity compared to no change. For the traditional com-
posite, gaining 2 lines of visual acuity improved quality of life by 2.55 points (95% CI 1.00,
3.77), while losing 2 lines decreased quality of life by -1.41 points (95% CI -2.67, -0.23). The
Task and Well-Being composites followed similar overall patterns but were but were slightly
different from one another when lost visual acuity was lost. For the Task composite, those
with improved visual acuity had quality of life 2.63 points (95% CI 0.95, 3.94) higher scores,
while those with worsened visual acuity had -1.72 points (95% CI -3.17, -0.41) lower scores.
Likewise, for the Well-Being composite, improved visual acuity raised quality of life 2.47
points (95% CI 0.88, 3.68), while worsened visual acuity reduced it -1.09 points (95% CI
-2.51, 0.20). Changes at 4 lines of visual acuity showed similar patterns, but the differences
between the Task and Well-Being composites was more pronounced: For the Task-related
quality of life, losing 4 lines of visual acuity led to a loss of -4.15 points (95% CI -6.36, -2.18),
compared to -2.38 points (95% CI -4.74, -0.43) of Well-Being-specific QOL.
49
3.4 Discussion
In this population-based study of Latino adults, we found that long-term change in
visual acuity impacts vision-specific quality of life. We also found that developing new
scoring algorithms for the NEI-VFQ-25 that calculate separate task and well-being focused
composite scores, as opposed to the traditional composite score that treats them as a single
dimension of visual functioning, improves the psychometric validity of the NEI-VFQ-25.
Patients may better understand the impact of 2 lines of improved vision on the ability to
complete vision-related tasks in their home compared to a general composite score. The
best fit was for individually reporting the 11 subscales. Composite and individual scales
may be useful to patients depending on the context of the VA change. We also analyzed an
additional4yearsoffollow-updatacomparedtoourpreviousreport. Manymoreparticipants
had changes in their vision at 8-year follow-up compared to 4-year follow-up, with 22.6%
with 2 lines or greater reduction in vision at 8 years vs. 1% at 4 years (6.3% vs. 2.6% 2
lines or greater improved vision). Consequently, changes in VS-QOL were also of a greater
magnitude, with almost all subscales increasing over 8 years. The traditional composite score
also increased from an average difference of 1.9 to 4.3.
Previous work in clinical and population-based settings suggested that vision loss was
associated with reduced QOL, including among those who had changes in vision due to
cataract surgery and glaucoma (17,86). More recently, the Singapore Malay Eye Study
Research Group found a longitudinal decrease in vision lead to worse VS-QOL (using a
different visual function assessment tool than the NEI-VFQ-25) (87). In studies of longi-
tudinal change, the change in the NEI-VFQ-25 composite score varied for those whose VA
improved, but the magnitude and direction of change were similar across studies. The Sub-
macular Surgery Trials Research Group, for instance, found a change of 8.3 point increase in
the NEI-VFQ-25 composite score among patients enrolled in clinical trials whose VA then
increased by at least 2 lines (17,88). The Age-Related Eye Disease Study Research Group
found a 10 point increase for patients in participants whose VA increased by at least 3 lines
50
(89). Notably, other studies tended to be clinically-based, which may result in differences in
change in VS-QOL due to the severity and specific types of eye disease among participants,
as well as treatments; our estimates were population-based. Our previous work suggested a
5 point difference (10,14). Our previous results are mathematically consistent with results
presented here, which represent more precise contrasts (exactly 2 lines change versus 2 lines
or greater change). A linear effect of 2.5 points per two lines creates an effect of 5 points on
average when categorized because categorization produces a weighted average of the impact
of all changes above that cutoff.
In our population-based estimates, the distribution for clinically relevant change in
quality of life centers around 2 points per 2 lines of VA, or 5 points for 2 lines change or
greater. Correctly estimated changes in quality of life are an essential metric for researchers
and clinicians interested in improving the lives of patients with or at risk for low vision.
Because quality of life also includes the participant’s perception of impact, it is also probable
that this heterogeneity is both a function of study designs and populations and also a natural
aspect of quality of life, not a limitation per se. An exact cutoff for clinically significant
quality of life probably doesn’t exist, but this and previous work suggest that a change of 2
points per 2 lines of VA or 5 points for 2 lines change or greater may be relevant to clinical
decision making.
In this study, the Task and Well-Being composites fit the data better than the tradi-
tional, single composite summary (CFI of 0.82 for Task and Well-Being composites vs. 0.78
for the traditional composite, P <.001). While the magnitudes of the test statistics were
similar, a 2 composite measure described by daily task and social well being may be more
meaningful to a patient trying to understand the potential risk and benefits of treatment
or change in vision than an abstract single composite summary measure. The impact of
visual acuity on VS-QOL as measured by the new composites differed depending on the
change in VA. Other authors have pointed out that the traditional NEI-VFQ-25 composite
may be multi-dimensional (54,60). When averaging the effect across dimensions, scores miss
51
subtleties within the dimensions, such as when Task and Well-Being scores may differ across
changes in VA.
Because the new composites are more specific to daily tasks and well-being, they may
be more straightforward for clinicians and patients to interpret. Notably, the 12 subscales
initially proposed for the NEI-VFQ-25 fit the data best. Using the individual scales may
be more meaningful when communicating to patients about the potential benefits (or risks)
of VA change due to disease or treatment. For example, the impact of VA improvement
following treatment on driving may be more meaningful to a patient than a 5-point change
in the summary composite score. However, when the overall summary measure is more
useful, then the Task and Well-Being composites may offer a more psychometrically valid
approach to summarizing scores. Together with previous evidence, our results suggest that
the Task and Well-Being composites may make a useful addition to the way researchers
measure VS-QOL with the NEI-VFQ-25.
While a strength of this study is its detailed ophthalmic examinations and in-home
interviews over almost a decade of follow-up, one potential limitation is the rate loss to
follow-up in LALES. Loss to follow-up is common in longitudinal studies and not necessarily
indicative of selection bias, but those who were lost to follow-up had worse VA and VS-QOL
at baseline and were older than those who stayed in the study. Worse levels of VA and
VS-QOL among those not included may underestimate the real difference in VS-QOL given
change in VA. The results of models weighted with the inverse probability of censoring to
address were similar (if slightly underestimated) to models without weights (see appendix).
It is possible that we misspecified the weights, leaving residual selection bias. However, our
results are consistent with our previous work on the relationship between QOL and VA. A
strength of this longitudinal analysis includes the addition of 4 years of additional VA data,
which allows us to examine the impact of a broader range of VA change with precision due
to an increased number of participants with 2 or more lines of VA change. While we were
not able to assess causes of VI at 8-years, we were able to assess VS-QOL by cause of VI at
52
4-years. Additional strengths of the paper include the population-based design of LALES,
the large sample size, and that the Latino participants represent an underserved population.
A potential limitation not addressed by this analysis is its use of classical test theory
for scoring. While the classical approach can be powerful under the right circumstances, one
of its significant assumptions is that each item contributes equally to the overall score, while
each potential response is equally spaced from the others. Item response theory addresses
these issues, and there have been several recent attempts to use this approach to improve
the NEI-VFQ-25 (54,90). In this paper, we report CTT for the comparability of findings to
our 4-year and baseline publications. Further, we address concerns of multidimensionality
by using separate Task and Well-Being composites. We use the IRT approach in chapter 5
to further investigate the measurement of VS-QOL and VS-QOL’s relationship with vision
loss. These scales may also be improved by modifying the underlying CFA model form, for
instance by adding items to more than one scale or modeling correlation between items.
We chose not to use this approach so that we can maintain the traditional scoring used
for the NEI-VFQ-25. However, we provide an exploration of how these modifications might
improve the model in the appendix. Future research should confirm the effectiveness of these
modifications.
In this longitudinal, population-based study of Latino adults, we found that a clinically
relevant change in visual acuity impacts vision-specific quality of life. We also found that
scoring the NEI-VFQ-25 tool to measure vision-specific well-being and task-related quality
of life fit the data better than the traditional composite scoring. These scores represent
different dimensions of vision-specific quality of life that are affected differently by a change
in VA with the potential to make a more meaningful interpretation of findings to patients.
We found that the magnitude of change was influenced both by the direction and severity of
VA change. Specifically, we found that while a 2-line change in visual acuity led to about a
2.5-point increase in both task-related and well-being-related quality of life, a 2-line loss in
VA caused about a 1-point loss in well-being-related quality of life vs. about a 2-point loss
53
in task-related quality of life. These results suggest that visual acuity influences quality of
life differently depending on the dimension of quality of life and direction of change. We also
found that many more participants had changes in vision after 8 years of follow-up compare
to 4 years, which led to a greater change in VS-QOL. This suggests that collecting long-term
data is important to understanding the impacts of changing vision on VS-QOL.
3.5 Acknowledgments
TheauthorswouldliketothanktheparticipantsofLALESthathavegenerouslydonated
their time and data to our study.
54
3.6 References
1. Jones GC, Rovner BW, Crews JE, et al. Effects of depressive symptoms on health
behavior practices among older adults with vision loss. Rehabilitation Psychology.
2009;54(2):164–172.
2. Lamoureux EL, Fenwick E, Moore K, et al. Impact of the Severity of Distance
and Near-Vision Impairment on Depression and Vision-Specific Quality of Life in
Older People Living in Residential Care. Investigative Ophthalmology & Visual Sci-
ence [electronicarticle]. 2009;50(9):4103–4109. (https://iovs.arvojournals.org/article.
aspx?articleid=2185912). (Accessed January 26, 2019)
3. Rovner BW, Ganguli M. Depression and Disability Associated with Impaired Vision:
The MoVIES Project. Journal of the American Geriatrics Society [electronic ar-
ticle]. 1998;46(5):617–619. (http://onlinelibrary.wiley.com/doi/abs/10.1111/j.1532-
5415.1998.tb01080.x). (Accessed January 26, 2019)
4. Zhang X, Bullard KM, Cotch MF, et al. Association Between Depression and Func-
tional Vision Loss in Persons 20 Years of Age or Older in the United States, NHANES
2005-2008. JAMA Ophthalmology [electronic article]. 2013;131(5):573–581. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/1660943). (Accessed Jan-
uary 26, 2019)
5. Crews JE, Campbell VA. Vision impairment and hearing loss among community-
dwelling older Americans: Implications for health and functioning. American Journal
of Public Health. 2004;94(5):823–829.
6. Ivers RQ, Norton R, Cumming RG, et al. Visual impairment and risk of hip fracture.
American Journal of Epidemiology. 2000;152(7):633–639.
7. Lee DJ, Gómez-Marín O, Lam BL, et al. Visual acuity impairment and mortality in
US adults. Archives of Ophthalmology (Chicago, Ill.: 1960). 2002;120(11):1544–1550.
55
8. 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 Ophthalmology [electronic article]. 2016;134(7):802–809. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/2523780). (Accessed Jan-
uary 26, 2019)
9. Choudhury F, Varma R, Klein R, et al. Age-Related Macular Degeneration and
Quality of Life in Latinos: The Los Angeles Latino Eye Study. JAMA Ophthalmol-
ogy [electronic article]. 2016;134(6):683–690. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/2514080). (Accessed March 12, 2018)
10. Globe DR, Wu J, Azen SP, et al. The impact of visual impairment on self-reported vi-
sual functioning in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(6):1141–1149. (http://www.aaojournal.org/article/S0161-
6420(04)00303-3/abstract). (Accessed March 19, 2018)
11. Mazhar K, Varma R, Choudhury F, et al. Severity of Diabetic Retinopathy and
Health-Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology
[electronic article]. 2011;118(4):649–655. (http://www.aaojournal.org/article/S0161-
6420(10)00819-5/abstract). (Accessed March 19, 2018)
12. PatinoCM,VarmaR,AzenSP,etal. TheImpactofChangeinVisualFieldonHealth-
Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2011;118(7):1310–1317. (http://www.aaojournal.org/article/s0161-
6420(10)01344-8/abstract). (Accessed March 19, 2018)
56
13. Varma R, Paz SH, Azen SP, et al. The Los Angeles Latino Eye Study: Design,
methods, and baseline data. Ophthalmology [electronic article]. 2004;111(6):1121–
1131. (http://www.aaojournal.org/article/S0161-6420(04)00291-X/abstract). (Ac-
cessed March 19, 2018)
14. McKean-Cowdin R, Varma R, Hays RD, et al. Longitudinal Changes in Visual Acu-
ity and Health-related Quality of Life: The Los Angeles Latino Eye Study. Ophthal-
mology [electronic article]. 2010;117(10):1900–1907.e1. (http://www.aaojournal.org/
article/S0161-6420(10)00132-6/abstract). (Accessed March 19, 2018)
15. Linder M, Chang TS, Scott IU, et al. Validity of the visual function index (vf-
14) in patients with retinal disease. Archives of Ophthalmology [electronic article].
1999;117(12):1611–1616. (http://dx.doi.org/10.1001/archopht.117.12.1611)
16. Mangione CM, Phillips RS, Lawrence MG, et al. Improved Visual Function
and Attenuation of Declines in Health-Related Quality of Life After Cataract
Extraction. Archives of Ophthalmology [electronic article]. 1994;112(11):1419–
1425. (https://jamanetwork.com/journals/jamaophthalmology/fullarticle/640890).
(Accessed March 19, 2018)
17. Health- and vision-related quality of life among patients with choroidal neovascular-
ization secondary to age-related macular degeneration at enrollment in randomized
trials of submacular surgery: SST report no. 4. American Journal of Ophthalmol-
ogy [electronic article]. 2004;138(1):91–108. (http://www.sciencedirect.com/science/
article/pii/S0002939404001370)
18. Parrish RK, Gedde SJ, Scott IU, et al. Visual function and quality of life
among patients with glaucoma. Archives of Ophthalmology (Chicago, Ill.: 1960).
1997;115(11):1447–1455.
57
19. Scott IU, Smiddy WE, Schiffman J, et al. Quality of life of low-vision patients and the
impact of low-vision services. American Journal of Ophthalmology [electronic article].
128(1):54–62. (http://dx.doi.org/10.1016/S0002-9394(99)00108-7). (Accessed March
20, 2018)
20. Healthy people 2020: vision. Healthy People 2020. (https://www.healthypeople.gov/
2020/topics-objectives/topic/vision)
21. Lamoureux E, Pesudovs K. Vision-Specific Quality-of-Life Research: A Need
to Improve the Quality. American Journal of Ophthalmology [electronic arti-
cle]. 2011;151(2):195–197.e2. (https://doi.org/10.1016/j.ajo.2010.09.020). (Accessed
March 18, 2020)
22. Coleman AL, Yu F, Keeler E, et al. Treatment of Uncorrected Refractive Error
Improves Vision-Specific Quality of Life. Journal of the American Geriatrics Society
[electronic article]. 2006;54(6):883–890. (http://onlinelibrary.wiley.com/doi/abs/10.
1111/j.1532-5415.2006.00817.x). (Accessed March 23, 2020)
23. BourneRR,StevensGA,WhiteRA,etal. Causesofvisionlossworldwide, 1990–2010:
A systematic analysis. The lancet global health. 2013;1(6):e339–e349.
24. Varma R, Chung J, Foong AWP, et al. Four-Year Incidence and Progression of Vi-
sual Impairment in Latinos: The Los Angeles Latino Eye Study (LALES). Amer-
ican journal of ophthalmology [electronic article]. 2010;149(5):713–727. (https:
//www.ncbi.nlm.nih.gov/pmc/articles/PMC2919324/). (Accessed January 26, 2019)
25. Varma R, Kim JS, Burkemper BS, et al. Prevalence and causes of visual impairment
and blindness in chinese american adults: The chinese american eye study. Jama
Ophthalmology. 2016;134(7):785–793.
58
26. Munoz B, West SK, Rubin GS, et al. Causes of blindness and visual impairment
in a population of older americans: The salisbury eye evaluation study. Archives of
ophthalmology. 2000;118(6):819–825.
27. Tang Y, Wang X, Wang J, et al. Prevalence and causes of visual impairment in a
chinese adult population: The taizhou eye study. Ophthalmology. 2015;122(7):1480–
1488.
28. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindness in a rural chinese adult population: The handan eye study. Ophthalmology.
2008;115(11):1965–1972.
29. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in beijing: The beijing eye study. Ophthalmology. 2006;113(7):1134–e1.
30. ZhaoJ,XuX,EllweinLB,etal. Causesofvisualimpairmentandblindnessinthe2006
and 2014 nine-province surveys in rural china. American Journal of Ophthalmology.
2019;197:80–87.
31. AtteboK,MitchellP,SmithW.Visualacuityandthecausesofvisuallossinaustralia:
The blue mountains eye study. Ophthalmology. 1996;103(3):357–364.
32. Hyman L, Wu S-Y, Connell AM, et al. Prevalence and causes of visual impairment
in the barbados eye study. Ophthalmology. 2001;108(10):1751–1756.
33. Klaver CC, Wolfs RC, Vingerling JR, et al. Age-specific prevalence and causes
of blindness and visual impairment in an older population: The rotterdam study.
Archives of ophthalmology. 1998;116(5):653–658.
34. Rodriguez J, Sanchez R, Munoz B, et al. Causes of blindness and visual impairment
in a population-based sample of US hispanics. Ophthalmology. 2002;109(4):737–743.
59
35. CongdonN,O’ColmainB,KlaverC,etal. Causesandprevalenceofvisualimpairment
among adults in the united states. Archives of Ophthalmology (Chicago, Ill.: 1960).
2004;122(4):477–485.
36. VanNewkirk MR, Weih L, McCarty CA, et al. Cause-specific prevalence of bilateral
visual impairment in victoria, australia: The visual impairment project. Ophthalmol-
ogy. 2001;108(5):960–967.
37. 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. Investigative Ophthalmology & Visual Science. 2002;43(11):3393–
3398.
38. Unzueta M, Globe D, Wu J, et al. Compliance with recommendations for follow-up
care in latinos: The los angeles latino eye study. Ethnicity & disease. 2004;14:285–91.
39. Varma R, Mohanty SA, Deneen J, et al. Burden and predictors of undetected eye
disease in mexican-americans. Medical care. 2008;46(5):497–506.
40. Patty LE, Morales LS, Lai M, et al. Self-reported utilization of eye care among latinos
in the los angeles latino eye study (LALES). Investigative Ophthalmology & Visual
Science. 2009;50(13):2502–2502.
41. Varma R, Torres M, McKean-Cowdin R, et al. Prevalence and risk factors for re-
fractive error in adult chinese americans: The chinese american eye study. American
Journal of Ophthalmology. 2017;175:201–212.
42. Jiang X, Varma R, Torres M, et al. Self-reported use of eye care among adult chinese
americans in the chinese american eye study. American journal of ophthalmology.
2017;176:183–193.
60
43. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by hispanics: Findings from HHANES 1982-84. American Journal of
Public Health. 1990;80(Suppl):11–19.
44. Zheng Y, Lamoureux EL, Ikram MK, et al. Impact of migration and acculturation
on prevalence of type 2 diabetes and related eye complications in indians living in a
newly urbanised society. PloS One. 2012;7(4):e34829.
45. Streiner DL, Cairney J, Norman GR. Health measurement scales: A practical guide
to their development and use. Oxford Univ. Press; 2015.
46. Ayala RJde. The theory and practice of item response theory. Guildford Press; 2009.
47. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Medical
Research Methodology [electronic article]. 2008;8:70. (https://www.ncbi.nlm.nih.gov/
pmc/articles/PMC2601045/). (Accessed November 21, 2018)
48. Hernán MA, Robins JM. Causal Inference. Boca Raton: Chapman Hall/CRC; 2019.
49. Steinberg EP, Tielsch JM, Schein OD, et al. The VF-14: An Index of Functional
ImpairmentinPatientsWithCataract. Archives of Ophthalmology [electronicarticle].
1994;112(5):630–638. (https://doi.org/10.1001/archopht.1994.01090170074026)
50. Mangione CM, Phillips RS, Seddon JM, et al. Development of the ’Activities of Daily
Vision Scale’. A measure of visual functional status. Medical Care. 1992;30(12):1111–
1126.
51. Margolis MK, Coyne K, Kennedy-Martin T, et al. Vision-specific instruments for the
assessment of health-related quality of life and visual functioning: A literature review.
PharmacoEconomics. 2002;20(12):791–812.
61
52. Mangione CM, Lee PP, Gutierrez PR, et al. Development of the 25-list-item
National Eye Institute Visual Function Questionnaire. Archives of Ophthalmol-
ogy [electronic article]. 2001;119(7):1050–1058. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/267134). (Accessed March 19, 2018)
53. Mangione CM, Lee PP, Pitts J, et al. Psychometric Properties of the National
Eye Institute Visual Function Questionnaire (NEI-VFQ). Archives of Ophthalmology
[electronic article]. 1998;116(11):1496–1504. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/264225). (Accessed March 19, 2018)
54. Pesudovs K, Gothwal VK, Wright T, et al. Remediating serious flaws in the Na-
tional Eye Institute Visual Function Questionnaire. Journal of Cataract & Refrac-
tive Surgery [electronicarticle]. 2010;36(5):718–732. (http://linkinghub.elsevier.com/
retrieve/pii/S0886335010002531). (Accessed March 12, 2018)
55. Kline RB. Principles and practice of structural equation modeling. The Guilford
Press; 2016.
56. Bentler PM. On tests and indices for evaluating structural models. Personal-
ity and Individual Differences [electronic article]. 2007;42(5):825–829. (http://
www.sciencedirect.com/science/article/pii/S0191886906003898). (Accessed March
18, 2020)
57. Brucker PS, Yost K, Cashy J, et al. General population and cancer pa-
tient norms for the functional assessment of cancer therapy-general (FACT-
g). Evaluation & the Health Professions [electronic article]. 2005;28(2):192–
211. (%20%0A%20%20%20%20%20%20%20%20https://doi.org/10.1177/
0163278705275341%0A%20%20%20%20%0A)
62
58. Wright BD. Solving MEASUREMENT PROBLEMS WITH THE Rasch MODEL.
Journal of Educational Measurement [electronic article]. 1977;14(2):97–116. (https:
//onlinelibrary.wiley.com/doi/abs/10.1111/j.1745-3984.1977.tb00031.x)
59. Harrell F. Regression modeling strategies: With applications to linear models, logistic
and ordinal regression, and survival analysis. springer; 2016.
60. Medeiros FA, Gracitelli CPB, Boer ER, et al. Longitudinal Changes in Quality of
Life and Rates of Progressive Visual Field Loss in Glaucoma Patients. Ophthalmol-
ogy [electronic article]. 122(2):293–301. (http://dx.doi.org/10.1016/j.ophtha.2014.08.
014). (Accessed March 20, 2018)
61. Association WM. World Medical Association Declaration of Helsinki: Ethical Prin-
ciples for Medical Research Involving Human Subjects. JAMA [electronic article].
2013;310(20):2191–2194. (https://doi.org/10.1001/jama.2013.281053)
62. GlobeDR,Schoua-GlusbergA,PazS,etal. Usingfocusgroupstodevelopaculturally
sensitive methodology for epidemiological surveys in a Latino population: Findings
from the Los Angeles Latino Eye Study (LALES). Ethnicity & disease [electronic
article]. 2002;12(2):259–266. (http://europepmc.org/abstract/med/12019936). (Ac-
cessed March 19, 2018)
63. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by Hispanics: Findings from HHANES 1982-84. American Journal
of Public Health [electronic article]. 1990;80(Suppl):11–19. (https://doi.org/10.2105/
AJPH.80.Suppl.11). (Accessed March 20, 2018)
64. Brody BL, Gamst AC, Williams RA, et al. Depression, visual acuity, comorbidity,
and disability associated with age-related macular degeneration. Ophthalmology [elec-
tronicarticle]. 108(10):1893–1900. (http://dx.doi.org/10.1016/S0161-6420(01)00754-
0). (Accessed March 20, 2018)
63
65. Globe DR, Varma R, Torres M, et al. Self-reported comorbidities and visual function
inapopulation-basedstudy: Thelosangeleslatinoeyestudy. Archives of Ophthalmol-
ogy [electronic article]. 2005;123(6):815–821. (http://dx.doi.org/10.1001/archopht.
123.6.815)
66. Linn Bernard S., Linn Margaret W., Gurel Lee. Cumulative illness rating scale.
Journal of the American Geriatrics Society [electronic article]. 2015;16(5):622–626.
(https://onlinelibrary.wiley.com/doi/full/10.1111/j.1532-5415.1968.tb02103.x). (Ac-
cessed March 19, 2018)
67. Azen SP, Varma R, Preston-Martin S, et al. Binocular Visual Acuity Sum-
mation and Inhibition in an Ocular Epidemiological Study: The Los Angeles
Latino Eye Study. Investigative Ophthalmology & Visual Science [electronic ar-
ticle]. 2002;43(6):1742–1748. (http://iovs.arvojournals.org/article.aspx?articleid=
2123581). (Accessed March 19, 2018)
68. Varma R, Fraser-Bell S, Tan S, et al. Prevalence of age-related macular degen-
eration in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(7):1288–1297. (http://www.aaojournal.org/article/S0161-
6420(04)00313-6/abstract). (Accessed March 19, 2018)
69. Varma R, Torres M, Peña F, et al. Prevalence of diabetic retinopathy in adult
Latinos: The Los Angeles Latino Eye Study. Ophthalmology [electronic article].
2004;111(7):1298–1306. (http://www.aaojournal.org/article/S0161-6420(04)00336-
7/abstract). (Accessed March 19, 2018)
64
70. Ware JE, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: Construction
ofScalesandPreliminaryTestsofReliabilityandValidity. Medical Care [electronicar-
ticle]. 1996;34(3):220. (https://journals.lww.com/lww-medicalcare/Abstract/1996/
03000/A_12_Item_Short_Form_Health_Survey__Construction.3.aspx). (Ac-
cessed March 19, 2018)
71. Ware J, A. Kosinski M, D. Keller S. SF-12: How to Score the SF-12 Physical and
Mental Health Summary Scales. 1998;
72. Rosenthal R. Parametric measures of effect size. In: The handbook of research syn-
thesis. New York, NY, US: Russell Sage Foundation; 1994:231–244.
73. Cohen J. A power primer. Psychological bulletin [electronic article]. 1992;112(1):155–
159. (http://libproxy.usc.edu/login?url=https://search-proquest-com.libproxy2.usc.
edu/docview/614317877?accountid=14749)
74. Hays RD, Brodsky M, Johnston MF, et al. Evaluating the Statistical Significance
of Health-Related Quality-Of-Life Change in Individual Patients. Evaluation & the
Health Professions [electronic article]. 2005;28(2):160–171. (https://doi.org/10.1177/
0163278705275339). (Accessed March 19, 2018)
75. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epi-
demiology (Cambridge, Mass.). 1999;10(1):37–48.
76. Greenland S. Quantifying biases in causal models: Classical confounding vs collider-
stratification bias. Epidemiology (Cambridge, Mass.). 2003;14(3):300–306.
77. Buuren S van. Flexible imputation of missing data. CRC Press; 2018.
65
78. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained
equations in r. Journal of Statistical Software [electronic article]. 2011;45(3):1–67.
(https://www.jstatsoft.org/v45/i03/)
79. R Core Team. R: A language and environment for statistical computing. Vienna,
Austria: RFoundationforStatisticalComputing; 2017.(https://www.R-project.org/)
80. Rosseel Y. lavaan: An R package for structural equation modeling. Journal of Sta-
tistical Software [electronic article]. 2012;48(2):1–36. (http://www.jstatsoft.org/v48/
i02/)
81. Epskamp S, Simon Stuber with contributions from. semPlot: Path diagrams and
visual analysis of various SEM packages’ output. 2017.(https://CRAN.R-project.
org/package=semPlot)
82. Harrell Jr FE. Rms: Regression modeling strategies. 2018.(https://CRAN.R-project.
org/package=rms)
83. Textor J, van der Zander B. Dagitty: Graphical analysis of structural causal models.
2016.
84. Barrett M. Ggdag: Analyze and create elegant directed acyclic graphs. 2018.
85. Wickham H. ggplot2: Elegant graphics for data analysis. Springer-Verlag New York;
2016.(http://ggplot2.org)
86. Alqudah A, Mansberger SL, Gardiner SK, et al. Vision-related quality of life in glau-
coma suspect or early glaucoma patients. Journal of glaucoma [electronic article].
2016;25(8):629–633. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4975427/).
(Accessed March 12, 2018)
66
87. Fenwick EK, Man RK, Ong P, et al. Association of changes in visual acuity
with vision-specific functioning in the singapore malay eye study. JAMA Oph-
thalmology [electronic article]. 2016;134(11):1299–1305. (http://dx.doi.org/10.1001/
jamaophthalmol.2016.3692)
88. Responsiveness of the National Eye Institute Visual Function Questionnaire
to Changes in Visual Acuity. Archives of ophthalmology [electronic article].
2003;121(4):531–539. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1434793/).
(Accessed March 19, 2018)
89. Lindblad AS, Clemons TE. Responsiveness of the National Eye Institute Visual Func-
tion Questionnaire to progression to advanced age-related macular degeneration, vi-
sion loss, and lens opacity: AREDS Report no. 14. Archives of Ophthalmology
(Chicago, Ill.: 1960). 2005;123(9):1207–1214.
90. Hays RD, Morales LS, Reise SP. Item Response Theory and Health Outcomes Mea-
surementin the21stCentury. Medical care [electronicarticle]. 2000;38(9Suppl):II28–
II42. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1815384/). (Accessed
March 19, 2018)
91. Wittenborn J, Rein D. The economic burden of vision loss and eye disorders in the
United States. University of Chicago; 2013.(https://www.preventblindness.org/sites/
default/files/national/documents/Economic%20Burden%20of%20Vision%20Final%
20Report_130611.pdf)
92. Wang X, Lamoureux E, Zheng Y, et al. Health Burden Associated with Visual Im-
pairment in Singapore: The Singapore Epidemiology of Eye Disease Study. Ophthal-
mology [electronic article]. 2014;121(9):1837–1842. (http://www.sciencedirect.com/
science/article/pii/S0161642014002413). (Accessed March 23, 2020)
67
93. Tsai S-Y, Chi L-Y, Cheng C-Y, et al. The Impact of Visual Impairment and Use of
Eye Services on Health-Related Quality of Life among the Elderly in Taiwan: The
Shihpai Eye Study. Quality of Life Research [electronic article]. 2004;13(8):1415–
1424. (https://doi.org/10.1023/B:QURE.0000040791.87602.fe). (Accessed March 23,
2020)
94. Vision impairment and major eye diseases reduce vision-specific emotional well-being
in a Chinese population - ProQuest. (https://search-proquest-com.libproxy1.usc.
edu/docview/1894846003?rfr_id=info%3Axri%2Fsid%3Aprimo). (Accessed March
23, 2020)
95. Varma R, Kim JS, Burkemper BS, et al. Prevalence and Causes of Visual Impairment
andBlindnessinChineseAmericanAdults: TheChineseAmericanEyeStudy. JAMA
Ophthalmology [electronic article]. 2016;134(7):785–793. (https://jamanetwork-
com.libproxy1.usc.edu/journals/jamaophthalmology/fullarticle/2523778). (Accessed
March 21, 2020)
96. Hsu W-M, Cheng C-Y, Liu J-H, et al. Prevalence and causes of visual impairment
in an elderly Chinese population in Taiwan: The Shihpai Eye Study. Ophthalmology.
2004;111(1):62–69.
97. Huang S, Zheng Y, Foster PJ, et al. Prevalence and causes of visual impairment in
Chinese adults in urban southern China. Archives of Ophthalmology (Chicago, Ill.:
1960). 2009;127(10):1362–1367.
98. Saw S-M, Foster PJ, Gazzard G, et al. Causes of blindness, low vision, and
questionnaire-assessed poor visual function in Singaporean Chinese adults: The Tan-
jong Pagar Survey. Ophthalmology. 2004;111(6):1161–1168.
68
99. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindnessinaruralchineseadultpopulation: TheHandanEyeStudy. Ophthalmology.
2008;115(11):1965–1972.
100. Zhao J, Ellwein LB, Cui H, et al. Prevalence of vision impairment in older adults in
rural China: The China Nine-Province Survey. Ophthalmology. 2010;117(3):409–416,
416.e1.
101. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in Beijing: The Beijing Eye Study. Ophthalmology. 2006;113(7):1134.e1–
11.
102. Varma R, Hsu C, Wang D, et al. The Chinese American Eye Study: Design and
Methods. Ophthalmic Epidemiology [electronic article]. 2013;20(6):335–347. (http:
//www.tandfonline.com/doi/full/10.3109/09286586.2013.823505). (Accessed March
21, 2020)
103. SuinnRM,AhunaC,KhooG.TheSuinn-LewAsianSelf-IdentityAcculturationScale:
Concurrent and factorial validation. Educational and Psychological Measurement.
1992;52(4):1041–1046.
104. Burkemper B, Torres M, Jiang X, et al. Factors Associated with Visual Impairment in
Chinese American Adults: The Chinese American Eye Study. Ophthalmic Epidemiol-
ogy [electronic article]. 2019;26(5):329–335. (https://doi.org/10.1080/09286586.2019.
1622737). (Accessed March 22, 2020)
105. Linden WJ van der. Handbook of item response theory. Chapman & Hall/CRC;
2019.
106. Wood SN. Generalized additive models: An introduction with r. CRC Press/Taylor
& Francis Group; 2017.
69
107. Rizopoulos D. Ltm: An r package for latent variable modelling and item response
theory analyses. Journal of Statistical Software [electronic article]. 2006;17(5):1–25.
(http://www.jstatsoft.org/v17/i05/)
108. Fenwick EK, Ong PG, Sabanayagam C, et al. Assessment of the psychometric proper-
ties of the Chinese Impact of Vision Impairment questionnaire in a population-based
study: FindingsfromtheSingaporeChineseEyeStudy. Qualityof LifeResearch [elec-
tronic article]. 2016;25(4):871–880. (https://doi.org/10.1007/s11136-015-1141-1).
(Accessed March 23, 2020)
109. Rees G, Tee HW, Marella M, et al. Vision-specific distress and depressive symptoms
in people with vision impairment. Investigative Ophthalmology & Visual Science.
2010;51(6):2891–2896.
110. Morales LS, Varma R, Paz SH, et al. Self-reported use of eye care among Latinos:
The Los Angeles Latino Eye Study. Ophthalmology. 2010;117(2):207–215.e1.
111. Chang W, Cheng J, Allaire J, et al. Shiny: Web application framework for r.
2020.(https://CRAN.R-project.org/package=shiny)
70
3.7 Tables and Figures
Table 3.1: Overall Change in Visual Acuity and QOL
Variable LALES I LALES III Difference P-
Value
% Down 5
Points
% Up 5
Points
NEI-VFQ-25 Scales
Ocular Pain 78.63
(19.78)
89.67
(13.94)
11.04
(21.46)
<.001 17.19 55.19
Near Vision 80.59
(19.15)
90.37
(15.29)
9.78
(21.66)
<.001 19.36 55.36
Distance Vision 86.82
(17.55)
93.4
(12.88)
6.58 (18.6) <.001 16.94 42.38
Composite (Task) 87.76
(14.1)
93 (11.56) 5.23 (15.1) <.001 16.56 39.7
Peripheral Vision 88.27
(19.54)
93.48
(14.66)
5.21
(21.86)
<.001 12.99 24.06
Composite (Traditional) 85.04
(13.03)
89.29
(10.36)
4.25
(13.42)
<.001 16.6 37.72
Role Function 89.13
(20.14)
92.54
(18.76)
3.42 (25.3) 0.03 14.41 26.13
Composite (Well-Being) 81.93
(13.33)
85.07
(10.56)
3.14
(14.09)
<.001 23.14 39.79
Driving 88.62
(16.41)
91.59
(15.83)
2.97
(17.79)
0.1 19.04 31.7
Color Vision 94.5
(14.61)
97.3
(10.01)
2.81
(16.35)
<.001 6.61 13.21
Social Function 93.57
(13.46)
96.03
(11.26)
2.46
(15.75)
<.001 12.11 21.4
Dependency 90.79
(18.43)
93.12
(18.52)
2.33
(23.26)
0.58 13.81 26.82
General Health 47.32
(23.06)
45.37
(21.59)
-1.96
(26.13)
0.69 31.27 24.4
Mental Health 77.55
(21.1)
78.42
(17.43)
0.87
(24.63)
0.04 43.26 41.83
71
Variable LALES I LALES III Difference P-
Value
% Down 5
Points
% Up 5
Points
General Vision 69.19
(16.28)
68.33
(14.83)
-0.86
(18.77)
0.08 29.47 26.15
SF-12 Scales
MCS (SF-12) 50.11
(10.74)
53.71
(7.41)
3.61 (12.4) <.001 23.46 39.62
PCS (SF-12) 46.75
(9.33)
46.47
(9.81)
-0.28
(11.32)
0.02 28.37 26.88
Visual Acuity
Presenting VA, Binocular -0.01
(0.15)
0.05 (0.21) 0.06 (0.18) <.001 – –
Presenting VA, Better Eye 0.03 (0.16) 0.09 (0.21) 0.06 (0.19) <.001 – –
Best-Corrected VA, Better Eye -0.02
(0.11)
0 (0.16) 0.02 (0.13) <.001 – –
N for each scale range from 1612 to 2257. Visual Acuity on LogMAR scale. P-values from log-transformed paired T-tests.
VA = visual acuity. NEI-VFQ-25 = National Eye Institute Visual Function Questionnaire. SF-12 = Short Form 12. PCS =
Physical Health Composite. MCS = Mental Health Composite.
72
Table 3.2: Confirmatory Factor Analysis of NEI-VFQ-25 Scales
CFA Model Chi-Sq DF GOF P-Value CFI RMSEA Likelihood P-Value
All Scales 2270.01 213 <.001 0.94 0.05 <.001
Task + Well-Being 6877.23 274 <.001 0.82 0.08 <.001
Traditional Composite 8459.89 275 <.001 0.78 0.09
Chi-Sq: Goodness-of-fit Chi-Square Statistic. DF: Degrees of Freedom. CFA: Confirmatory Factor Analysis. CFI: Comparative
Fit Index. RMSEA: Root Mean Square Error of Approximation. Goodness-of-fit P-values from Chi-Square statistic tests.
Likelihood P-values from nested likelihood ratio tests of CFA models.
73
Table 3.3: Change in VS-QOL by Change in Visual Acuity (Best Presenting Eye)
Vision Loss No VA
Change
VA
Improvement
Difference
(SD)
ES Difference
(SD)
ES Difference (SD) ES
NEI-VFQ-25 Composites
Composite (Traditional) 2.37 (14.37) 0.18 4.05 (12.71) 0.33 8.41 (15.84) 0.48
Composite (Task) 2.67 (15.84) 0.2 5.06 (14.45) 0.38 9.75 (17.38) 0.52
Composite (Well-Being) 2.02 (15.07) 0.15 2.94 (13.46) 0.23 6.94 (16.4) 0.4
NEI-VFQ-25 Scales
General Health -3.16 (27.09) -0.14 -2.24 (25.43) -0.1 -0.7 (30.21) -0.03
General Vision -2.8 (19.21) -0.18 -0.93 (18.48) -0.06 4 (19.67) 0.24
Color Vision 2 (17.31) 0.13 2.62 (15.82) 0.19 5.71 (17.32) 0.32
Peripheral Vision 2.8 (22.65) 0.14 4.8 (21.4) 0.25 9.57 (22.69) 0.44
Ocular Pain 8.83 (22.38) 0.47 11.3 (20.96) 0.58 13.3 (22.62) 0.59
Near Vision 7.63 (21.75) 0.41 9.68 (21.35) 0.52 14.13 (23.8) 0.64
Distance Vision 3.68 (19.58) 0.21 6.24 (17.89) 0.37 12.27 (20.58) 0.59
Social Function 1.75 (18.24) 0.13 2.16 (14.66) 0.17 5.85 (16.87) 0.37
Mental Health 1.3 (25.43) 0.06 0.36 (24.17) 0.02 5.97 (26.36) 0.23
Role Function 0.88 (27.16) 0.05 3.1 (24.65) 0.16 7.71 (25.03) 0.3
Dependency 1.21 (25.32) 0.07 1.88 (22.31) 0.11 5.08 (25.58) 0.19
Driving -2.05 (18.77) -0.14 2.93 (16.43) 0.2 6.81 (25.26) 0.28
SF-12 Scales
PCS (SF-12) -0.43 (11.27) -0.05 -0.53 (11.24) -0.06 0.48 (12.27) 0.05
MCS (SF-12) 3.58 (12.91) 0.32 3.53 (12.12) 0.33 3.71 (13.75) 0.33
N for each scale range from 1612 to 2257. Vision loss represents losing 2 lines or greater. VA Improvement represents gaining 2
lines or greater. ES = effect size. SD = standard deviation. VA = visual acuity. NEI-VFQ-25 = National Eye Institute Visual
Function Questionnaire. PCS = Physical Health Composite. MCS = Mental Health Composite.
74
Table 3.4: Predicted Mean Change of VS-QOL at Levels of Visual Acuity Change
Vision Improved Vision Declined
Two Lines Four Lines Two Lines Four Lines
Difference (95% CI) Difference (95% CI) Difference (95% CI) Difference (95% CI)
Composite 2.55 (1.00, 3.77) 6.67 (1.31, 10.59) -1.41 (-2.67, -0.23) -3.30 (-5.45, -1.31)
Task Subcomposite 2.63 (0.95, 3.94) 6.64 (0.93, 10.92) -1.72 (-3.17, -0.41) -4.15 (-6.36, -2.18)
Well-being
Subcomposite
2.47 (0.88, 3.68) 6.68 (0.99, 10.80) -1.09 (-2.51, 0.20) -2.38 (-4.74, -0.43)
N = 2234. Fit with ordinary least squares regression and calculated with the parametric G-formula. Adjusted for age, sex,
cataracts, comorbidities, education, employment, acculturation, baseline visual acuity, and insurance. Weighted by stabilized
inverse probability of censoring weights. CI = confidence intervals.
75
Figure 3.1: Change in Vision-Related Quality of Life and Visual Acuity by Subscale, Multivariably Adjusted
N = 2234. A: y-axis represents change in vision-specific quality of life. B: y-axis represents the reliable change index. Fit
with ordinary least squares regression, adjusted for age, sex, cataracts, comorbidities, education, employment, acculturation,
baseline visual acuity, and insurance. Weighted by stabilized inverse probability of censoring weights.
76
Figure 3.2: G-Formula Calculated Changes in Vision-Related Quality of Life at 2 and 4 Lines Change in
Visual Acuity
N = 2234. Fit with ordinary least squares regression and calculated with the parametric G-formula. Adjusted for age, sex,
cataracts, comorbidities, education, employment, acculturation, baseline visual acuity, and insurance. Weighted by stabilized
inverse probability of censoring weights.
77
CHAPTER 4
EVALUATING THE EFFECT OF VISION LOSS ON QUALITY OF LIFE IN THE
CHINESE EYE STUDY (CHES) WITH ITEM RESPONSE THEORY
4.1 Abstract
We sought to describe and analyze the distribution of vision-specific quality of life (VS-
QOL) by visual acuity (VA) in a population of Chinese Americans using item response
theory (IRT) to measure VS-QOL. 4578 self-identified Chinese adults in a population-based,
cross-sectional cohort study in Monterrey Park, CA, completed comprehensive eye exams,
including measurement of presenting VA, from 2010 to 2013. We calculated VS-QOL using
the NEI-VFQ-25. We used IRT to estimate participant’s Task and Well-Being NEI-VFQ-
25 scores, as well as to understand the psychometric properties of these two domains. We
modeled whether VS-QOL differed by VA using covariate-adjusted OLS models, which we
then used to predict mean change in VS-QOL at 2 and 4 lines of change in VA using the
parametric G-formula. For the Task composite, participants with exactly 2 lines of visual
acuity better than normal vision had 0.39 points (95% CI -2.60, 3.52) higher than those
with 20/20 vision, while participants with 2 lines worse VA had a mean quality of life score
difference of -3.86 (95% CI -4.92, -2.78) lower. For the Well-Being composite, 2 lines of visual
acuity better than normal vision were associated with a better quality of life score of 3.75
points (95% CI 1.00, 6.50), while those with 2 lines worse VA had a mean quality of life score
difference of -2.52 (95% CI -3.41, -1.59). At four lines of VA worse, participants had 8.51
fewer points (95% CI -10.28, -6.66) for the Task composite and 6.22 (95% CI -7.74, -4.72)
fewer points for the Well-Being composite. For both Task and Well-Being, the relationship
with VA was non-linear: for those without visual impairment but slightly worse than normal
vision, VS-QOL was not strongly impacted; however, VS-QOL curves strongly downward
78
for those with visual impairment. We also found that the NEI-VFQ-25 functions best for
those with below-average VS-QOL for both the Task and Well-Being composites. This study
confirms the association between VA and VS-QOL in Chinese Americans and offers guidance
for future use and interpretation of IRT-based NEI-VFQ-25 scores.
79
Visual impairment and blindness are increasingly important areas of public health,
particularly as the US population ages. A 2015 report estimated that 3.22 million US adults
over 40 were visually impaired and that another 1.02 million were blind (8); the same report
estimates that prevalence of visual impairment and blindness may double by 2050. The
economic burden of this public health problem is high. A 2013 report estimated that the
total cost of visual impairment and blindness in the US was $139 billion (91). Additionally,
loss of vision may lead to depression and isolation (5,6). Previous work shows that these
conditions reduce both general health-related quality of life (HRQOL) and vision-specific
quality of life (VS-QOL) (13–19,87). Understanding how this relationship may differ among
racial and ethnic populations is thus critical for setting public health priorities and making
clinical decisions.
While many population-based studies show the association between visual acuity (VA)
and VS-QOL, including the results for Latinos discussed in Chapter 4 and many other multi-
cultural populations in and out of the United States, very few population-based studies have
examinedtheeffectofvisionlossonVS-QOLforChineseindividuals(13–19,87). Population-
based studies in China, Sinapore, and Taiwan demonstrate a loss in general HRQOL associ-
ated with lower vision, while only one population-based has studied the effect of vision loss
on VS-QOL, which showed a loss in vision-related emotional well-being (92–94). Notably,
none of these studies are of Chinese Americans. Our previous work on the prevalence of
visual impairment and its causes in Chinese Americans demonstrated a similar prevalence
to other ethnic groups in the US but a slightly lower prevalence of VI than in China-based
studies (95–101). In CHES, the most frequent causes of VI were cataract, degenerative my-
opia, glaucoma, corneal opacity, macular degeneration, diabetic retinopathy, and other optic
nerve damage (95). Notably, of those with visual impairment, many more were affected by
uncorrected refractive error than in Chinese studies; this finding is important given that a
randomized control trial of the effect of correcting refractive error on VS-QOL showed that
such interventions could increase VS-QOL (22). A standard tool for measuring VS-QOL is
80
the National Eye Institute Visual Function Questionnaire (NEI-VFQ-25) (52,53). Although
researchers designed this tool using classical test theory (CTT), recent research has favored
item response theory approaches that address concerns about the scale’s psychometric prop-
erties (21,54). In Chapter 4, we present a CTT-based approach that addresses some of these
concerns, but item response scores offer additional advantages, such as distinguishing item
performance and importance (46).
The Chinese American Eye Study is a large, comprehensive cross-sectional study of
eye disease among Chinese Americans 50 years or older, the first of its kind (102). In
this analysis, we describe the epidemiology of HRQOL and VS-QOL in Chinese Americans
and assess the impact that visual acuity has on VS-QOL. We also estimate the Task and
Well-Being composites of the NEI-VFQ-25 developed in Chapter 4 using IRT. In addition to
evaluating the effect vision loss has on these scores, we also seek to estimate the item and test
information functions of the NEI-VFQ-25, psychometric properties that have not previously
been reported. To our knowledge, this is the first study of the effect of visual acuity on
VS-QOL in Chinese Americans and the most extensive analysis of the NEI-VFQ-25 in this
population.
4.2 Methods
These data were collected as part of the Chinese American Eye Study (CHES), a
population-based study of eye disease in Chinese individuals living in Monterey Park, Cali-
fornia. The CHES Study Group described data collection and the study design in a previous
report (102). Briefly, eligible participants were 50 years old or over and of self-identified Chi-
nese ancestry living in 10 census tracks in Monterrey Park. Data for CHES were collected
from February 2010 to October 2013, and participants provided informed consent before as-
sessments. Consenting participants completed a home survey and a comprehensive eye exam
81
at a local eye examination center. The institutional review boards for Los Angles County
and the University of Southern California Medical Center approved CHES. All procedures
in CHES adhered to the recommendations of the Declaration of Helsinki as well as current
regulations specified by the Health Insurance Portability and Accountability Act (61).
4.2.1 Sociodemographic and Clinical Data
Eligible and consenting members of the study community participated in interviews
that assessed detailed demographic information, medical and ophthalmic history, access to
health care, insurance status (medical and vision), and level of Western acculturation. Par-
ticipants responded to the Suinn-Lew Asian Self-Identity Acculturation (SL-ASIA) scale,
a tool that measures acculturation to Western culture from 1 (lowest acculturation) to 5
(highest) (103). Based on medical history, we also calculated a composite score for comor-
bidities summarizing the history of 12 medical conditions: diabetes mellitus, arthritis, stroke
or brain hemorrhage, high blood pressure, angina, heart attack, heart failure, asthma, skin
cancer, other types of cancers, back problems, and deafness or hearing problems; this com-
posite score uses a validated approach to reducing the dimensions of comorbidities (64–66).
Study ophthalmologists performed comprehensive eye examinations. In addition to visual
acuity, ophthalmologists assessed the presence of several eye disorders, including glaucoma,
age-related macular degeneration, and diabetic retinopathy.
4.2.2 Visual Acuity
Visual acuity testing procedures in CHES have been described in detail in previous
reports (95,102,104). Briefly, study ophthalmologists measured vision for each eye with
presenting correction, if applicable, at 4 meters using modified Early Treatment Diabetic
82
Retinopathy Study distance charts trans-illuminated with the chart illuminator (Precision
Vision, La Salle, IL). If presenting visual acuity (VA) was 20/20 or better (>=55 letters) in
each eye according to the Early Treatment Diabetic Retinopathy Study protocol (ETDRS),
refraction was the presenting spectacle correction or plano. Otherwise, clinicians used auto-
mated refraction (Carl Zeiss Meditec, Dublin, CA) and subjective refraction. If a participant
was unable to read 20 letters at 4 meters (20/100 Snellen), clinicians instead measured VA at
1 meter. For participants unable to read standard charts, clinicians used an LEA chart. We
calculated VA as the total number of lines read converted into the logarithm of the minimum
angle of resolution (logMAR). We similarly calculated best-corrected VA, the best distance
VA measured via subjective refraction for the participant’s better-seeing eye.
4.2.3 Health-Related and Vision-Specific Quality of Life
4.2.3.1 Medical Outcomes Study 12-Item Short-Form Health Survey and the National Eye
Institute Visual Function Questionnaire
We used the Medical Outcomes Study 12-Item Short-Form Health Survey (SF-12) to
measure general health-related quality of life. The SF-12 has two summary scores: the
Physical Component Summary (PCS) and the Mental Component Summary (MCS), where
higher scores indicate a higher quality of life (70,71). The PCS and MCS are calibrated to
the general United States population. Both summaries are norm-based T-scores, where the
mean is set to 50, and the SD is set to 10.
To assess vision-specific quality of life, we used the NEI-VFQ-25 (52,53). The NEI-
VFQ-25 measures visual functioning related to general health and well-being, as well as
areas more specific to daily functioning related to vision. This scale is typically calculated
using classical test theory (CTT), an approach to psychometric measurement that assumes a
person’sobservedscoreonagivenscaleisequaltotherealunderlyingscoreplusmeasurement
83
error (see Chapter 4 for more details on this calculation) (45,52,53). The NEI-VFQ-25 has
25 items yielding 12 domains: general health, general vision, near and distance vision, ocular
pain, vision-related social function, vision-related role function, vision-related mental health,
vision-related dependency, driving difficulties, color vision, and peripheral vision. The scales
are calculated from 1 to 4 related questions, depending on the domain. We used the standard
scoring method, which produces scales ranging from 0 to 100 (52); higher scores indicate
better visual functioning and quality of life. The traditional composite score averages 11 of
the 12 scale scores (excluding the general health rating question). In Chapter 4, I present
an alternative to the composite score that addresses the potential lack of unidimensionality
(a key assumption in pyschometrics) when reporting a single composite score. We addressed
this concern with the calculation of two separate composite that summarize task-related
and well-being-related vision-specific quality of life (21,54). The task composite averages
the following domains: near vision, distance vision, driving, color vision, peripheral vision,
and vision-related role function. The well-being averages the remaining domains, except
for general health: general vision, vision-related dependency, vision-related mental health,
ocular pain, and vision-related social function.
4.2.3.2 Item Response Model Estimation of Vision-Specific Quality of Life
The NEI-VFQ-25 is typically scored using an approach based on CTT and requires sev-
eral assumptions that may not always be true (45). Several researchers have recently raised
concerns that the NEI-VFQ-25’s composite score may violate these assumptions (21,54).
Firstly, the composite score averages the items of the scale together without differential
weighting, and within each item, responses are also spaced equally (e.g., a valid response for
an item may be 1, 2, 3, or 4). Additionally, using a composite score assumes that the score
is unidimensional. Previous work, as well as the work in Chapter 4, suggests that this may
not be true for the NEI-VFQ-25 (21,54).
84
An alternative way of measuring latent variables is through item response theory (IRT).
Item response theory is a set of models that generalize the Rasch model, a tool developed in
educational assessment to understand how well items perform at measuring ability across its
spectrum and how well they discriminate between levels of ability (46,58). An extension of
IRT is the graded response model (GRM), a two-parameter model closely related to ordinal
logistic regression (105). GRMs are capable of handling polytomous items, such as Likert-
type scales. The two parameters fit by GRMs are difficulty and discrimination; the first
estimates the location of the item along a range of scores, while the second estimates the
shape of the information distribution that the item explains. We used GRMs to fit a model
that estimated VS-QOL scores for the task and well-being composites. Because IRT strongly
assumes unidimensionality, we did not estimate scores for the traditional composite, which
may not meet this assumption. Additionally, we rescaled the item response scores, which
are estimated as standard normal distributions, to range from 0 to 100, as the NEI-VFQ-25
traditionally ranges.
4.2.4 Statistical Analyses
We examined the distribution of health-related and vision-specific quality of life by
presenting visual acuity in the better-seeing eye. We calculated the mean and standard
deviation for each of the NEI-VFQ-25 domains, the traditional CTT composite, and the
IRT-based and CTT-based Task and Well-being composites, as well as the SF-12 PCS and
MCS scores. Additionally, we calculated Spearman’s correlation coefficient between each
score and presenting visual acuity. We computed the confidence intervals for the correlation
coefficient using the bootstrap. We also visualize the distribution of IRT-based Task and
Well-being composites with density plots. We additionally explore the relationship between
VS-QOL and VA through scatter plots using generalized additive models to fit a non-linear
regression line between these variables (106).
85
In addition to estimating item response scores, we also used IRT to estimate item and
test information functions (46,105). Here, item information functions visualize the range
of vision-specific quality of life for which the item is informative. When a discrimination
parameter is fit, as it is in the GRM, then item information functions also show the amount
of information each item provides at the given range. Test information functions similarly
visualize the range of (latent) VS-QOL scores for which the test (the NEI-VFQ-25) functions
well.
We modeled the impact of VA on VS-QOL for the task and well-being composites using
ordinary least squares models (OLS). We modeled VA using restricted cubic splines to allow
non-linearity in the relationship with VS-QOL (59). We then predicted the mean quality of
life scores at 2 lines and 4 lines of visual acuity above (worse than) a logMAR value of 0
(20/20), as well as at 2 line below (better than) a logMAR value of 0. To calculate these
predicted means, we used the parametric G-formula (48). We computed confidence intervals
for the predicted difference in VS-QOL at 2 and 4 lines of visual acuity via the bootstrap.
We selected covariates for these models using a combination of a priori knowledge of
risk factors for VA and VS-QOL and causal structural models (47,75,76). Causal structural
modelsarecommonlydrawnasdirectedacyclicgraphsandusedtodeterminetheminimalset
of covariates required to estimate an effect without bias due to confounding. The minimal set
for this study included age, cataracts, comorbidities, education, employment, acculturation,
and insurance status. We restricted analyses to those with complete data on VA and VS-
QOL, and we imputed missing covariate data with multiple imputation. The imputation
model used chained equations with all other covariates and visual acuity (77,78). In each
OLS and imputation model, we modeled continuous covariates non-linearly using restricted
cubic splines. We calculated predictions by setting continuous covariates to their mean and
categorical covariates to the most frequently occurring category.
We conducted all analyses in R 3.6.3 (79) using the following R packages: ltm (107)
86
(item response models), rms (82) (restricted cubic splines), mice (78) (imputation), dagitty
and ggdag (83,84) (directed acyclic graphs), and ggplot2 (85) (all data visualizations).
4.3 Results
4.3.1 Description of Study Population
This analysis includes 4555 CHES participants with complete information for VA and
QOL. 4578 participants had complete VA data from comprehensive eye exams. Of those, 23
were missing data on well-being-related items for the NEI-VFQ-25 and 14 were missing data
ontaskrelateditems. Theprevalenceofvisualimpairmentinthebetter-seeingeyewas14.3%
for those missing VS-QOL data and 6.5% for those with complete VS-QOL data. However,
the group with missing VS-QOL data was small, with only 23 participants. Those with
complete VA and QOL data had a mean age of 61.8, and 63.2% were female; 50.1% reported
working, but only 11.8% reported having vision insurance; 32.0% of participants reported
having less than high school education; participants had, on average, 1.2 comorbidities.
4.3.2 Health-Related Quality of Life and Visual Acuity
In Table 4.1, we present the overall means of the NEI-VFQ-25 and SF-12 scales. For the
NEI-VFQ-25, we present both the traditional (CTT) scoring and IRT scoring. We also show
the Spearman’s correlation of the scales with presenting visual acuity in the better-seeing
eye and the difference in mean scores for those with 2 or greater lines worse than normal
(20/20 or better) vision. IRT-based Task Composite scores were higher on average than
IRT-based Well-Being Composite scores (78.1 ± 20.6 vs. 63.0 ± 16.7, respectively). This
pattern was also present in the CTT-based scores of these two composites, although less
pronounced (94.2 ± 11.3 vs. 84.4 ± 14.4, respectively). All CTT-based composite scores
87
were high relative to their range (0-100), indicating higher VS-QOL, with the Traditional
composite having a mean and SD of 89.6 ± 12.3. Among the CTT-based subscales, the
General Health domain was lowest (40.6 ± 24.7), and the Color Vision domain the highest
(99.2 ± 5.9).
All scales negatively correlated with rising logMAR values of presenting visual acuity,
suggesting that worse VA was associated with worse quality of life for all scores. The IRT-
based Task and Well-Being Composites had correlations of -0.27 (95% CI -0.30, -0.25) and
-0.21 (95% CI -0.24, -0.18), respectively, while the CTT-based Traditional Composite had
a correlation of -0.23 (95% CI -0.26, -0.20). Notably, for Task and Well-Being Composites,
the IRT-based scores had slightly different correlations from one another, while they were
very similar for the CTT-based scores, likely because they are less flexible than IRT-based
scores. Similarly, the differences in the Task and Well-Being scores for those with 2 or greater
lines worse VA were more pronounced for IRT than CTT (IRT: -12.3 (95% CI -13.7, -10.9)
vs. -7.3 (95% CI -8.4, -6.2), CTT: -6.1 (95% CI -7.0, -5.2) vs. -6.5 (95% CI -7.5, -5.4)).
For the Traditional Composite, the mean difference was -6.5 (95% CI -7.4, -5.5). Among
the subscales, the highest absolute correlation was with the Driving domain (-0.15 (95% CI
-0.18, -0.11)) and the lowest was with the Ocular Pain domain (-0.08 (95% CI -0.11, -0.05)).
The greatest difference by VA was for the Role Function domain (-10.5 (95% CI -12.2, -8.7))
while the smallest was for the Color Vision domain (-1.3 (95% CI -1.8, -0.8)).
In Figure 4.1, we present the distributions of the IRT-based Task and Well-Being Com-
posites, as well as a non-linear regression of their relationship with presenting VA. In general,
Well-Being scores were lower than Task scores. Additionally, the Task score was noticeably
bi-modal, withapotentialceilingeffectatthehighestrangeofthescale, whiletheWell-Being
score was not. Despite this, the IRT scores were less susceptible to ceiling effects than the
CTT-based scores, for which both composites had strong effects (see appendix). Both IRT
and CTT-based scores lowered with worse visual acuity. While the Well-Being composite
was mostly linear, the Task composite had a shallower slope for those with the worse VA
88
compared to lower ranges of logMAR values, although confidence intervals were much wider
for upper ranges of logMAR values.
4.3.3 Item and Test Information Curves
In Figure 4.2, we present the Test Information Functions (TIFs) and Item Information
Functions (IIFs) for the IRT-based Task and Well-Being composites and the items used to
estimate them, respectively. Both composites captured much more information for below-
average (less than 0 logits) VS-QOL than for above-average VS-QOL. Higher information
for below-average scores indicates that the scales are better able to discriminate between
worse VS-QOL scores than for better VS-QOL scores. The Well-being TIF suggests that it
captures slightly more information for those with higher VS-QOL scores than for the Task
composite TIF. For the Task composite, most items better discriminated in this lower range.
Two items related to driving (difficulty driving during the day and at night) provided little
information across the range of the Task composite, although it did provide a small amount
of information for those with the worst Task composite scores. Several items provided
broad information across below-average Task composite scores, while one item (difficulty
going to events) provided much more specific information closer to 0 logits, suggesting it
helps discriminate scores better in this range. For the Well-Being composite, self-rating
overall vision and worrying about eyesight both captured information broadly across the
range of Well-Being composite scores. Difficulty seeing the reactions of others discriminated
well for those with particularly low scores. Needing to rely on the sight of others sharply
discriminated within a narrower range of lower scores. The least informative item appeared
to be difficulty visiting others, although it did provide a small amount of information.
89
4.3.4 Predicted Mean Change in VS-QOL by Visual Acuity
In Figure 4.3a, we present predictions for VS-QOL based on logMAR VA values, cal-
culated with non-linear multivariably-adjusted OLS regression. For both composites, differ-
ences in VS-QOL were modest for those who were not visually impaired (logMAR < 0.3,
20/40). VS-QOL, as measured by both composites, declined steeply for those with visual
impairment. While both scores followed a similar pattern, the Task composite had a higher
intercept than the Well-Being composite, as suggested by the higher mean of the former.
In Figure 4.3b and Table 4.2, we present the predicted mean change in VS-QOL when
comparing participants with normal vision (logMAR = 0, 20/20) to those with 2 lines better
VA, 2 lines worse, and 4 lines worse. For the Task composite, 2 lines of visual acuity better
than normal vision were associated with a slightly better quality of life score of 0.39 points
(95% CI -2.60, 3.52), while participants with 2 lines worse VA had a mean quality of life
score difference of -3.86 (95% CI -4.92, -2.78). The difference in the Task composite at four
lines worse vision was more pronounced with -8.51 fewer points (95% CI -10.28, -6.66). The
Well-Being composite followed a similar pattern but, compared to the Task composite, had
a larger magnitude for better VA and a smaller magnitude for worse VA. For the Well-Being
composite, 2 lines of visual acuity better than normal vision were associated with a better
quality of life score of 3.75 points (95% CI 1.00, 6.50). Participants with 2 lines worse VA
had a mean quality of life score difference of -2.52 (95% CI -3.41, -1.59), and those with 4
lines worse had a mean quality of life score difference of -6.22 (95% CI -7.74, -4.72). CTT-
based estimates of VS-QOL also lowered with worse VA, but the two composites were much
more similar to one another in both the magnitude of the effects and the overall pattern (see
appendix).
90
4.4 Discussion
In this cross-sectional, population-based study of Chinese American adults, the first of
its kind, we found that worse visual acuity was associated with lower vision-specific quality
of life scores. We applied item response models to the calculation of VS-QOL scores. In
multivariably-adjusted models, VS-QOL was lower for those with impaired vision (2 or more
lines from logMar = 0), and the impact of impaired vision on VS-QOL was larger for the
Task composite. The IRT-based Task and Well-Being composites were also different in their
magnitude of effect. For instance, at a clinically meaningful 2-line difference in VA, the
difference in Task scores were, on average, 5 points worse than Well-being scores (12.3 vs
7.3), while CTT-based Task and Well-being scores were closer (6.1 vs 6.5). In IRT-based
scores, the differences between Task and Well-being scores also depended on the participant’s
vision, with Task scores being much lower with worse VA and Well-being scores being better
for those with the best VA, an effect not seen in their CTT-based scores. The relationship
between VA and VS-QOL was non-linear, with the impact on VS-QOL being much worse
for those with visual impairment. In linking the magnitude of difference in IRT-based VS-
QOL scores to 2- and 4-line differences in visual acuity, clinically meaningful level of vision
loss (mild and severe visual impairment, respectively), we sought to give context to the
interpretation of these scores. Thus, a 4-point improvement in Task scores or 2.5-point
improvement in Well-being scores might be interpreted as equivalent to the gains you would
in improving vision by 2 lines. In using IRT to calculate these scores, we found that the
NEI-VFQ-25 operates best for those with below-average VS-QOL for both the Task and
Well-Being composites. The composites did less well for those with higher levels of VS-
QOL. Additionally, several items in both composites added little information.
Our results confirmed the link between vision loss and lower VS-QOL in Chinese Amer-
icans. We found that, at precisely 2 lines of difference, participants had around a 3.9-point
decrease in Task-related quality of life and a 2.5-point decrease in Well-Being-related quality
of life. We used item response theory to calculate the NEI-VFQ-25 composites designed
91
in Chapter 4, a method that may have several benefits, including decreased measurement
error and better statistical power when applied to longitudinal data. However, this scale was
designed using classical test theory and is usually calculated using that method. The exist-
ing literature is based on results presented using CTT, making it difficult to compare our
results to previous studies. As the relationship between VA and VS-QOL is well-established,
however, we chose to examine the effect of VA on VS-QOL at 2 lines of difference, a clinically
significant level of vision loss, to better anchor the interpretation of these scores in future
studies. We suggest that these effect estimates may additionally serve as clinically-important
quantities of change in VS-QOL in both VA and other settings related to clinical decision
making, as they relate to a meaningful change in vision.
The estimates presented here are slightly higher than their CTT-counterparts in the
literature, although they are not necessarily on the same scale. In previous studies using
the traditional composite, estimates hovered at around 2 points difference for exactly 2-lines
change in VA and around 5 points difference for 2-lines or greater change in VA (13–19,87).
Our previous results in the Los Angeles Latino Eye Study closely followed this pattern at
baseline, 4-year, and 8-year (see Chapter 4) follow-up (13,14). While we used the Task and
Well-Being composites, our results here and in Chapter 4 suggest that, for the CTT-based
version of the scale, the traditional composite is an average of the Task and Well-Being
composites, with Well-being scores being slightly lower than Task scores (84.4 vs. 89.6 in
this study). Our results imply that the IRT-based scores better-distinguish between these
two domains, but that the overall pattern is comparable to previous results. While the
Task and Well-being scores, representing distinct latent variables, are difficult to compare
directly to each other, the magnitude in change relative to the baseline mean was higher
for Well-being scores. This may mean that treatments or correction that improve vision
has the most impact on a patient’s well-being. Compared to participants in LALES, CHES
participants who had 2 or more lines worse VA than normal had slightly worse VS-QOL
(using the composite CTT-based score: -5.5 in LALES vs. -6.5 in CHES). There were also
92
differences in the importance of CTT-based domains: In LALES, driving impacted VS-QOL
greatly, while in CHES, it was less important; in CHES, role functioning, mental health, and
dependency were most important (differences in >= 2 lines worse VA ranging from to -10.5
to -8.4 vs. -5.9 for driving) (10,14).
Few studies have been conducted that report on VS-QOL among adults of Chinese
descent, and, toourknowledge, thisisthefirstpopulation-basedstudyofVS-QOLinChinese
Americans. Studies of Chinese individuals in the Singapore Epidemiology of Eye Disease
Study and the Shihpai Eye Study (in Taiwan) have connected reduced visual acuity to
general health-related quality of life (92,93). A study of Singaporean Chinese participants
in the Singapore Chinese Eye Study looked more closely at the effect visual impairment
had on vision-specific emotional well-being using a Rasch variant of the Chinese Impact
of Vision Impairment questionnaire (94,108). Researchers found that emotional well-being
scores decreased as vision impairment worsened. Notably, in our study, the Well-Being score
was lower overall than the Task score. Given that the same researchers found that VS-
QOL mediates visual impairment and depression, this may be a key area of intervention for
Chinese Americans (109). These studies used a different instrument to measure VS-QOL,
making it difficult to contrast results with those presented here; however, one benefit of
the IRT approach is the ability to equate scales. The increasing use of IRT in ophthalmic
epidemiology may allow researchers to better compare results in future work.
While there is less work on Task-based VS-QOL in Chinese participants, previous re-
searchinSingaporeandChinaonrelatedissuessuggestsitisacrucialaspectofhealth. Inthe
Tanjong Pagar Survey of Singaporean Chinese adults, for example, researchers found that
task-related aspects of vision, such as reading the newspaper, reading signs, and navigating
stairs, were all more limited in those with visual impairment compared to those without
(98). Additionally, results from the Singapore Epidemiology of Eye Disease Study suggest
that those with visual impairment have reduced quality-adjusted life-years (92).
93
In our analysis of test and item information functions for the NEI-VFQ-25, we found
that both scales performed much better for lower ranges of VS-QOL. While the Well-Being
composite did provide some information about those with higher VS-QOL, the Task compos-
ite only measured well for those with below-average VS-QOL. As those with low VS-QOL
are the people we are most concerned about, this may not be a serious flaw, but it does
account for the ceiling effect in the IRT-based Task score and both CTT-based composites
(see appendix). Accordingly, it may be appropriate to develop items that capture more
information in these upper regions to distinguish better those with high VS-QOL.
Because we used a two-parameter IRT model—the graded response model—each item
contributed a different amount of maximum information (discrimination). Most items pro-
vided a reasonable amount of information for lower ranges of VS-QOL, with some items, like
difficulty going to events (in the Task composite) and relying on what others say they see
(in the Well-Being composite) were particularly informative. Others, however, performed
less well. The two driving-related items in the Task composite, for example, only captured a
smallamountofinformation. Whileitmaybeappropriatetoremoveitemsbasedonthetheir
discrimination and information functions, as well as other psychometrics, another approach
related to IRT, computerized adaptive testing (CAT), may be useful to researchers (46).
Rather than removing these items, a CAT version of the NEI-VFQ-25 would only present
participants with these items when it seems they have unusually low VS-QOL. CAT-based
scales often require fewer items to measure the underlying trait, and so it may also reduce
participant burden.
A strength of this study is the detailed ophthalmic data recorded from comprehensive
eye exams. CHES is also a large, population-based study of Chinese Americans; it is the
first study to report visual impairment and VS-QOL at the population level for this group.
One limitation of this report is the cross-sectional nature of the data, which prohibited
us from looking at how the longitudinal change in visual acuity affects VS-QOL. While
reverse causality (poor VS-QOL causing poor eyesight) is not a serious concern, the estimand
94
estimated here is slightly different than in longitudinal studies (2-line difference between
patients compared to 2-line change within patients). Notably, though, work in the Los
Angeles Latino Study found that these two estimates were similar in magnitude at baseline
and 4-year data collection; Chapter 4 also shows that this remained consistent at 8-years
of data collection. Longitudinal data should be collected for Chinese Americans to confirm
this.
While utilizing item response theory to estimate VS-QOL is a strength because of how
it may improve measurement, it comes with a few limitations. First, we were not able
to use the traditional composite score in our models because IRT assumes that scales are
unidimensional, an assumption that the traditional composite may not meet. Many previous
studies report a single, CTT-based composite score, making it difficult to compare to the
estimates here. However, we did include the traditional composite in our description of these
data. Secondly, while the IRT and CTT versions of these scales seek to estimate the same
latent variables, they may not be directly comparable. However, we sought to offset these
limitations by linking the IRT scores to a clinically significant marker, 2-lines of difference
in visual acuity: knowing the relationship between these scores and a 2-line difference in
visual acuity may help other studies interpret IRT-based VS-QOL scores better, as well as
to permit more direct comparisons between CTT-based findings in the literature to more
recent IRT based-composite scores. An additional strength of the study is the completeness
of the data. All participants had visual acuity data, and over 99% of participants had
complete VS-QOL data. While those missing VS-QOL data did have higher rates of visual
impairment for their better-seeing eye (14.3% vs. 6.5% in those without and with VS-QOL
data, respectively), this group only represents 23 participants. It is, therefore, unlikely that
missing outcome data had any effect on our results. The same was true of covariate data:
over 99% of participants had data on all covariates included in the model, except for the
presence of cataracts, which was 93% complete. Missing covariate data were also imputed
using multiple imputation and thus did not reduce the sample size for the regression models.
95
In this cross-sectional, population-based study of Chinese American adults, we found
that worsening visual acuity was associated with lower vision-specific quality of life for both
Task and Well-Being-related QOL. We found that, for 2 lines of difference in visual acuity,
participants had around a 3.9-point decrease in Task-related quality of life and a 2.5-point
decrease in Well-Being-related quality of life. We also found that the NEI-VFQ-25 measures
VS-QOLwellforthosewithbelow-averageVS-QOLbutlesswellforthosewithhigherscores.
This study confirms the relationship between vision loss and reduced VS-QOL in Chinese
Americans. Because a substantial number of participants had uncorrected refractive error,
Chinese Americans may also benefit from simple ophthalmic interventions, as correcting
refractive error leads to increased VS-QOL (22,95).
4.5 Acknowledgments
The authors would like to thank the participants of CHES that have generously donated
their time and data to our study.
96
4.6 References
1. Jones GC, Rovner BW, Crews JE, et al. Effects of depressive symptoms on health
behavior practices among older adults with vision loss. Rehabilitation Psychology.
2009;54(2):164–172.
2. Lamoureux EL, Fenwick E, Moore K, et al. Impact of the Severity of Distance
and Near-Vision Impairment on Depression and Vision-Specific Quality of Life in
Older People Living in Residential Care. Investigative Ophthalmology & Visual Sci-
ence [electronicarticle]. 2009;50(9):4103–4109. (https://iovs.arvojournals.org/article.
aspx?articleid=2185912). (Accessed January 26, 2019)
3. Rovner BW, Ganguli M. Depression and Disability Associated with Impaired Vision:
The MoVIES Project. Journal of the American Geriatrics Society [electronic ar-
ticle]. 1998;46(5):617–619. (http://onlinelibrary.wiley.com/doi/abs/10.1111/j.1532-
5415.1998.tb01080.x). (Accessed January 26, 2019)
4. Zhang X, Bullard KM, Cotch MF, et al. Association Between Depression and Func-
tional Vision Loss in Persons 20 Years of Age or Older in the United States, NHANES
2005-2008. JAMA Ophthalmology [electronic article]. 2013;131(5):573–581. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/1660943). (Accessed Jan-
uary 26, 2019)
5. Crews JE, Campbell VA. Vision impairment and hearing loss among community-
dwelling older Americans: Implications for health and functioning. American Journal
of Public Health. 2004;94(5):823–829.
6. Ivers RQ, Norton R, Cumming RG, et al. Visual impairment and risk of hip fracture.
American Journal of Epidemiology. 2000;152(7):633–639.
7. Lee DJ, Gómez-Marín O, Lam BL, et al. Visual acuity impairment and mortality in
US adults. Archives of Ophthalmology (Chicago, Ill.: 1960). 2002;120(11):1544–1550.
97
8. 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 Ophthalmology [electronic article]. 2016;134(7):802–809. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/2523780). (Accessed Jan-
uary 26, 2019)
9. Choudhury F, Varma R, Klein R, et al. Age-Related Macular Degeneration and
Quality of Life in Latinos: The Los Angeles Latino Eye Study. JAMA Ophthalmol-
ogy [electronic article]. 2016;134(6):683–690. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/2514080). (Accessed March 12, 2018)
10. Globe DR, Wu J, Azen SP, et al. The impact of visual impairment on self-reported vi-
sual functioning in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(6):1141–1149. (http://www.aaojournal.org/article/S0161-
6420(04)00303-3/abstract). (Accessed March 19, 2018)
11. Mazhar K, Varma R, Choudhury F, et al. Severity of Diabetic Retinopathy and
Health-Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology
[electronic article]. 2011;118(4):649–655. (http://www.aaojournal.org/article/S0161-
6420(10)00819-5/abstract). (Accessed March 19, 2018)
12. PatinoCM,VarmaR,AzenSP,etal. TheImpactofChangeinVisualFieldonHealth-
Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2011;118(7):1310–1317. (http://www.aaojournal.org/article/s0161-
6420(10)01344-8/abstract). (Accessed March 19, 2018)
98
13. Varma R, Paz SH, Azen SP, et al. The Los Angeles Latino Eye Study: Design,
methods, and baseline data. Ophthalmology [electronic article]. 2004;111(6):1121–
1131. (http://www.aaojournal.org/article/S0161-6420(04)00291-X/abstract). (Ac-
cessed March 19, 2018)
14. McKean-Cowdin R, Varma R, Hays RD, et al. Longitudinal Changes in Visual Acu-
ity and Health-related Quality of Life: The Los Angeles Latino Eye Study. Ophthal-
mology [electronic article]. 2010;117(10):1900–1907.e1. (http://www.aaojournal.org/
article/S0161-6420(10)00132-6/abstract). (Accessed March 19, 2018)
15. Linder M, Chang TS, Scott IU, et al. Validity of the visual function index (vf-
14) in patients with retinal disease. Archives of Ophthalmology [electronic article].
1999;117(12):1611–1616. (http://dx.doi.org/10.1001/archopht.117.12.1611)
16. Mangione CM, Phillips RS, Lawrence MG, et al. Improved Visual Function
and Attenuation of Declines in Health-Related Quality of Life After Cataract
Extraction. Archives of Ophthalmology [electronic article]. 1994;112(11):1419–
1425. (https://jamanetwork.com/journals/jamaophthalmology/fullarticle/640890).
(Accessed March 19, 2018)
17. Health- and vision-related quality of life among patients with choroidal neovascular-
ization secondary to age-related macular degeneration at enrollment in randomized
trials of submacular surgery: SST report no. 4. American Journal of Ophthalmol-
ogy [electronic article]. 2004;138(1):91–108. (http://www.sciencedirect.com/science/
article/pii/S0002939404001370)
18. Parrish RK, Gedde SJ, Scott IU, et al. Visual function and quality of life
among patients with glaucoma. Archives of Ophthalmology (Chicago, Ill.: 1960).
1997;115(11):1447–1455.
99
19. Scott IU, Smiddy WE, Schiffman J, et al. Quality of life of low-vision patients and the
impact of low-vision services. American Journal of Ophthalmology [electronic article].
128(1):54–62. (http://dx.doi.org/10.1016/S0002-9394(99)00108-7). (Accessed March
20, 2018)
20. Healthy people 2020: vision. Healthy People 2020. (https://www.healthypeople.gov/
2020/topics-objectives/topic/vision)
21. Lamoureux E, Pesudovs K. Vision-Specific Quality-of-Life Research: A Need
to Improve the Quality. American Journal of Ophthalmology [electronic arti-
cle]. 2011;151(2):195–197.e2. (https://doi.org/10.1016/j.ajo.2010.09.020). (Accessed
March 18, 2020)
22. Coleman AL, Yu F, Keeler E, et al. Treatment of Uncorrected Refractive Error
Improves Vision-Specific Quality of Life. Journal of the American Geriatrics Society
[electronic article]. 2006;54(6):883–890. (http://onlinelibrary.wiley.com/doi/abs/10.
1111/j.1532-5415.2006.00817.x). (Accessed March 23, 2020)
23. BourneRR,StevensGA,WhiteRA,etal. Causesofvisionlossworldwide, 1990–2010:
A systematic analysis. The lancet global health. 2013;1(6):e339–e349.
24. Varma R, Chung J, Foong AWP, et al. Four-Year Incidence and Progression of Vi-
sual Impairment in Latinos: The Los Angeles Latino Eye Study (LALES). Amer-
ican journal of ophthalmology [electronic article]. 2010;149(5):713–727. (https:
//www.ncbi.nlm.nih.gov/pmc/articles/PMC2919324/). (Accessed January 26, 2019)
25. Varma R, Kim JS, Burkemper BS, et al. Prevalence and causes of visual impairment
and blindness in chinese american adults: The chinese american eye study. Jama
Ophthalmology. 2016;134(7):785–793.
100
26. Munoz B, West SK, Rubin GS, et al. Causes of blindness and visual impairment
in a population of older americans: The salisbury eye evaluation study. Archives of
ophthalmology. 2000;118(6):819–825.
27. Tang Y, Wang X, Wang J, et al. Prevalence and causes of visual impairment in a
chinese adult population: The taizhou eye study. Ophthalmology. 2015;122(7):1480–
1488.
28. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindness in a rural chinese adult population: The handan eye study. Ophthalmology.
2008;115(11):1965–1972.
29. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in beijing: The beijing eye study. Ophthalmology. 2006;113(7):1134–e1.
30. ZhaoJ,XuX,EllweinLB,etal. Causesofvisualimpairmentandblindnessinthe2006
and 2014 nine-province surveys in rural china. American Journal of Ophthalmology.
2019;197:80–87.
31. AtteboK,MitchellP,SmithW.Visualacuityandthecausesofvisuallossinaustralia:
The blue mountains eye study. Ophthalmology. 1996;103(3):357–364.
32. Hyman L, Wu S-Y, Connell AM, et al. Prevalence and causes of visual impairment
in the barbados eye study. Ophthalmology. 2001;108(10):1751–1756.
33. Klaver CC, Wolfs RC, Vingerling JR, et al. Age-specific prevalence and causes
of blindness and visual impairment in an older population: The rotterdam study.
Archives of ophthalmology. 1998;116(5):653–658.
34. Rodriguez J, Sanchez R, Munoz B, et al. Causes of blindness and visual impairment
in a population-based sample of US hispanics. Ophthalmology. 2002;109(4):737–743.
101
35. CongdonN,O’ColmainB,KlaverC,etal. Causesandprevalenceofvisualimpairment
among adults in the united states. Archives of Ophthalmology (Chicago, Ill.: 1960).
2004;122(4):477–485.
36. VanNewkirk MR, Weih L, McCarty CA, et al. Cause-specific prevalence of bilateral
visual impairment in victoria, australia: The visual impairment project. Ophthalmol-
ogy. 2001;108(5):960–967.
37. 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. Investigative Ophthalmology & Visual Science. 2002;43(11):3393–
3398.
38. Unzueta M, Globe D, Wu J, et al. Compliance with recommendations for follow-up
care in latinos: The los angeles latino eye study. Ethnicity & disease. 2004;14:285–91.
39. Varma R, Mohanty SA, Deneen J, et al. Burden and predictors of undetected eye
disease in mexican-americans. Medical care. 2008;46(5):497–506.
40. Patty LE, Morales LS, Lai M, et al. Self-reported utilization of eye care among latinos
in the los angeles latino eye study (LALES). Investigative Ophthalmology & Visual
Science. 2009;50(13):2502–2502.
41. Varma R, Torres M, McKean-Cowdin R, et al. Prevalence and risk factors for re-
fractive error in adult chinese americans: The chinese american eye study. American
Journal of Ophthalmology. 2017;175:201–212.
42. Jiang X, Varma R, Torres M, et al. Self-reported use of eye care among adult chinese
americans in the chinese american eye study. American journal of ophthalmology.
2017;176:183–193.
102
43. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by hispanics: Findings from HHANES 1982-84. American Journal of
Public Health. 1990;80(Suppl):11–19.
44. Zheng Y, Lamoureux EL, Ikram MK, et al. Impact of migration and acculturation
on prevalence of type 2 diabetes and related eye complications in indians living in a
newly urbanised society. PloS One. 2012;7(4):e34829.
45. Streiner DL, Cairney J, Norman GR. Health measurement scales: A practical guide
to their development and use. Oxford Univ. Press; 2015.
46. Ayala RJde. The theory and practice of item response theory. Guildford Press; 2009.
47. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Medical
Research Methodology [electronic article]. 2008;8:70. (https://www.ncbi.nlm.nih.gov/
pmc/articles/PMC2601045/). (Accessed November 21, 2018)
48. Hernán MA, Robins JM. Causal Inference. Boca Raton: Chapman Hall/CRC; 2019.
49. Steinberg EP, Tielsch JM, Schein OD, et al. The VF-14: An Index of Functional
ImpairmentinPatientsWithCataract. Archives of Ophthalmology [electronicarticle].
1994;112(5):630–638. (https://doi.org/10.1001/archopht.1994.01090170074026)
50. Mangione CM, Phillips RS, Seddon JM, et al. Development of the ’Activities of Daily
Vision Scale’. A measure of visual functional status. Medical Care. 1992;30(12):1111–
1126.
51. Margolis MK, Coyne K, Kennedy-Martin T, et al. Vision-specific instruments for the
assessment of health-related quality of life and visual functioning: A literature review.
PharmacoEconomics. 2002;20(12):791–812.
103
52. Mangione CM, Lee PP, Gutierrez PR, et al. Development of the 25-list-item
National Eye Institute Visual Function Questionnaire. Archives of Ophthalmol-
ogy [electronic article]. 2001;119(7):1050–1058. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/267134). (Accessed March 19, 2018)
53. Mangione CM, Lee PP, Pitts J, et al. Psychometric Properties of the National
Eye Institute Visual Function Questionnaire (NEI-VFQ). Archives of Ophthalmology
[electronic article]. 1998;116(11):1496–1504. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/264225). (Accessed March 19, 2018)
54. Pesudovs K, Gothwal VK, Wright T, et al. Remediating serious flaws in the Na-
tional Eye Institute Visual Function Questionnaire. Journal of Cataract & Refrac-
tive Surgery [electronicarticle]. 2010;36(5):718–732. (http://linkinghub.elsevier.com/
retrieve/pii/S0886335010002531). (Accessed March 12, 2018)
55. Kline RB. Principles and practice of structural equation modeling. The Guilford
Press; 2016.
56. Bentler PM. On tests and indices for evaluating structural models. Personal-
ity and Individual Differences [electronic article]. 2007;42(5):825–829. (http://
www.sciencedirect.com/science/article/pii/S0191886906003898). (Accessed March
18, 2020)
57. Brucker PS, Yost K, Cashy J, et al. General population and cancer pa-
tient norms for the functional assessment of cancer therapy-general (FACT-
g). Evaluation & the Health Professions [electronic article]. 2005;28(2):192–
211. (%20%0A%20%20%20%20%20%20%20%20https://doi.org/10.1177/
0163278705275341%0A%20%20%20%20%0A)
104
58. Wright BD. Solving MEASUREMENT PROBLEMS WITH THE Rasch MODEL.
Journal of Educational Measurement [electronic article]. 1977;14(2):97–116. (https:
//onlinelibrary.wiley.com/doi/abs/10.1111/j.1745-3984.1977.tb00031.x)
59. Harrell F. Regression modeling strategies: With applications to linear models, logistic
and ordinal regression, and survival analysis. springer; 2016.
60. Medeiros FA, Gracitelli CPB, Boer ER, et al. Longitudinal Changes in Quality of
Life and Rates of Progressive Visual Field Loss in Glaucoma Patients. Ophthalmol-
ogy [electronic article]. 122(2):293–301. (http://dx.doi.org/10.1016/j.ophtha.2014.08.
014). (Accessed March 20, 2018)
61. Association WM. World Medical Association Declaration of Helsinki: Ethical Prin-
ciples for Medical Research Involving Human Subjects. JAMA [electronic article].
2013;310(20):2191–2194. (https://doi.org/10.1001/jama.2013.281053)
62. GlobeDR,Schoua-GlusbergA,PazS,etal. Usingfocusgroupstodevelopaculturally
sensitive methodology for epidemiological surveys in a Latino population: Findings
from the Los Angeles Latino Eye Study (LALES). Ethnicity & disease [electronic
article]. 2002;12(2):259–266. (http://europepmc.org/abstract/med/12019936). (Ac-
cessed March 19, 2018)
63. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by Hispanics: Findings from HHANES 1982-84. American Journal
of Public Health [electronic article]. 1990;80(Suppl):11–19. (https://doi.org/10.2105/
AJPH.80.Suppl.11). (Accessed March 20, 2018)
64. Brody BL, Gamst AC, Williams RA, et al. Depression, visual acuity, comorbidity,
and disability associated with age-related macular degeneration. Ophthalmology [elec-
tronicarticle]. 108(10):1893–1900. (http://dx.doi.org/10.1016/S0161-6420(01)00754-
0). (Accessed March 20, 2018)
105
65. Globe DR, Varma R, Torres M, et al. Self-reported comorbidities and visual function
inapopulation-basedstudy: Thelosangeleslatinoeyestudy. Archives of Ophthalmol-
ogy [electronic article]. 2005;123(6):815–821. (http://dx.doi.org/10.1001/archopht.
123.6.815)
66. Linn Bernard S., Linn Margaret W., Gurel Lee. Cumulative illness rating scale.
Journal of the American Geriatrics Society [electronic article]. 2015;16(5):622–626.
(https://onlinelibrary.wiley.com/doi/full/10.1111/j.1532-5415.1968.tb02103.x). (Ac-
cessed March 19, 2018)
67. Azen SP, Varma R, Preston-Martin S, et al. Binocular Visual Acuity Sum-
mation and Inhibition in an Ocular Epidemiological Study: The Los Angeles
Latino Eye Study. Investigative Ophthalmology & Visual Science [electronic ar-
ticle]. 2002;43(6):1742–1748. (http://iovs.arvojournals.org/article.aspx?articleid=
2123581). (Accessed March 19, 2018)
68. Varma R, Fraser-Bell S, Tan S, et al. Prevalence of age-related macular degen-
eration in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(7):1288–1297. (http://www.aaojournal.org/article/S0161-
6420(04)00313-6/abstract). (Accessed March 19, 2018)
69. Varma R, Torres M, Peña F, et al. Prevalence of diabetic retinopathy in adult
Latinos: The Los Angeles Latino Eye Study. Ophthalmology [electronic article].
2004;111(7):1298–1306. (http://www.aaojournal.org/article/S0161-6420(04)00336-
7/abstract). (Accessed March 19, 2018)
106
70. Ware JE, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: Construction
ofScalesandPreliminaryTestsofReliabilityandValidity. Medical Care [electronicar-
ticle]. 1996;34(3):220. (https://journals.lww.com/lww-medicalcare/Abstract/1996/
03000/A_12_Item_Short_Form_Health_Survey__Construction.3.aspx). (Ac-
cessed March 19, 2018)
71. Ware J, A. Kosinski M, D. Keller S. SF-12: How to Score the SF-12 Physical and
Mental Health Summary Scales. 1998;
72. Rosenthal R. Parametric measures of effect size. In: The handbook of research syn-
thesis. New York, NY, US: Russell Sage Foundation; 1994:231–244.
73. Cohen J. A power primer. Psychological bulletin [electronic article]. 1992;112(1):155–
159. (http://libproxy.usc.edu/login?url=https://search-proquest-com.libproxy2.usc.
edu/docview/614317877?accountid=14749)
74. Hays RD, Brodsky M, Johnston MF, et al. Evaluating the Statistical Significance
of Health-Related Quality-Of-Life Change in Individual Patients. Evaluation & the
Health Professions [electronic article]. 2005;28(2):160–171. (https://doi.org/10.1177/
0163278705275339). (Accessed March 19, 2018)
75. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epi-
demiology (Cambridge, Mass.). 1999;10(1):37–48.
76. Greenland S. Quantifying biases in causal models: Classical confounding vs collider-
stratification bias. Epidemiology (Cambridge, Mass.). 2003;14(3):300–306.
77. Buuren S van. Flexible imputation of missing data. CRC Press; 2018.
107
78. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained
equations in r. Journal of Statistical Software [electronic article]. 2011;45(3):1–67.
(https://www.jstatsoft.org/v45/i03/)
79. R Core Team. R: A language and environment for statistical computing. Vienna,
Austria: RFoundationforStatisticalComputing; 2017.(https://www.R-project.org/)
80. Rosseel Y. lavaan: An R package for structural equation modeling. Journal of Sta-
tistical Software [electronic article]. 2012;48(2):1–36. (http://www.jstatsoft.org/v48/
i02/)
81. Epskamp S, Simon Stuber with contributions from. semPlot: Path diagrams and
visual analysis of various SEM packages’ output. 2017.(https://CRAN.R-project.
org/package=semPlot)
82. Harrell Jr FE. Rms: Regression modeling strategies. 2018.(https://CRAN.R-project.
org/package=rms)
83. Textor J, van der Zander B. Dagitty: Graphical analysis of structural causal models.
2016.
84. Barrett M. Ggdag: Analyze and create elegant directed acyclic graphs. 2018.
85. Wickham H. ggplot2: Elegant graphics for data analysis. Springer-Verlag New York;
2016.(http://ggplot2.org)
86. Alqudah A, Mansberger SL, Gardiner SK, et al. Vision-related quality of life in glau-
coma suspect or early glaucoma patients. Journal of glaucoma [electronic article].
2016;25(8):629–633. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4975427/).
(Accessed March 12, 2018)
108
87. Fenwick EK, Man RK, Ong P, et al. Association of changes in visual acuity
with vision-specific functioning in the singapore malay eye study. JAMA Oph-
thalmology [electronic article]. 2016;134(11):1299–1305. (http://dx.doi.org/10.1001/
jamaophthalmol.2016.3692)
88. Responsiveness of the National Eye Institute Visual Function Questionnaire
to Changes in Visual Acuity. Archives of ophthalmology [electronic article].
2003;121(4):531–539. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1434793/).
(Accessed March 19, 2018)
89. Lindblad AS, Clemons TE. Responsiveness of the National Eye Institute Visual Func-
tion Questionnaire to progression to advanced age-related macular degeneration, vi-
sion loss, and lens opacity: AREDS Report no. 14. Archives of Ophthalmology
(Chicago, Ill.: 1960). 2005;123(9):1207–1214.
90. Hays RD, Morales LS, Reise SP. Item Response Theory and Health Outcomes Mea-
surementin the21stCentury. Medical care [electronicarticle]. 2000;38(9Suppl):II28–
II42. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1815384/). (Accessed
March 19, 2018)
91. Wittenborn J, Rein D. The economic burden of vision loss and eye disorders in the
United States. University of Chicago; 2013.(https://www.preventblindness.org/sites/
default/files/national/documents/Economic%20Burden%20of%20Vision%20Final%
20Report_130611.pdf)
92. Wang X, Lamoureux E, Zheng Y, et al. Health Burden Associated with Visual Im-
pairment in Singapore: The Singapore Epidemiology of Eye Disease Study. Ophthal-
mology [electronic article]. 2014;121(9):1837–1842. (http://www.sciencedirect.com/
science/article/pii/S0161642014002413). (Accessed March 23, 2020)
109
93. Tsai S-Y, Chi L-Y, Cheng C-Y, et al. The Impact of Visual Impairment and Use of
Eye Services on Health-Related Quality of Life among the Elderly in Taiwan: The
Shihpai Eye Study. Quality of Life Research [electronic article]. 2004;13(8):1415–
1424. (https://doi.org/10.1023/B:QURE.0000040791.87602.fe). (Accessed March 23,
2020)
94. Vision impairment and major eye diseases reduce vision-specific emotional well-being
in a Chinese population - ProQuest. (https://search-proquest-com.libproxy1.usc.
edu/docview/1894846003?rfr_id=info%3Axri%2Fsid%3Aprimo). (Accessed March
23, 2020)
95. Varma R, Kim JS, Burkemper BS, et al. Prevalence and Causes of Visual Impairment
andBlindnessinChineseAmericanAdults: TheChineseAmericanEyeStudy. JAMA
Ophthalmology [electronic article]. 2016;134(7):785–793. (https://jamanetwork-
com.libproxy1.usc.edu/journals/jamaophthalmology/fullarticle/2523778). (Accessed
March 21, 2020)
96. Hsu W-M, Cheng C-Y, Liu J-H, et al. Prevalence and causes of visual impairment
in an elderly Chinese population in Taiwan: The Shihpai Eye Study. Ophthalmology.
2004;111(1):62–69.
97. Huang S, Zheng Y, Foster PJ, et al. Prevalence and causes of visual impairment in
Chinese adults in urban southern China. Archives of Ophthalmology (Chicago, Ill.:
1960). 2009;127(10):1362–1367.
98. Saw S-M, Foster PJ, Gazzard G, et al. Causes of blindness, low vision, and
questionnaire-assessed poor visual function in Singaporean Chinese adults: The Tan-
jong Pagar Survey. Ophthalmology. 2004;111(6):1161–1168.
110
99. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindnessinaruralchineseadultpopulation: TheHandanEyeStudy. Ophthalmology.
2008;115(11):1965–1972.
100. Zhao J, Ellwein LB, Cui H, et al. Prevalence of vision impairment in older adults in
rural China: The China Nine-Province Survey. Ophthalmology. 2010;117(3):409–416,
416.e1.
101. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in Beijing: The Beijing Eye Study. Ophthalmology. 2006;113(7):1134.e1–
11.
102. Varma R, Hsu C, Wang D, et al. The Chinese American Eye Study: Design and
Methods. Ophthalmic Epidemiology [electronic article]. 2013;20(6):335–347. (http:
//www.tandfonline.com/doi/full/10.3109/09286586.2013.823505). (Accessed March
21, 2020)
103. SuinnRM,AhunaC,KhooG.TheSuinn-LewAsianSelf-IdentityAcculturationScale:
Concurrent and factorial validation. Educational and Psychological Measurement.
1992;52(4):1041–1046.
104. Burkemper B, Torres M, Jiang X, et al. Factors Associated with Visual Impairment in
Chinese American Adults: The Chinese American Eye Study. Ophthalmic Epidemiol-
ogy [electronic article]. 2019;26(5):329–335. (https://doi.org/10.1080/09286586.2019.
1622737). (Accessed March 22, 2020)
105. Linden WJ van der. Handbook of item response theory. Chapman & Hall/CRC;
2019.
106. Wood SN. Generalized additive models: An introduction with r. CRC Press/Taylor
& Francis Group; 2017.
111
107. Rizopoulos D. Ltm: An r package for latent variable modelling and item response
theory analyses. Journal of Statistical Software [electronic article]. 2006;17(5):1–25.
(http://www.jstatsoft.org/v17/i05/)
108. Fenwick EK, Ong PG, Sabanayagam C, et al. Assessment of the psychometric proper-
ties of the Chinese Impact of Vision Impairment questionnaire in a population-based
study: FindingsfromtheSingaporeChineseEyeStudy. Qualityof LifeResearch [elec-
tronic article]. 2016;25(4):871–880. (https://doi.org/10.1007/s11136-015-1141-1).
(Accessed March 23, 2020)
109. Rees G, Tee HW, Marella M, et al. Vision-specific distress and depressive symptoms
in people with vision impairment. Investigative Ophthalmology & Visual Science.
2010;51(6):2891–2896.
110. Morales LS, Varma R, Paz SH, et al. Self-reported use of eye care among Latinos:
The Los Angeles Latino Eye Study. Ophthalmology. 2010;117(2):207–215.e1.
111. Chang W, Cheng J, Allaire J, et al. Shiny: Web application framework for r.
2020.(https://CRAN.R-project.org/package=shiny)
112
4.7 Tables and Figures
Table 4.1: Overall Correlation and Difference in Visual Acuity and QOL
Variable Mean
(SD)
Spearman’s Correlation
(95% CI)
Difference >= 2 Lines Worse
(95% CI)
NEI-VFQ-25 Composites (IRT)
Task Composite 78.1
(20.6)
-0.27 (-0.30, -0.25) -12.3 (-13.7, -10.9)
Well-Being Composite 63.0
(16.7)
-0.21 (-0.24, -0.18) -7.3 (-8.4, -6.2)
NEI-VFQ-25 Composites (CTT)
Well-Being Composite 84.4
(14.4)
-0.19 (-0.22, -0.16) -6.5 (-7.5, -5.4)
Traditional Composite 89.6
(12.3)
-0.23 (-0.26, -0.20) -6.5 (-7.4, -5.5)
Task Composite 94.2
(11.3)
-0.23 (-0.26, -0.21) -6.1 (-7.0, -5.2)
NEI-VFQ-25 Subscales (CTT)
Role Function 87.3
(23.7)
-0.20 (-0.23, -0.17) -10.5 (-12.2, -8.7)
Mental Health 83.3
(25.0)
-0.17 (-0.21, -0.14) -10.0 (-11.7, -8.2)
Dependency 92.8
(21.1)
-0.17 (-0.21, -0.14) -8.4 (-10.0, -6.7)
Near Vision 93.5
(14.4)
-0.19 (-0.22, -0.16) -6.6 (-7.8, -5.5)
General Health 40.6
(24.7)
-0.14 (-0.17, -0.10) -6.3 (-7.9, -4.8)
General Vision 62.3
(14.5)
-0.22 (-0.24, -0.19) -6.2 (-7.2, -5.2)
Distance Vision 95.6
(12.4)
-0.23 (-0.25, -0.20) -6.1 (-7.2, -5.1)
Driving 93.4
(16.4)
-0.15 (-0.18, -0.11) -5.9 (-7.8, -4.1)
113
Variable Mean
(SD)
Spearman’s Correlation
(95% CI)
Difference >= 2 Lines Worse
(95% CI)
Ocular Pain 84.7
(21.5)
-0.08 (-0.11, -0.05) -5.0 (-6.5, -3.5)
Peripheral Vision 97.9 (9.3) -0.18 (-0.20, -0.15) -3.4 (-4.2, -2.7)
Social Function 98.8 (7.2) -0.18 (-0.21, -0.14) -2.7 (-3.3, -2.0)
Color Vision 99.2 (5.9) -0.10 (-0.12, -0.07) -1.3 (-1.8, -0.8)
SF-12 Scales
PCS (SF-12) 49.1 (9.3) -0.18 (-0.21, -0.15) -3.7 (-4.4, -3.0)
MCS (SF-12) 55.8
(10.1)
-0.09 (-0.12, -0.06) -2.2 (-2.9, -1.5)
N for each scale range from 3225 to 4564. Scores labeled IRT were calculated using graded response models and re-scaled
from 0 to 100. Scores labeled CTT were calculated using standard NEI-VFQ-25 scoring formulas. Mean values are for overall
scale means for those with complete data required to calculate the score. Spearman’s Correlations were with logMAR values of
presenting visual acuity in the better-seeing eye; 95% CIs were calculated using the bootstrap. Difference for >= 2 lines worse
were the difference and 95% confidence interval for those with 2 or greater lines more than 20/20 (logMAR value of 0). Scales
are arranged by the descending magnitude of this value within each heading. NEI-VFQ-25 = National Eye Institute Visual
Function Questionnaire. SF-12 = Short Form 12. PCS = Physical Health Composite. MCS = Mental Health Composite.
114
Figure 4.1: Distribution of Vision-Specific Quality of Life and Relationship with Visual Acuity
A: Density plot of Task and Well-Being composites of Vision-Specific Quality of Life. x-axis represents VS-QOL and
y-axis represents density of VS-QOL values among CHES participants. B: Smoothed general additive models of the unadjusted
relationship between visual acuity and VS-QOL. x-axis represents the presenting VA in the participant’s better-seeing eye in
logMAR units. Participants with visual acuity over 0.3 logMAR units are visually impaired. y-axis represents VS-QOL values.
115
A: Test Information Function (TIF) of the Task and Well-Being Composites. x-axis represents VS-QOL, expressed
in logits, and y-axis represents information captured by the scale. Higher values of information mean that the scale better
discriminates values of VS-QOL within that range of VS-QOL. 0 logits signify average VS-QOL, and higher values mean higher
VS-QOL B: Item Information Functions (IIF) of each item in the Task and Well-Being Composites. x-axis represents VS-QOL,
expressed in logits, and y-axis represents information captured by the item. Higher values of information mean that the item
better discriminates values of VS-QOL within that range of VS-QOL. Arranged by the value of VS-QOL for which the item
captures the maximum amount of information.
116
Figure 4.2: Test and Item Information Functions of the Task and Well-Being Composites
117
Table 4.2: Predicted Mean Difference of VS-QOL at Levels of Visual Acuity Difference
Vision Better than 20/20 Vision Worse than 20/20
Two Lines Two Lines Four Lines
Difference (95% CI) Difference (95% CI) Difference (95% CI)
Task Composite 0.39 (-2.60, 3.52) -3.86 (-4.92, -2.78) -8.51 (-10.28, -6.66)
Well-being Composite 3.75 (1.00, 6.50) -2.52 (-3.41, -1.59) -6.22 (-7.74, -4.72)
N = 4564 for Task models and N = 4555 for Well-Being models. Outcomes calculated using graded response models
(IRT). Fit with ordinary least squares regression and calculated with the parametric G-formula. Estimates are the difference
in mean vision-specific quality of life at 2 or 4 lines of presenting visual acuity (better-seeing eye) better and worse than 20/20
compared to 20/20 (logMAR value of 0). Adjusted for age, cataracts, comorbidities, education, employment, acculturation,
and insurance. Covariates imputed using multiple imputation. CI = confidence intervals.
118
Figure 4.3: Difference in Vision-Related Quality of Life and Visual Acuity by IRT-based Composite
N = 4564 for Task models and N = 4555 for Well-Being models. Outcomes calculated using graded response models
(IRT). Fit with ordinary least squares regression and calculated with the parametric G-formula. Adjusted for age, cataracts,
comorbidities, education, employment, acculturation, and insurance. Covariates imputed using multiple imputation. A: Pre-
dicted mean VS-QOL over a range of VA and 95% confidence interval. B: Parametric G-formula point estimates of differences
in VS-QOL at exact points of VA. Estimates are the difference in mean vision-specific quality of life at 2 or 4 lines of presenting
visual acuity (better-seeing eye) better and worse than 20/20 compared to 20/20 (logMAR value of 0). VA: Visual Acuity.
VS-QOL: Vision-Specific Quality of Life.
119
CHAPTER 5
SUMMARY OF FINDINGS AND FUTURE DIRECTIONS
5.1 Summary
Vision loss remains a serious public health concern, and its effect on quality of life is
one of the ways it most meaningfully impacts the aging population in the US. The shifting
demographics of the US require that we understand the impact of vision loss in large popu-
lations, such as Latinos and Chinese Americans. In this dissertation, I sought to contribute
to our understanding of the epidemiology of central vision loss and vision-specific quality
of life (VS-QOL), as well as to improve the measurement of VS-QOL. The primary aims
of this dissertation were thus to describe and model the epidemiology of vision loss and
VS-QOL at 8 years of follow-up in Latinos and baseline for Chinese Americans; to develop
a new approach to using the NEI-VFQ-25 in measuring VS-QOL that better represents its
multidimensionality; and to analyze and score the NEI-VFQ-25 from the perspective of item
response theory to improve its use and measurement.
Chapter 4 presents a longitudinal analysis of VA and VS-QOL in the Los Angeles Latino
Eye Study. At 8 years of follow-up, we found that Latino adults with a clinically relevant
change in visual acuity also saw their VS-QOL change. We found that scoring the NEI-VFQ-
25 tool to measure vision-specific Well-Being and Task-related quality of life fit the data
better than the traditional approach to scoring the NEI-VFQ-25. As this approach treats
VS-QOL as having two primary dimensions, these scores also behaved slightly differently.
A 2 line change in visual acuity led to about a 2.5 point increase in both Task-related and
Well-Being-related quality of life, a 2 line loss in VA caused about a 1 point loss in Well-
Being-related quality of life as opposed to a 2 point loss in Task-related quality of life. In
this study, the traditional approach appeared to average the effects of the two dimensions.
120
Chapter 5 presents a cross-sectional analysis of VA and VS-QOL in the Chinese Amer-
ican Eye Study. At baseline data collection, we found that Chinese American adults with
worsening visual acuity also had lower vision-specific quality of life for both Task and Well-
Being-related QOL. For 2 lines of difference in visual acuity, participants had around a 3.9-
point decrease in Task-related quality of life and a 2.5-point decrease in Well-Being-related
quality of life. In adjusted models, VS-QOL decreased slightly for those with little to mild
visual impairment and more sharply for those with more severe impairment, particularly
in the Task domain. We also found that the IRT-based Task and Well-Being composites
were different in their magnitude of effect. Task scores were lower with worse VA compared
to Well-being scores, which themselves were better for those with the best VA. We linked
the magnitude of difference in VS-QOL to 2 lines difference in visual acuity, a clinically
meaningful level of vision loss, to help future interpretations of these scores. In using IRT to
calculate these scores, we found that the NEI-VFQ-25 operates best for those with below-
average VS-QOL for both the Task and Well-Being composites. The composites did not
capture as much information for those with higher levels of VS-QOL. We also found several
items that performed poorly and suggested that computerized adaptive testing could reduce
the number of items answered by participants.
5.2 Future Directions and Public Health Impacts
Vision loss remains a significant public health problem. Understanding how it affects
VS-QOLisessentialforbothdirectingourlimitedpublichealthresourcesandmakingclinical
decisions. There are several areas of future research related to the work in this dissertation
that may benefit researchers, clinicians, and patients. Here, I briefly describe how we may
contribute further to epidemiologic and methodological questions in the area of vision loss
and VS-QOL.
121
While the relationship between vision loss and VS-QOL is now well established, there
remain several epidemiologic questions. In Chapter 5, we present cross-sectional results of
thisrelationshipinChineseAmericans; wefoundthatvisionlossaffectedthisgroupatsimilar
rates to others. There are very few population-based studies of people of Chinese descent
that have examined this question, and, including ours, all have been cross-sectional (94,102).
The state of the research indicates a need for longitudinal data in China, Singapore, the US,
and other countries with large Chinese populations. While results from the Los Angeles
Latino Eyes study suggest that baseline estimates may be similar to longitudinal estimates,
this needs to be confirmed, particularly in this important subgroup of Americans (13,14).
Research in Latinos has been more extensive, thanks in large part to LALES, but other
questions have not been examined related to vision loss and VS-QOL, such as the impact
of other aspects of vision like contrast sensitivity. Additionally, recent research in Singapore
suggests that VS-QOL may mediate the relationship between vision loss and depression
(109); both this mediated relationship and the potential impact of ophthalmic interventions
should be investigated in these and other populations.
Related to the remaining epidemiologic questions are those concerning how we may
address this issue from a public health perspective. Research supports that ophthalmic
interventions, such as cataract surgery and refractive error correction, are what improve
VS-QOL (22,86). In CHES, for example, many participants had particularly high rates of
uncorrected refractive error, an issue that is often easy to address with corrective lenses and
other treatments. Because treatment of diseases is such an important driver of VS-QOL,
the study of incident VI due to ophtlamic disease remains an important topic in improving
VS-QOL for both epidemiologic studies and randomized controlled trials. A related issue is
lack of access to care, which contributes to VS-QOL by worsening vision-related outcomes
(110). An important issue here is the lack of wide-spread access to vision insurance. Many
treatments for vision loss, such as corrective lenses, are easy wins in terms of improving
VS-QOL, and policies that support better access to vision insurance may have an important
122
role to play (22). More research is thus needed in how interventions like mobile vision clinics
and policy changes for access to vision care might impact VS-QOL. An additional area that
may be valuable is incorporating VS-QOL measurement into clinical practice to help make
decisions about treatment.
A key issue related to lack of access to care is acculturation; a primary way accultur-
ation affects VS-QOL is through treatment (38–42). Acculturation is multi-dimensional,
but the aspect that seems to be most related to access to care is language (43). From a
causal perspective, language mediates the effect of acculturation on treatment. Addressing
these questions statistically identifies practical ways to increase care access, such as bilingual
providers and language services. An additional key aspect of this mediator is familiarity with
the US insurance system, which may be more fragmented or stratified compared to a par-
ticipant’s home country. Notably, a study that addresses the effect of acculturation on eye
disease treatment can easily extend to VS-QOL. Both RCTs and observational mediation
studies may identify the effect of acculturation and improved language services on VS-QOL.
Several methodological issues remain relevant. In this dissertation, I use a novel ap-
proach to addressing some of the concerns about the NEI-VFQ-25, an instrument meant
to measure VS-QOL. We developed separate Task and Well-Being composites that fit the
data slightly better than the single dimension composite traditionally used to summarize
scores. In Chapter 4, we present these scales as structural equation models and analyze
them using a classical approach. In Chapter 5, we used item response theory to analyze
these scores. While the two separate composites functioned slightly better in our data, these
results need to be confirmed. Other aspects of the dimensionality of the NEI-VFQ-25 need
to be investigated, as well. In Chapter 4, using the 12 subscales of the NEI-VFQ-25 fit the
data best. It is unclear why this may be. One explanation is that 2 dimensions (Task and
Well-Being) are still not sufficiently separate to model well. Another explanation is that
the 12 subscale model fits a separate but related latent variable, such as symptom clusters.
These subscales and their component items may be easy to understand for patients, as well,
123
and so examining them at this close level may be of benefit for interpretation, as well. One
approach to investigate improving these scales further is to use modification indices analyses
on the scales, as presented in Chapter 4 (appendix). Modifications to the scales may include
adding items to more than one scale (as in the case of a question related to role functioning)
or modeling correlation between items within a scale. We did not pursue these modifications
to maintain the standard scoring of the NEI-VFQ-25, but they merit further investigation.
In Chapter 5, I analyzed the NEI-VFQ-25 from the perspective of item response theory.
A unique contribution was fitting test and item information functions, which allowed me to
analyze the performance of the test overall and by item. The results suggest that the NEI-
VFQ-25 performs well for below-average VS-QOL score but has a harder time distinguishing
scores for above-average VS-QOL. While this area is arguably less critical, it is worth ex-
ploring how measurement for this range may be improved. A straightforward approach is to
use the previous implementation of the NEI-VFQ-25, which had 51 items, to determine if
any of these items improve the test information function (52,53). We also found that some
items, such as driving-related items, were not always crucial for each person’s measurement,
as they only had limited ranges where the provided information. Computerized adaptive
testing can account for these issues, only showing participants enough items to accurately
measure their score (46). The driving-related items, for instance, only measure information
for those with the worst VS-QOL, so an adapted test would only show the items to these
participants. Researchers should explore a central way of offering CAT for the NEI-VFQ-25,
such as APIs that data collection services can use or data science web frameworks such as
Shiny that can score items interactively (111). A related issue is equating the NEI-VFQ-25
with other scales that measure VS-QOL (46). Researchers could collect equating data into
a central service to compare scores better across studies.
124
5.3 Conclusion
In this dissertation, I attempt to describe and analyze the epidemiology of the impact of
visionlossonvision-specificqualityoflifeandtoimprovethemeasurementandinterpretation
of VS-QOL. There remains work to do in both areas, principally to understand the subtleties
ofVS-QOLinotherareasofvisionlossandtomakeimprovedmeasurementtoolsmorewidely
available to researchers. However, it is clear that the impact vision loss had on VS-QOL is a
significant public health concern, especially as the US population ages. The findings in this
dissertation may help clinicians and patients make important treatment decisions based on
the impact it has on VS-QOL. It may also improve future research on VS-QOL by providing
a path for better measurement and interpretation.
125
5.4 References
1. Jones GC, Rovner BW, Crews JE, et al. Effects of depressive symptoms on health
behavior practices among older adults with vision loss. Rehabilitation Psychology.
2009;54(2):164–172.
2. Lamoureux EL, Fenwick E, Moore K, et al. Impact of the Severity of Distance
and Near-Vision Impairment on Depression and Vision-Specific Quality of Life in
Older People Living in Residential Care. Investigative Ophthalmology & Visual Sci-
ence [electronicarticle]. 2009;50(9):4103–4109. (https://iovs.arvojournals.org/article.
aspx?articleid=2185912). (Accessed January 26, 2019)
3. Rovner BW, Ganguli M. Depression and Disability Associated with Impaired Vision:
The MoVIES Project. Journal of the American Geriatrics Society [electronic ar-
ticle]. 1998;46(5):617–619. (http://onlinelibrary.wiley.com/doi/abs/10.1111/j.1532-
5415.1998.tb01080.x). (Accessed January 26, 2019)
4. Zhang X, Bullard KM, Cotch MF, et al. Association Between Depression and Func-
tional Vision Loss in Persons 20 Years of Age or Older in the United States, NHANES
2005-2008. JAMA Ophthalmology [electronic article]. 2013;131(5):573–581. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/1660943). (Accessed Jan-
uary 26, 2019)
5. Crews JE, Campbell VA. Vision impairment and hearing loss among community-
dwelling older Americans: Implications for health and functioning. American Journal
of Public Health. 2004;94(5):823–829.
6. Ivers RQ, Norton R, Cumming RG, et al. Visual impairment and risk of hip fracture.
American Journal of Epidemiology. 2000;152(7):633–639.
7. Lee DJ, Gómez-Marín O, Lam BL, et al. Visual acuity impairment and mortality in
US adults. Archives of Ophthalmology (Chicago, Ill.: 1960). 2002;120(11):1544–1550.
126
8. 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 Ophthalmology [electronic article]. 2016;134(7):802–809. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/2523780). (Accessed Jan-
uary 26, 2019)
9. Choudhury F, Varma R, Klein R, et al. Age-Related Macular Degeneration and
Quality of Life in Latinos: The Los Angeles Latino Eye Study. JAMA Ophthalmol-
ogy [electronic article]. 2016;134(6):683–690. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/2514080). (Accessed March 12, 2018)
10. Globe DR, Wu J, Azen SP, et al. The impact of visual impairment on self-reported vi-
sual functioning in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(6):1141–1149. (http://www.aaojournal.org/article/S0161-
6420(04)00303-3/abstract). (Accessed March 19, 2018)
11. Mazhar K, Varma R, Choudhury F, et al. Severity of Diabetic Retinopathy and
Health-Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology
[electronic article]. 2011;118(4):649–655. (http://www.aaojournal.org/article/S0161-
6420(10)00819-5/abstract). (Accessed March 19, 2018)
12. PatinoCM,VarmaR,AzenSP,etal. TheImpactofChangeinVisualFieldonHealth-
Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2011;118(7):1310–1317. (http://www.aaojournal.org/article/s0161-
6420(10)01344-8/abstract). (Accessed March 19, 2018)
127
13. Varma R, Paz SH, Azen SP, et al. The Los Angeles Latino Eye Study: Design,
methods, and baseline data. Ophthalmology [electronic article]. 2004;111(6):1121–
1131. (http://www.aaojournal.org/article/S0161-6420(04)00291-X/abstract). (Ac-
cessed March 19, 2018)
14. McKean-Cowdin R, Varma R, Hays RD, et al. Longitudinal Changes in Visual Acu-
ity and Health-related Quality of Life: The Los Angeles Latino Eye Study. Ophthal-
mology [electronic article]. 2010;117(10):1900–1907.e1. (http://www.aaojournal.org/
article/S0161-6420(10)00132-6/abstract). (Accessed March 19, 2018)
15. Linder M, Chang TS, Scott IU, et al. Validity of the visual function index (vf-
14) in patients with retinal disease. Archives of Ophthalmology [electronic article].
1999;117(12):1611–1616. (http://dx.doi.org/10.1001/archopht.117.12.1611)
16. Mangione CM, Phillips RS, Lawrence MG, et al. Improved Visual Function
and Attenuation of Declines in Health-Related Quality of Life After Cataract
Extraction. Archives of Ophthalmology [electronic article]. 1994;112(11):1419–
1425. (https://jamanetwork.com/journals/jamaophthalmology/fullarticle/640890).
(Accessed March 19, 2018)
17. Health- and vision-related quality of life among patients with choroidal neovascular-
ization secondary to age-related macular degeneration at enrollment in randomized
trials of submacular surgery: SST report no. 4. American Journal of Ophthalmol-
ogy [electronic article]. 2004;138(1):91–108. (http://www.sciencedirect.com/science/
article/pii/S0002939404001370)
18. Parrish RK, Gedde SJ, Scott IU, et al. Visual function and quality of life
among patients with glaucoma. Archives of Ophthalmology (Chicago, Ill.: 1960).
1997;115(11):1447–1455.
128
19. Scott IU, Smiddy WE, Schiffman J, et al. Quality of life of low-vision patients and the
impact of low-vision services. American Journal of Ophthalmology [electronic article].
128(1):54–62. (http://dx.doi.org/10.1016/S0002-9394(99)00108-7). (Accessed March
20, 2018)
20. Healthy people 2020: vision. Healthy People 2020. (https://www.healthypeople.gov/
2020/topics-objectives/topic/vision)
21. Lamoureux E, Pesudovs K. Vision-Specific Quality-of-Life Research: A Need
to Improve the Quality. American Journal of Ophthalmology [electronic arti-
cle]. 2011;151(2):195–197.e2. (https://doi.org/10.1016/j.ajo.2010.09.020). (Accessed
March 18, 2020)
22. Coleman AL, Yu F, Keeler E, et al. Treatment of Uncorrected Refractive Error
Improves Vision-Specific Quality of Life. Journal of the American Geriatrics Society
[electronic article]. 2006;54(6):883–890. (http://onlinelibrary.wiley.com/doi/abs/10.
1111/j.1532-5415.2006.00817.x). (Accessed March 23, 2020)
23. BourneRR,StevensGA,WhiteRA,etal. Causesofvisionlossworldwide, 1990–2010:
A systematic analysis. The lancet global health. 2013;1(6):e339–e349.
24. Varma R, Chung J, Foong AWP, et al. Four-Year Incidence and Progression of Vi-
sual Impairment in Latinos: The Los Angeles Latino Eye Study (LALES). Amer-
ican journal of ophthalmology [electronic article]. 2010;149(5):713–727. (https:
//www.ncbi.nlm.nih.gov/pmc/articles/PMC2919324/). (Accessed January 26, 2019)
25. Varma R, Kim JS, Burkemper BS, et al. Prevalence and causes of visual impairment
and blindness in chinese american adults: The chinese american eye study. Jama
Ophthalmology. 2016;134(7):785–793.
129
26. Munoz B, West SK, Rubin GS, et al. Causes of blindness and visual impairment
in a population of older americans: The salisbury eye evaluation study. Archives of
ophthalmology. 2000;118(6):819–825.
27. Tang Y, Wang X, Wang J, et al. Prevalence and causes of visual impairment in a
chinese adult population: The taizhou eye study. Ophthalmology. 2015;122(7):1480–
1488.
28. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindness in a rural chinese adult population: The handan eye study. Ophthalmology.
2008;115(11):1965–1972.
29. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in beijing: The beijing eye study. Ophthalmology. 2006;113(7):1134–e1.
30. ZhaoJ,XuX,EllweinLB,etal. Causesofvisualimpairmentandblindnessinthe2006
and 2014 nine-province surveys in rural china. American Journal of Ophthalmology.
2019;197:80–87.
31. AtteboK,MitchellP,SmithW.Visualacuityandthecausesofvisuallossinaustralia:
The blue mountains eye study. Ophthalmology. 1996;103(3):357–364.
32. Hyman L, Wu S-Y, Connell AM, et al. Prevalence and causes of visual impairment
in the barbados eye study. Ophthalmology. 2001;108(10):1751–1756.
33. Klaver CC, Wolfs RC, Vingerling JR, et al. Age-specific prevalence and causes
of blindness and visual impairment in an older population: The rotterdam study.
Archives of ophthalmology. 1998;116(5):653–658.
34. Rodriguez J, Sanchez R, Munoz B, et al. Causes of blindness and visual impairment
in a population-based sample of US hispanics. Ophthalmology. 2002;109(4):737–743.
130
35. CongdonN,O’ColmainB,KlaverC,etal. Causesandprevalenceofvisualimpairment
among adults in the united states. Archives of Ophthalmology (Chicago, Ill.: 1960).
2004;122(4):477–485.
36. VanNewkirk MR, Weih L, McCarty CA, et al. Cause-specific prevalence of bilateral
visual impairment in victoria, australia: The visual impairment project. Ophthalmol-
ogy. 2001;108(5):960–967.
37. 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. Investigative Ophthalmology & Visual Science. 2002;43(11):3393–
3398.
38. Unzueta M, Globe D, Wu J, et al. Compliance with recommendations for follow-up
care in latinos: The los angeles latino eye study. Ethnicity & disease. 2004;14:285–91.
39. Varma R, Mohanty SA, Deneen J, et al. Burden and predictors of undetected eye
disease in mexican-americans. Medical care. 2008;46(5):497–506.
40. Patty LE, Morales LS, Lai M, et al. Self-reported utilization of eye care among latinos
in the los angeles latino eye study (LALES). Investigative Ophthalmology & Visual
Science. 2009;50(13):2502–2502.
41. Varma R, Torres M, McKean-Cowdin R, et al. Prevalence and risk factors for re-
fractive error in adult chinese americans: The chinese american eye study. American
Journal of Ophthalmology. 2017;175:201–212.
42. Jiang X, Varma R, Torres M, et al. Self-reported use of eye care among adult chinese
americans in the chinese american eye study. American journal of ophthalmology.
2017;176:183–193.
131
43. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by hispanics: Findings from HHANES 1982-84. American Journal of
Public Health. 1990;80(Suppl):11–19.
44. Zheng Y, Lamoureux EL, Ikram MK, et al. Impact of migration and acculturation
on prevalence of type 2 diabetes and related eye complications in indians living in a
newly urbanised society. PloS One. 2012;7(4):e34829.
45. Streiner DL, Cairney J, Norman GR. Health measurement scales: A practical guide
to their development and use. Oxford Univ. Press; 2015.
46. Ayala RJde. The theory and practice of item response theory. Guildford Press; 2009.
47. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Medical
Research Methodology [electronic article]. 2008;8:70. (https://www.ncbi.nlm.nih.gov/
pmc/articles/PMC2601045/). (Accessed November 21, 2018)
48. Hernán MA, Robins JM. Causal Inference. Boca Raton: Chapman Hall/CRC; 2019.
49. Steinberg EP, Tielsch JM, Schein OD, et al. The VF-14: An Index of Functional
ImpairmentinPatientsWithCataract. Archives of Ophthalmology [electronicarticle].
1994;112(5):630–638. (https://doi.org/10.1001/archopht.1994.01090170074026)
50. Mangione CM, Phillips RS, Seddon JM, et al. Development of the ’Activities of Daily
Vision Scale’. A measure of visual functional status. Medical Care. 1992;30(12):1111–
1126.
51. Margolis MK, Coyne K, Kennedy-Martin T, et al. Vision-specific instruments for the
assessment of health-related quality of life and visual functioning: A literature review.
PharmacoEconomics. 2002;20(12):791–812.
132
52. Mangione CM, Lee PP, Gutierrez PR, et al. Development of the 25-list-item
National Eye Institute Visual Function Questionnaire. Archives of Ophthalmol-
ogy [electronic article]. 2001;119(7):1050–1058. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/267134). (Accessed March 19, 2018)
53. Mangione CM, Lee PP, Pitts J, et al. Psychometric Properties of the National
Eye Institute Visual Function Questionnaire (NEI-VFQ). Archives of Ophthalmology
[electronic article]. 1998;116(11):1496–1504. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/264225). (Accessed March 19, 2018)
54. Pesudovs K, Gothwal VK, Wright T, et al. Remediating serious flaws in the Na-
tional Eye Institute Visual Function Questionnaire. Journal of Cataract & Refrac-
tive Surgery [electronicarticle]. 2010;36(5):718–732. (http://linkinghub.elsevier.com/
retrieve/pii/S0886335010002531). (Accessed March 12, 2018)
55. Kline RB. Principles and practice of structural equation modeling. The Guilford
Press; 2016.
56. Bentler PM. On tests and indices for evaluating structural models. Personal-
ity and Individual Differences [electronic article]. 2007;42(5):825–829. (http://
www.sciencedirect.com/science/article/pii/S0191886906003898). (Accessed March
18, 2020)
57. Brucker PS, Yost K, Cashy J, et al. General population and cancer pa-
tient norms for the functional assessment of cancer therapy-general (FACT-
g). Evaluation & the Health Professions [electronic article]. 2005;28(2):192–
211. (%20%0A%20%20%20%20%20%20%20%20https://doi.org/10.1177/
0163278705275341%0A%20%20%20%20%0A)
133
58. Wright BD. Solving MEASUREMENT PROBLEMS WITH THE Rasch MODEL.
Journal of Educational Measurement [electronic article]. 1977;14(2):97–116. (https:
//onlinelibrary.wiley.com/doi/abs/10.1111/j.1745-3984.1977.tb00031.x)
59. Harrell F. Regression modeling strategies: With applications to linear models, logistic
and ordinal regression, and survival analysis. springer; 2016.
60. Medeiros FA, Gracitelli CPB, Boer ER, et al. Longitudinal Changes in Quality of
Life and Rates of Progressive Visual Field Loss in Glaucoma Patients. Ophthalmol-
ogy [electronic article]. 122(2):293–301. (http://dx.doi.org/10.1016/j.ophtha.2014.08.
014). (Accessed March 20, 2018)
61. Association WM. World Medical Association Declaration of Helsinki: Ethical Prin-
ciples for Medical Research Involving Human Subjects. JAMA [electronic article].
2013;310(20):2191–2194. (https://doi.org/10.1001/jama.2013.281053)
62. GlobeDR,Schoua-GlusbergA,PazS,etal. Usingfocusgroupstodevelopaculturally
sensitive methodology for epidemiological surveys in a Latino population: Findings
from the Los Angeles Latino Eye Study (LALES). Ethnicity & disease [electronic
article]. 2002;12(2):259–266. (http://europepmc.org/abstract/med/12019936). (Ac-
cessed March 19, 2018)
63. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by Hispanics: Findings from HHANES 1982-84. American Journal
of Public Health [electronic article]. 1990;80(Suppl):11–19. (https://doi.org/10.2105/
AJPH.80.Suppl.11). (Accessed March 20, 2018)
64. Brody BL, Gamst AC, Williams RA, et al. Depression, visual acuity, comorbidity,
and disability associated with age-related macular degeneration. Ophthalmology [elec-
tronicarticle]. 108(10):1893–1900. (http://dx.doi.org/10.1016/S0161-6420(01)00754-
0). (Accessed March 20, 2018)
134
65. Globe DR, Varma R, Torres M, et al. Self-reported comorbidities and visual function
inapopulation-basedstudy: Thelosangeleslatinoeyestudy. Archives of Ophthalmol-
ogy [electronic article]. 2005;123(6):815–821. (http://dx.doi.org/10.1001/archopht.
123.6.815)
66. Linn Bernard S., Linn Margaret W., Gurel Lee. Cumulative illness rating scale.
Journal of the American Geriatrics Society [electronic article]. 2015;16(5):622–626.
(https://onlinelibrary.wiley.com/doi/full/10.1111/j.1532-5415.1968.tb02103.x). (Ac-
cessed March 19, 2018)
67. Azen SP, Varma R, Preston-Martin S, et al. Binocular Visual Acuity Sum-
mation and Inhibition in an Ocular Epidemiological Study: The Los Angeles
Latino Eye Study. Investigative Ophthalmology & Visual Science [electronic ar-
ticle]. 2002;43(6):1742–1748. (http://iovs.arvojournals.org/article.aspx?articleid=
2123581). (Accessed March 19, 2018)
68. Varma R, Fraser-Bell S, Tan S, et al. Prevalence of age-related macular degen-
eration in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(7):1288–1297. (http://www.aaojournal.org/article/S0161-
6420(04)00313-6/abstract). (Accessed March 19, 2018)
69. Varma R, Torres M, Peña F, et al. Prevalence of diabetic retinopathy in adult
Latinos: The Los Angeles Latino Eye Study. Ophthalmology [electronic article].
2004;111(7):1298–1306. (http://www.aaojournal.org/article/S0161-6420(04)00336-
7/abstract). (Accessed March 19, 2018)
135
70. Ware JE, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: Construction
ofScalesandPreliminaryTestsofReliabilityandValidity. Medical Care [electronicar-
ticle]. 1996;34(3):220. (https://journals.lww.com/lww-medicalcare/Abstract/1996/
03000/A_12_Item_Short_Form_Health_Survey__Construction.3.aspx). (Ac-
cessed March 19, 2018)
71. Ware J, A. Kosinski M, D. Keller S. SF-12: How to Score the SF-12 Physical and
Mental Health Summary Scales. 1998;
72. Rosenthal R. Parametric measures of effect size. In: The handbook of research syn-
thesis. New York, NY, US: Russell Sage Foundation; 1994:231–244.
73. Cohen J. A power primer. Psychological bulletin [electronic article]. 1992;112(1):155–
159. (http://libproxy.usc.edu/login?url=https://search-proquest-com.libproxy2.usc.
edu/docview/614317877?accountid=14749)
74. Hays RD, Brodsky M, Johnston MF, et al. Evaluating the Statistical Significance
of Health-Related Quality-Of-Life Change in Individual Patients. Evaluation & the
Health Professions [electronic article]. 2005;28(2):160–171. (https://doi.org/10.1177/
0163278705275339). (Accessed March 19, 2018)
75. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epi-
demiology (Cambridge, Mass.). 1999;10(1):37–48.
76. Greenland S. Quantifying biases in causal models: Classical confounding vs collider-
stratification bias. Epidemiology (Cambridge, Mass.). 2003;14(3):300–306.
77. Buuren S van. Flexible imputation of missing data. CRC Press; 2018.
136
78. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained
equations in r. Journal of Statistical Software [electronic article]. 2011;45(3):1–67.
(https://www.jstatsoft.org/v45/i03/)
79. R Core Team. R: A language and environment for statistical computing. Vienna,
Austria: RFoundationforStatisticalComputing; 2017.(https://www.R-project.org/)
80. Rosseel Y. lavaan: An R package for structural equation modeling. Journal of Sta-
tistical Software [electronic article]. 2012;48(2):1–36. (http://www.jstatsoft.org/v48/
i02/)
81. Epskamp S, Simon Stuber with contributions from. semPlot: Path diagrams and
visual analysis of various SEM packages’ output. 2017.(https://CRAN.R-project.
org/package=semPlot)
82. Harrell Jr FE. Rms: Regression modeling strategies. 2018.(https://CRAN.R-project.
org/package=rms)
83. Textor J, van der Zander B. Dagitty: Graphical analysis of structural causal models.
2016.
84. Barrett M. Ggdag: Analyze and create elegant directed acyclic graphs. 2018.
85. Wickham H. ggplot2: Elegant graphics for data analysis. Springer-Verlag New York;
2016.(http://ggplot2.org)
86. Alqudah A, Mansberger SL, Gardiner SK, et al. Vision-related quality of life in glau-
coma suspect or early glaucoma patients. Journal of glaucoma [electronic article].
2016;25(8):629–633. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4975427/).
(Accessed March 12, 2018)
137
87. Fenwick EK, Man RK, Ong P, et al. Association of changes in visual acuity
with vision-specific functioning in the singapore malay eye study. JAMA Oph-
thalmology [electronic article]. 2016;134(11):1299–1305. (http://dx.doi.org/10.1001/
jamaophthalmol.2016.3692)
88. Responsiveness of the National Eye Institute Visual Function Questionnaire
to Changes in Visual Acuity. Archives of ophthalmology [electronic article].
2003;121(4):531–539. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1434793/).
(Accessed March 19, 2018)
89. Lindblad AS, Clemons TE. Responsiveness of the National Eye Institute Visual Func-
tion Questionnaire to progression to advanced age-related macular degeneration, vi-
sion loss, and lens opacity: AREDS Report no. 14. Archives of Ophthalmology
(Chicago, Ill.: 1960). 2005;123(9):1207–1214.
90. Hays RD, Morales LS, Reise SP. Item Response Theory and Health Outcomes Mea-
surementin the21stCentury. Medical care [electronicarticle]. 2000;38(9Suppl):II28–
II42. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1815384/). (Accessed
March 19, 2018)
91. Wittenborn J, Rein D. The economic burden of vision loss and eye disorders in the
United States. University of Chicago; 2013.(https://www.preventblindness.org/sites/
default/files/national/documents/Economic%20Burden%20of%20Vision%20Final%
20Report_130611.pdf)
92. Wang X, Lamoureux E, Zheng Y, et al. Health Burden Associated with Visual Im-
pairment in Singapore: The Singapore Epidemiology of Eye Disease Study. Ophthal-
mology [electronic article]. 2014;121(9):1837–1842. (http://www.sciencedirect.com/
science/article/pii/S0161642014002413). (Accessed March 23, 2020)
138
93. Tsai S-Y, Chi L-Y, Cheng C-Y, et al. The Impact of Visual Impairment and Use of
Eye Services on Health-Related Quality of Life among the Elderly in Taiwan: The
Shihpai Eye Study. Quality of Life Research [electronic article]. 2004;13(8):1415–
1424. (https://doi.org/10.1023/B:QURE.0000040791.87602.fe). (Accessed March 23,
2020)
94. Vision impairment and major eye diseases reduce vision-specific emotional well-being
in a Chinese population - ProQuest. (https://search-proquest-com.libproxy1.usc.
edu/docview/1894846003?rfr_id=info%3Axri%2Fsid%3Aprimo). (Accessed March
23, 2020)
95. Varma R, Kim JS, Burkemper BS, et al. Prevalence and Causes of Visual Impairment
andBlindnessinChineseAmericanAdults: TheChineseAmericanEyeStudy. JAMA
Ophthalmology [electronic article]. 2016;134(7):785–793. (https://jamanetwork-
com.libproxy1.usc.edu/journals/jamaophthalmology/fullarticle/2523778). (Accessed
March 21, 2020)
96. Hsu W-M, Cheng C-Y, Liu J-H, et al. Prevalence and causes of visual impairment
in an elderly Chinese population in Taiwan: The Shihpai Eye Study. Ophthalmology.
2004;111(1):62–69.
97. Huang S, Zheng Y, Foster PJ, et al. Prevalence and causes of visual impairment in
Chinese adults in urban southern China. Archives of Ophthalmology (Chicago, Ill.:
1960). 2009;127(10):1362–1367.
98. Saw S-M, Foster PJ, Gazzard G, et al. Causes of blindness, low vision, and
questionnaire-assessed poor visual function in Singaporean Chinese adults: The Tan-
jong Pagar Survey. Ophthalmology. 2004;111(6):1161–1168.
139
99. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindnessinaruralchineseadultpopulation: TheHandanEyeStudy. Ophthalmology.
2008;115(11):1965–1972.
100. Zhao J, Ellwein LB, Cui H, et al. Prevalence of vision impairment in older adults in
rural China: The China Nine-Province Survey. Ophthalmology. 2010;117(3):409–416,
416.e1.
101. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in Beijing: The Beijing Eye Study. Ophthalmology. 2006;113(7):1134.e1–
11.
102. Varma R, Hsu C, Wang D, et al. The Chinese American Eye Study: Design and
Methods. Ophthalmic Epidemiology [electronic article]. 2013;20(6):335–347. (http:
//www.tandfonline.com/doi/full/10.3109/09286586.2013.823505). (Accessed March
21, 2020)
103. SuinnRM,AhunaC,KhooG.TheSuinn-LewAsianSelf-IdentityAcculturationScale:
Concurrent and factorial validation. Educational and Psychological Measurement.
1992;52(4):1041–1046.
104. Burkemper B, Torres M, Jiang X, et al. Factors Associated with Visual Impairment in
Chinese American Adults: The Chinese American Eye Study. Ophthalmic Epidemiol-
ogy [electronic article]. 2019;26(5):329–335. (https://doi.org/10.1080/09286586.2019.
1622737). (Accessed March 22, 2020)
105. Linden WJ van der. Handbook of item response theory. Chapman & Hall/CRC;
2019.
106. Wood SN. Generalized additive models: An introduction with r. CRC Press/Taylor
& Francis Group; 2017.
140
107. Rizopoulos D. Ltm: An r package for latent variable modelling and item response
theory analyses. Journal of Statistical Software [electronic article]. 2006;17(5):1–25.
(http://www.jstatsoft.org/v17/i05/)
108. Fenwick EK, Ong PG, Sabanayagam C, et al. Assessment of the psychometric proper-
ties of the Chinese Impact of Vision Impairment questionnaire in a population-based
study: FindingsfromtheSingaporeChineseEyeStudy. Qualityof LifeResearch [elec-
tronic article]. 2016;25(4):871–880. (https://doi.org/10.1007/s11136-015-1141-1).
(Accessed March 23, 2020)
109. Rees G, Tee HW, Marella M, et al. Vision-specific distress and depressive symptoms
in people with vision impairment. Investigative Ophthalmology & Visual Science.
2010;51(6):2891–2896.
110. Morales LS, Varma R, Paz SH, et al. Self-reported use of eye care among Latinos:
The Los Angeles Latino Eye Study. Ophthalmology. 2010;117(2):207–215.e1.
111. Chang W, Cheng J, Allaire J, et al. Shiny: Web application framework for r.
2020.(https://CRAN.R-project.org/package=shiny)
141
APPENDIX A
SUPPLEMENT TO CHAPTER 3
A.1 A primer on causal structural models and directed acyclic graphs
Here, I provide a brief introduction to causal directed acyclic graphs (DAGs), directed
because they show arrows from one variable (or node) to another in a single direction and
acyclic because there are no feedback loops. DAGs are created from three types of relation-
ships: Chains, forks, and inverted forks (paths with colliders). Briefly, each arrow represents
a causal assumption, and you can think of causality as flowing through the arrows like wa-
ter in a pipe. If you follow the directed path from one variable to another, causality flows
through open paths. Chains and forks are open, but inverted forks are closed. Controlling
for a variable, as in a regression model, blocks that path. Figure 1a is a demonstration of
classic confounding. We are analyzing the causal effect of x on y, although there isn’t one.
There is also an open path from x to y through the forked path at z, sometimes called a
back-door path. A crude estimate of the effect of x on y will appear as if there is one because
of this open path; we need to close it by adjusting for z (Figure 1b). In more complicated
DAGs, we don’t necessarily need to block open paths at multiple points along the same
back-door path, although we may have to block more than one path. We often talk about
confounders, but really we should talk about confounding, because it is about the pathway
more than any particular node along the path. The set of variables you need to block all
paths is sometimes called a minimal adjustment set, and often there are several sets that all
provide an unbiased effect.
Controlling for variables at inverted forks, also called colliders because of the way the
arrows meet at the node, has the opposite effect. Figure 2a shows a collider, m. Assessing
the crude effect of x on y will give us the correct answer, because the path is closed at m.
142
Controlling for m is actually the wrong decision: it induces a false association between x and
y and will bias results even when there is an effect. This is selection bias, and most cases of
selection bias can be shown as a DAG. For instance, if you imagine that m is our sampling
scheme, which is dependent on the outcome and exposure, our study will produce a biased
result because it is inherently conditioned on m.
143
z
x y
A
z
x y
{z}
adjusted
adjusted
adjusted
unadjusted
B
Figure A.1: Confounding as a Directed Acyclic Graph: Blocking Back-Door Paths
144
x y
m
A
x y
m
activated by
adjustment
for collider
adjusted
unadjusted
adjusted
adjusted
unadjusted
B
Figure A.2: Selection Bias as a Directed Acyclic Graph: Stratifying on a Collider
145
APPENDIX B
SUPPLEMENT TO CHAPTER 4
B.0.1 Scoring for new subscales
Table B.1: Traditional and Improved Scales for Items in the 25-Item National Eye Institute Visual Function
Questionnaire
Traditional Subscale Improved subcomposite Item
General health N/A 5-Level overall health rating
General Vision Well-being 5-Level overall vision rating
Ocular Pain Well-being How much pain or discomfort
have you had in and around your
eyes (for example, burning,
itching, or aching)?
Well-being How much does pain or
discomfort in or around your
eyes, for example, burning,
itching, or aching, keep you from
doing what you’d like to be
doing?
Near Activities Task How much difficulty do you have
reading ordinary print in
newspapers?
146
Traditional Subscale Improved subcomposite Item
Task How much difficulty do you have
doing work or hobbies that
require you to see well up close,
such as cooking, sewing, fixing
things around the house, or using
hand tools
Task Because of your eyesight, how
much difficulty do you have
finding something on a crowded
shelf?
Distance Activities Task How much difficulty do you have
reading street signs or the names
of stores?
Task Because of your eyesight, how
much difficulty do you have going
down steps, stairs, or curbs in
dim light or at night?
Task Because of your eyesight, how
much difficulty do you have going
out to see movies, plays, or
sports events?
Social Functioning Well-being Because of your eyesight, how
much difficulty do you have
seeing how people react to things
you say?
147
Traditional Subscale Improved subcomposite Item
Well-being Because of your eyesight, how
much difficulty do you have
visiting with people in their
homes, at parties, or in
restaurants?
Mental Health Well-being How much of the time do you
worry about your eyesight?
Well-being I feel frustrated a lot of the time
because of my eyesight.
Well-being I have much less control over
what I do, because of my
eyesight.
Well-being I worry about doing things that
will embarrass myself or others,
because of my eyesight
Role Difficulties Task Do you accomplish less than you
would like because of your vision?
Task Are you limited in how long you
can work or do other activities
because of your vision?
Dependency Well-being I stay home most of the time
because of my eyesight.
Well-being Because of my eyesight, I have to
rely too much on what other
people tell me.
Well-being I need a lot of help from others
because of my eyesight
148
Traditional Subscale Improved subcomposite Item
Driving Task How much difficulty do you have
driving during the daytime in
familiar places?
Task How much difficulty do you have
driving at night?
Task How much difficulty do you have
driving in difficult conditions,
such as in bad weather, during
rush hour, on the freeway, or in
city traffic?
Color Vision Task Because of your eyesight, how
much difficulty do you have
picking out and matching your
own clothes?
Peripheral Vision Task Because of your eyesight, how
much difficulty do you have
noticing objects off to the side
while you are walking along?
B.0.2 DAG of causal assumptions for selecting covariates
The adjustment sets to identify the unbiased effect of visual acuity on quality of life,
assuming this DAG is correct, are: {Visual Field (Wave 3), Co-morbidity Score}, {Co-
morbidity Score, Cataracts, Glaucoma, Insurance/Access to Care, Age}, {Co-morbidity
Score, Cataracts, Insurance/AccesstoCare, Age, Sex, Education, SES}, {DiabeticRetinopa-
thy, AMD, Cataracts, Insurance/Access to Care, Age}.
We opted for the middle set.
149
Figure B.1: Causal Diagram for the Effect of Visual Acuity on Quality of Life
150
B.0.3 Contrasts without Inverse Probability of Censoring Weights
Inthissensitivityanalysis, wecalculatethecontrastsusedinthemainpaperbutwithout
weighting by the inverse probability of censoring. Overall, the results were very similar. In
general, they suggest that analyses of VA and QOl that do not account for loss to follow-up
may produce slightly underestimated effects, but that the magnitude of this bias may be
minimal.
Table B.2: Predicted Mean Change of VS-QOL at Levels of Visual Acuity Change without IPCW
Vision Improved Vision Declined
Two Lines Four Lines Two Lines Four Lines
Difference (95% CI) Difference (95% CI) Difference (95% CI) Difference (95% CI)
Composite 2.59 (1.12, 3.90) 6.86 (1.81, 10.89) -1.36 (-2.54, -0.29) -3.20 (-5.19, -1.39)
Task Subcomposite 2.66 (1.13, 4.04) 6.79 (1.65, 11.41) -1.65 (-2.95, -0.41) -3.97 (-6.16, -2.05)
Well-being
Subcomposite
2.53 (1.13, 3.78) 6.94 (1.59, 10.82) -1.06 (-2.39, 0.15) -2.37 (-4.53, -0.41)
N = 2234. Fit with ordinary least squares regression and calculated with the parametric G-formula. Adjusted for age,
sex, cataracts, comorbidities, education, employment, acculturation, baseline visual acuity, and insurance.
N = 2234. Fit with ordinary least squares regression and calculated with the parametric G-formula. Adjusted for age, sex,
cataracts, comorbidities, education, employment, acculturation, baseline visual acuity, and insurance.
151
Figure B.2: G-Formula Calculated Changes in Vision-Related Quality of Life at 2 and 4 Lines Change in
Visual Acuity without IPCW
B.0.3.1 VS-QOL by Cause of Visual Impairment
Table B.3: VS-QOL among those with incident VI/blindness by cause
Cause N % Composite Task Well-Being
Diabetic Retinopathy 8 14.0% -15.5 -18.4 -12.5
Glaucoma 3 5.3% -10.0 -12.0 -7.6
Cataract 25 43.9% -0.5 1.0 -1.7
Other 17 29.8% -0.3 0.8 -1.2
AMD 4 7.0% 6.2 8.3 4.0
VS-QOL: Vision-Specific Quality of Life. VI: Visual impairment. Total N for each group represents participants with complete
visual acuity and VS-QOL data at baseline and year 4. VS-QOL scores represent the change for the given scale from baseline
to year 4, when the incident visual impairment occurred.
152
B.0.3.2 Modification Indices
Table B.4: Top 10 Modification Indices for Task and Well-Being CFA
Variable 1 Variable 2 MI (Chi-Sq)
RF1 RF2 686.3
Task SF1 632.9
DP2 DP3 582.1
OP1 OP2 413.0
DV3 SF2 298.3
PV1 SF1 265.2
CV1 SF1 222.7
Well-Being RF2 218.7
DP3 MH4 217.3
DV2 PV1 216.4
CFA: confirmatory factor analysis. MI: Modification indices. Each variable is either an observed item (e.g., “OP1,” “PV1”)
as asked in the traditional scoring of the NEI-VFQ-25 or a latent scale (e.g., “Task,” “Well-Being”). When one Variable 1 is
latent, the modification indices are the increase in chi-square values when Variable 2 is added to the scale. When Variable 1 is
observed, modification indices are the increase in chi-square values when the correlation between Variable 1 and Variable 2 are
explicitly modeled in the CFA.
153
APPENDIX C
SUPPLEMENT TO CHAPTER 5
C.0.1 NEI-VFQ-25 Item Descriptions and Discrimination
Table C.1: NEI-VFQ-25 Item Descriptions and Discrimination
Item Short Code Discrimination
5-Level overall health rating Overall health rating –
5-Level overall vision rating Overall vision rating 2.26
How much pain or discomfort have you had in and around your eyes (for example, burning, itching, or
aching)?
Pain in and around
eyes
-1.31
How much does pain or discomfort in or around your eyes, for example, burning, itching, or aching, keep
you from doing what you’d like to be doing?
Pain limits activities 2.67
How much difficulty do you have reading ordinary print in newspapers? Difficulty reading
print
1.60
How much difficulty do you have doing work or hobbies that require you to see well up close, such as
cooking, sewing, fixing things around the house, or using hand tools
Difficulty with close
work
2.58
Because of your eyesight, how much difficulty do you have finding something on a crowded shelf? Difficulty finding
things on shelf
2.57
How much difficulty do you have reading street signs or the names of stores? Difficulty reading
signs
1.79
Because of your eyesight, how much difficulty do you have going down steps, stairs, or curbs in dim light
or at night?
Difficulty with stairs
at night
3.08
Because of your eyesight, how much difficulty do you have going out to see movies, plays, or sports events? Difficulty going to
events
4.56
Because of your eyesight, how much difficulty do you have seeing how people react to things you say? Difficulty seeing
reactions
2.18
Because of your eyesight, how much difficulty do you have visiting with people in their homes, at parties,
or in restaurants?
Difficulty visiting
others
1.47
How much of the time do you worry about your eyesight? Worry about eyesight -1.90
I feel frustrated a lot of the time because of my eyesight. Frustrated with
eyesight
3.90
I have much less control over what I do, because of my eyesight. Less control over
activities
4.54
I worry about doing things that will embarrass myself or others, because of my eyesight Worry about
embarassing self
3.71
Do you accomplish less than you would like because of your vision? Accomplished less 1.58
154
Item Short Code Discrimination
Are you limited in how long you can work or do other activities because of your vision? Limits work or
activities
1.96
I stay home most of the time because of my eyesight. Stay at home more 2.62
Because of my eyesight, I have to rely too much on what other people tell me. Rely on what others
see
5.45
I need a lot of help from others because of my eyesight Need help from others 4.72
How much difficulty do you have driving during the daytime in familiar places? Driving during day 0.85
How much difficulty do you have driving at night? Driving at night 0.58
Because of your eyesight, how much difficulty do you have picking out and matching your own clothes? Difficulty matching
clothes
2.23
Because of your eyesight, how much difficulty do you have noticing objects off to the side while you are
walking along?
Difficulty seeing to
side
1.33
155
C.0.2 CTT-Based Task and Well-Being Composites
Figure C.1: Distribution of CTT-Based Vision-Specific Quality of Life and Relationship with Visual Acuity
A: Density plot of CTT-based Task and Well-Being composites of Vision-Specific Quality of Life. x-axis represents
VS-QOL and y-axis represents density of VS-QOL values among CHES participants. B: Smoothed general additive models of
the relationship between visual acuity and VS-QOL. x-axis represents the presenting VA in the participant’s better-seeing eye
in logMAR units. Participants with visual acuity over 0.3 logMAR units are visually impaired. y-axis represents CTT-based
VS-QOL values.
156
Table C.2: Predicted Mean Difference of VS-QOL (CTT) at Levels of Visual Acuity Difference
Vision Better than 20/20 Vision Worse than 20/20
Two Lines Two Lines Four Lines
Difference (95% CI) Difference (95% CI) Difference (95% CI)
Task Composite 1.10 (-0.30, 2.31) -1.88 (-2.70, -1.16) -6.27 (-7.47, -5.03)
Well-being Composite 2.56 (0.43, 4.56) -2.05 (-2.93, -1.19) -6.27 (-7.75, -4.80)
N = 4564 for Task models and N = 4564 for Well-Being models. Outcomes calculated using scoring described in Chapter
4 (CTT). Fit with ordinary least squares regression and calculated with the parametric G-formula. Estimates are the difference
in mean vision-specific quality of life at 2 or 4 lines of presenting visual acuity (better-seeing eye) better and worse than 20/20
compared to 20/20 (logMAR value of 0). Adjusted for age, cataracts, comorbidities, education, employment, acculturation,
and insurance. Covariates imputed using multiple imputation. CI = confidence intervals.
157
Figure C.2: Difference in Vision-Related Quality of Life and Visual Acuity by CTT-based Composite
N = 4564 for Task models and N = 4564 for Well-Being models. Outcomes calculated using scoring described in
Chapter 4 (CTT). Fit with ordinary least squares regression and calculated with the parametric G-formula. Adjusted for age,
cataracts, comorbidities, education, employment, acculturation, and insurance. Covariates imputed using multiple imputation.
A: Predicted mean VS-QOL over a range of VA and 95% confidence interval. B: Parametric G-formula point estimates of
differences in VS-QOL at exact points of VA. Estimates are the difference in mean vision-specific quality of life at 2 or 4 lines of
presenting visual acuity (better-seeing eye) better and worse than 20/20 compared to 20/20 (logMAR value of 0). VA: Visual
Acuity. VS-QOL: Vision-Specific Quality of Life.
158
C.0.3 DAG of causal assumptions for selecting covariates
Figure C.3: Causal Diagram for the Effect of Visual Acuity on Quality of Life
The adjustment sets to identify the unbiased effect of visual acuity on quality of life,
assuming this DAG is correct, are: {Visual Field, Co-morbidity Score}, {Co-morbidity
Score, Cataracts, Glaucoma, Insurance/Access to Care, Age}, {Co-morbidity Score,
Cataracts, Insurance/Access to Care, Age, Sex, Education, SES}, {Diabetic Retinopathy,
AMD, Cataracts, Insurance/Access to Care, Age}.
We opted for the middle set.
159
References
1. Jones GC, Rovner BW, Crews JE, et al. Effects of depressive symptoms on health
behavior practices among older adults with vision loss. Rehabilitation Psychology.
2009;54(2):164–172.
2. Lamoureux EL, Fenwick E, Moore K, et al. Impact of the Severity of Distance
and Near-Vision Impairment on Depression and Vision-Specific Quality of Life in
Older People Living in Residential Care. Investigative Ophthalmology & Visual Sci-
ence [electronicarticle]. 2009;50(9):4103–4109. (https://iovs.arvojournals.org/article.
aspx?articleid=2185912). (Accessed January 26, 2019)
3. Rovner BW, Ganguli M. Depression and Disability Associated with Impaired Vision:
The MoVIES Project. Journal of the American Geriatrics Society [electronic ar-
ticle]. 1998;46(5):617–619. (http://onlinelibrary.wiley.com/doi/abs/10.1111/j.1532-
5415.1998.tb01080.x). (Accessed January 26, 2019)
4. Zhang X, Bullard KM, Cotch MF, et al. Association Between Depression and Func-
tional Vision Loss in Persons 20 Years of Age or Older in the United States, NHANES
2005-2008. JAMA Ophthalmology [electronic article]. 2013;131(5):573–581. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/1660943). (Accessed Jan-
uary 26, 2019)
5. Crews JE, Campbell VA. Vision impairment and hearing loss among community-
dwelling older Americans: Implications for health and functioning. American Journal
of Public Health. 2004;94(5):823–829.
6. Ivers RQ, Norton R, Cumming RG, et al. Visual impairment and risk of hip fracture.
American Journal of Epidemiology. 2000;152(7):633–639.
160
7. Lee DJ, Gómez-Marín O, Lam BL, et al. Visual acuity impairment and mortality in
US adults. Archives of Ophthalmology (Chicago, Ill.: 1960). 2002;120(11):1544–1550.
8. 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 Ophthalmology [electronic article]. 2016;134(7):802–809. (http://
jamanetwork.com/journals/jamaophthalmology/fullarticle/2523780). (Accessed Jan-
uary 26, 2019)
9. Choudhury F, Varma R, Klein R, et al. Age-Related Macular Degeneration and
Quality of Life in Latinos: The Los Angeles Latino Eye Study. JAMA Ophthalmol-
ogy [electronic article]. 2016;134(6):683–690. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/2514080). (Accessed March 12, 2018)
10. Globe DR, Wu J, Azen SP, et al. The impact of visual impairment on self-reported vi-
sual functioning in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(6):1141–1149. (http://www.aaojournal.org/article/S0161-
6420(04)00303-3/abstract). (Accessed March 19, 2018)
11. Mazhar K, Varma R, Choudhury F, et al. Severity of Diabetic Retinopathy and
Health-Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology
[electronic article]. 2011;118(4):649–655. (http://www.aaojournal.org/article/S0161-
6420(10)00819-5/abstract). (Accessed March 19, 2018)
12. PatinoCM,VarmaR,AzenSP,etal. TheImpactofChangeinVisualFieldonHealth-
Related Quality of Life: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2011;118(7):1310–1317. (http://www.aaojournal.org/article/s0161-
6420(10)01344-8/abstract). (Accessed March 19, 2018)
161
13. Varma R, Paz SH, Azen SP, et al. The Los Angeles Latino Eye Study: Design,
methods, and baseline data. Ophthalmology [electronic article]. 2004;111(6):1121–
1131. (http://www.aaojournal.org/article/S0161-6420(04)00291-X/abstract). (Ac-
cessed March 19, 2018)
14. McKean-Cowdin R, Varma R, Hays RD, et al. Longitudinal Changes in Visual Acu-
ity and Health-related Quality of Life: The Los Angeles Latino Eye Study. Ophthal-
mology [electronic article]. 2010;117(10):1900–1907.e1. (http://www.aaojournal.org/
article/S0161-6420(10)00132-6/abstract). (Accessed March 19, 2018)
15. Linder M, Chang TS, Scott IU, et al. Validity of the visual function index (vf-
14) in patients with retinal disease. Archives of Ophthalmology [electronic article].
1999;117(12):1611–1616. (http://dx.doi.org/10.1001/archopht.117.12.1611)
16. Mangione CM, Phillips RS, Lawrence MG, et al. Improved Visual Function
and Attenuation of Declines in Health-Related Quality of Life After Cataract
Extraction. Archives of Ophthalmology [electronic article]. 1994;112(11):1419–
1425. (https://jamanetwork.com/journals/jamaophthalmology/fullarticle/640890).
(Accessed March 19, 2018)
17. Health- and vision-related quality of life among patients with choroidal neovascular-
ization secondary to age-related macular degeneration at enrollment in randomized
trials of submacular surgery: SST report no. 4. American Journal of Ophthalmol-
ogy [electronic article]. 2004;138(1):91–108. (http://www.sciencedirect.com/science/
article/pii/S0002939404001370)
18. Parrish RK, Gedde SJ, Scott IU, et al. Visual function and quality of life
among patients with glaucoma. Archives of Ophthalmology (Chicago, Ill.: 1960).
1997;115(11):1447–1455.
162
19. Scott IU, Smiddy WE, Schiffman J, et al. Quality of life of low-vision patients and the
impact of low-vision services. American Journal of Ophthalmology [electronic article].
128(1):54–62. (http://dx.doi.org/10.1016/S0002-9394(99)00108-7). (Accessed March
20, 2018)
20. Healthy people 2020: vision. Healthy People 2020. (https://www.healthypeople.gov/
2020/topics-objectives/topic/vision)
21. Lamoureux E, Pesudovs K. Vision-Specific Quality-of-Life Research: A Need
to Improve the Quality. American Journal of Ophthalmology [electronic arti-
cle]. 2011;151(2):195–197.e2. (https://doi.org/10.1016/j.ajo.2010.09.020). (Accessed
March 18, 2020)
22. Coleman AL, Yu F, Keeler E, et al. Treatment of Uncorrected Refractive Error
Improves Vision-Specific Quality of Life. Journal of the American Geriatrics Society
[electronic article]. 2006;54(6):883–890. (http://onlinelibrary.wiley.com/doi/abs/10.
1111/j.1532-5415.2006.00817.x). (Accessed March 23, 2020)
23. BourneRR,StevensGA,WhiteRA,etal. Causesofvisionlossworldwide, 1990–2010:
A systematic analysis. The lancet global health. 2013;1(6):e339–e349.
24. Varma R, Chung J, Foong AWP, et al. Four-Year Incidence and Progression of Vi-
sual Impairment in Latinos: The Los Angeles Latino Eye Study (LALES). Amer-
ican journal of ophthalmology [electronic article]. 2010;149(5):713–727. (https:
//www.ncbi.nlm.nih.gov/pmc/articles/PMC2919324/). (Accessed January 26, 2019)
25. Varma R, Kim JS, Burkemper BS, et al. Prevalence and causes of visual impairment
and blindness in chinese american adults: The chinese american eye study. Jama
Ophthalmology. 2016;134(7):785–793.
163
26. Munoz B, West SK, Rubin GS, et al. Causes of blindness and visual impairment
in a population of older americans: The salisbury eye evaluation study. Archives of
ophthalmology. 2000;118(6):819–825.
27. Tang Y, Wang X, Wang J, et al. Prevalence and causes of visual impairment in a
chinese adult population: The taizhou eye study. Ophthalmology. 2015;122(7):1480–
1488.
28. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindness in a rural chinese adult population: The handan eye study. Ophthalmology.
2008;115(11):1965–1972.
29. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in beijing: The beijing eye study. Ophthalmology. 2006;113(7):1134–e1.
30. ZhaoJ,XuX,EllweinLB,etal. Causesofvisualimpairmentandblindnessinthe2006
and 2014 nine-province surveys in rural china. American Journal of Ophthalmology.
2019;197:80–87.
31. AtteboK,MitchellP,SmithW.Visualacuityandthecausesofvisuallossinaustralia:
The blue mountains eye study. Ophthalmology. 1996;103(3):357–364.
32. Hyman L, Wu S-Y, Connell AM, et al. Prevalence and causes of visual impairment
in the barbados eye study. Ophthalmology. 2001;108(10):1751–1756.
33. Klaver CC, Wolfs RC, Vingerling JR, et al. Age-specific prevalence and causes
of blindness and visual impairment in an older population: The rotterdam study.
Archives of ophthalmology. 1998;116(5):653–658.
34. Rodriguez J, Sanchez R, Munoz B, et al. Causes of blindness and visual impairment
in a population-based sample of US hispanics. Ophthalmology. 2002;109(4):737–743.
164
35. CongdonN,O’ColmainB,KlaverC,etal. Causesandprevalenceofvisualimpairment
among adults in the united states. Archives of Ophthalmology (Chicago, Ill.: 1960).
2004;122(4):477–485.
36. VanNewkirk MR, Weih L, McCarty CA, et al. Cause-specific prevalence of bilateral
visual impairment in victoria, australia: The visual impairment project. Ophthalmol-
ogy. 2001;108(5):960–967.
37. 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. Investigative Ophthalmology & Visual Science. 2002;43(11):3393–
3398.
38. Unzueta M, Globe D, Wu J, et al. Compliance with recommendations for follow-up
care in latinos: The los angeles latino eye study. Ethnicity & disease. 2004;14:285–91.
39. Varma R, Mohanty SA, Deneen J, et al. Burden and predictors of undetected eye
disease in mexican-americans. Medical care. 2008;46(5):497–506.
40. Patty LE, Morales LS, Lai M, et al. Self-reported utilization of eye care among latinos
in the los angeles latino eye study (LALES). Investigative Ophthalmology & Visual
Science. 2009;50(13):2502–2502.
41. Varma R, Torres M, McKean-Cowdin R, et al. Prevalence and risk factors for re-
fractive error in adult chinese americans: The chinese american eye study. American
Journal of Ophthalmology. 2017;175:201–212.
42. Jiang X, Varma R, Torres M, et al. Self-reported use of eye care among adult chinese
americans in the chinese american eye study. American journal of ophthalmology.
2017;176:183–193.
165
43. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by hispanics: Findings from HHANES 1982-84. American Journal of
Public Health. 1990;80(Suppl):11–19.
44. Zheng Y, Lamoureux EL, Ikram MK, et al. Impact of migration and acculturation
on prevalence of type 2 diabetes and related eye complications in indians living in a
newly urbanised society. PloS One. 2012;7(4):e34829.
45. Streiner DL, Cairney J, Norman GR. Health measurement scales: A practical guide
to their development and use. Oxford Univ. Press; 2015.
46. Ayala RJde. The theory and practice of item response theory. Guildford Press; 2009.
47. Shrier I, Platt RW. Reducing bias through directed acyclic graphs. BMC Medical
Research Methodology [electronic article]. 2008;8:70. (https://www.ncbi.nlm.nih.gov/
pmc/articles/PMC2601045/). (Accessed November 21, 2018)
48. Hernán MA, Robins JM. Causal Inference. Boca Raton: Chapman Hall/CRC; 2019.
49. Steinberg EP, Tielsch JM, Schein OD, et al. The VF-14: An Index of Functional
ImpairmentinPatientsWithCataract. Archives of Ophthalmology [electronicarticle].
1994;112(5):630–638. (https://doi.org/10.1001/archopht.1994.01090170074026)
50. Mangione CM, Phillips RS, Seddon JM, et al. Development of the ’Activities of Daily
Vision Scale’. A measure of visual functional status. Medical Care. 1992;30(12):1111–
1126.
51. Margolis MK, Coyne K, Kennedy-Martin T, et al. Vision-specific instruments for the
assessment of health-related quality of life and visual functioning: A literature review.
PharmacoEconomics. 2002;20(12):791–812.
166
52. Mangione CM, Lee PP, Gutierrez PR, et al. Development of the 25-list-item
National Eye Institute Visual Function Questionnaire. Archives of Ophthalmol-
ogy [electronic article]. 2001;119(7):1050–1058. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/267134). (Accessed March 19, 2018)
53. Mangione CM, Lee PP, Pitts J, et al. Psychometric Properties of the National
Eye Institute Visual Function Questionnaire (NEI-VFQ). Archives of Ophthalmology
[electronic article]. 1998;116(11):1496–1504. (https://jamanetwork.com/journals/
jamaophthalmology/fullarticle/264225). (Accessed March 19, 2018)
54. Pesudovs K, Gothwal VK, Wright T, et al. Remediating serious flaws in the Na-
tional Eye Institute Visual Function Questionnaire. Journal of Cataract & Refrac-
tive Surgery [electronicarticle]. 2010;36(5):718–732. (http://linkinghub.elsevier.com/
retrieve/pii/S0886335010002531). (Accessed March 12, 2018)
55. Kline RB. Principles and practice of structural equation modeling. The Guilford
Press; 2016.
56. Bentler PM. On tests and indices for evaluating structural models. Personal-
ity and Individual Differences [electronic article]. 2007;42(5):825–829. (http://
www.sciencedirect.com/science/article/pii/S0191886906003898). (Accessed March
18, 2020)
57. Brucker PS, Yost K, Cashy J, et al. General population and cancer pa-
tient norms for the functional assessment of cancer therapy-general (FACT-
g). Evaluation & the Health Professions [electronic article]. 2005;28(2):192–
211. (%20%0A%20%20%20%20%20%20%20%20https://doi.org/10.1177/
0163278705275341%0A%20%20%20%20%0A)
167
58. Wright BD. Solving MEASUREMENT PROBLEMS WITH THE Rasch MODEL.
Journal of Educational Measurement [electronic article]. 1977;14(2):97–116. (https:
//onlinelibrary.wiley.com/doi/abs/10.1111/j.1745-3984.1977.tb00031.x)
59. Harrell F. Regression modeling strategies: With applications to linear models, logistic
and ordinal regression, and survival analysis. springer; 2016.
60. Medeiros FA, Gracitelli CPB, Boer ER, et al. Longitudinal Changes in Quality of
Life and Rates of Progressive Visual Field Loss in Glaucoma Patients. Ophthalmol-
ogy [electronic article]. 122(2):293–301. (http://dx.doi.org/10.1016/j.ophtha.2014.08.
014). (Accessed March 20, 2018)
61. Association WM. World Medical Association Declaration of Helsinki: Ethical Prin-
ciples for Medical Research Involving Human Subjects. JAMA [electronic article].
2013;310(20):2191–2194. (https://doi.org/10.1001/jama.2013.281053)
62. GlobeDR,Schoua-GlusbergA,PazS,etal. Usingfocusgroupstodevelopaculturally
sensitive methodology for epidemiological surveys in a Latino population: Findings
from the Los Angeles Latino Eye Study (LALES). Ethnicity & disease [electronic
article]. 2002;12(2):259–266. (http://europepmc.org/abstract/med/12019936). (Ac-
cessed March 19, 2018)
63. Solis JM, Marks G, Garcia M, et al. Acculturation, access to care, and use of pre-
ventive services by Hispanics: Findings from HHANES 1982-84. American Journal
of Public Health [electronic article]. 1990;80(Suppl):11–19. (https://doi.org/10.2105/
AJPH.80.Suppl.11). (Accessed March 20, 2018)
64. Brody BL, Gamst AC, Williams RA, et al. Depression, visual acuity, comorbidity,
and disability associated with age-related macular degeneration. Ophthalmology [elec-
tronicarticle]. 108(10):1893–1900. (http://dx.doi.org/10.1016/S0161-6420(01)00754-
0). (Accessed March 20, 2018)
168
65. Globe DR, Varma R, Torres M, et al. Self-reported comorbidities and visual function
inapopulation-basedstudy: Thelosangeleslatinoeyestudy. Archives of Ophthalmol-
ogy [electronic article]. 2005;123(6):815–821. (http://dx.doi.org/10.1001/archopht.
123.6.815)
66. Linn Bernard S., Linn Margaret W., Gurel Lee. Cumulative illness rating scale.
Journal of the American Geriatrics Society [electronic article]. 2015;16(5):622–626.
(https://onlinelibrary.wiley.com/doi/full/10.1111/j.1532-5415.1968.tb02103.x). (Ac-
cessed March 19, 2018)
67. Azen SP, Varma R, Preston-Martin S, et al. Binocular Visual Acuity Sum-
mation and Inhibition in an Ocular Epidemiological Study: The Los Angeles
Latino Eye Study. Investigative Ophthalmology & Visual Science [electronic ar-
ticle]. 2002;43(6):1742–1748. (http://iovs.arvojournals.org/article.aspx?articleid=
2123581). (Accessed March 19, 2018)
68. Varma R, Fraser-Bell S, Tan S, et al. Prevalence of age-related macular degen-
eration in Latinos: The Los Angeles Latino Eye Study. Ophthalmology [elec-
tronic article]. 2004;111(7):1288–1297. (http://www.aaojournal.org/article/S0161-
6420(04)00313-6/abstract). (Accessed March 19, 2018)
69. Varma R, Torres M, Peña F, et al. Prevalence of diabetic retinopathy in adult
Latinos: The Los Angeles Latino Eye Study. Ophthalmology [electronic article].
2004;111(7):1298–1306. (http://www.aaojournal.org/article/S0161-6420(04)00336-
7/abstract). (Accessed March 19, 2018)
169
70. Ware JE, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: Construction
ofScalesandPreliminaryTestsofReliabilityandValidity. Medical Care [electronicar-
ticle]. 1996;34(3):220. (https://journals.lww.com/lww-medicalcare/Abstract/1996/
03000/A_12_Item_Short_Form_Health_Survey__Construction.3.aspx). (Ac-
cessed March 19, 2018)
71. Ware J, A. Kosinski M, D. Keller S. SF-12: How to Score the SF-12 Physical and
Mental Health Summary Scales. 1998;
72. Rosenthal R. Parametric measures of effect size. In: The handbook of research syn-
thesis. New York, NY, US: Russell Sage Foundation; 1994:231–244.
73. Cohen J. A power primer. Psychological bulletin [electronic article]. 1992;112(1):155–
159. (http://libproxy.usc.edu/login?url=https://search-proquest-com.libproxy2.usc.
edu/docview/614317877?accountid=14749)
74. Hays RD, Brodsky M, Johnston MF, et al. Evaluating the Statistical Significance
of Health-Related Quality-Of-Life Change in Individual Patients. Evaluation & the
Health Professions [electronic article]. 2005;28(2):160–171. (https://doi.org/10.1177/
0163278705275339). (Accessed March 19, 2018)
75. Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epi-
demiology (Cambridge, Mass.). 1999;10(1):37–48.
76. Greenland S. Quantifying biases in causal models: Classical confounding vs collider-
stratification bias. Epidemiology (Cambridge, Mass.). 2003;14(3):300–306.
77. Buuren S van. Flexible imputation of missing data. CRC Press; 2018.
170
78. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained
equations in r. Journal of Statistical Software [electronic article]. 2011;45(3):1–67.
(https://www.jstatsoft.org/v45/i03/)
79. R Core Team. R: A language and environment for statistical computing. Vienna,
Austria: RFoundationforStatisticalComputing; 2017.(https://www.R-project.org/)
80. Rosseel Y. lavaan: An R package for structural equation modeling. Journal of Sta-
tistical Software [electronic article]. 2012;48(2):1–36. (http://www.jstatsoft.org/v48/
i02/)
81. Epskamp S, Simon Stuber with contributions from. semPlot: Path diagrams and
visual analysis of various SEM packages’ output. 2017.(https://CRAN.R-project.
org/package=semPlot)
82. Harrell Jr FE. Rms: Regression modeling strategies. 2018.(https://CRAN.R-project.
org/package=rms)
83. Textor J, van der Zander B. Dagitty: Graphical analysis of structural causal models.
2016.
84. Barrett M. Ggdag: Analyze and create elegant directed acyclic graphs. 2018.
85. Wickham H. ggplot2: Elegant graphics for data analysis. Springer-Verlag New York;
2016.(http://ggplot2.org)
86. Alqudah A, Mansberger SL, Gardiner SK, et al. Vision-related quality of life in glau-
coma suspect or early glaucoma patients. Journal of glaucoma [electronic article].
2016;25(8):629–633. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4975427/).
(Accessed March 12, 2018)
171
87. Fenwick EK, Man RK, Ong P, et al. Association of changes in visual acuity
with vision-specific functioning in the singapore malay eye study. JAMA Oph-
thalmology [electronic article]. 2016;134(11):1299–1305. (http://dx.doi.org/10.1001/
jamaophthalmol.2016.3692)
88. Responsiveness of the National Eye Institute Visual Function Questionnaire
to Changes in Visual Acuity. Archives of ophthalmology [electronic article].
2003;121(4):531–539. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1434793/).
(Accessed March 19, 2018)
89. Lindblad AS, Clemons TE. Responsiveness of the National Eye Institute Visual Func-
tion Questionnaire to progression to advanced age-related macular degeneration, vi-
sion loss, and lens opacity: AREDS Report no. 14. Archives of Ophthalmology
(Chicago, Ill.: 1960). 2005;123(9):1207–1214.
90. Hays RD, Morales LS, Reise SP. Item Response Theory and Health Outcomes Mea-
surementin the21stCentury. Medical care [electronicarticle]. 2000;38(9Suppl):II28–
II42. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1815384/). (Accessed
March 19, 2018)
91. Wittenborn J, Rein D. The economic burden of vision loss and eye disorders in the
United States. University of Chicago; 2013.(https://www.preventblindness.org/sites/
default/files/national/documents/Economic%20Burden%20of%20Vision%20Final%
20Report_130611.pdf)
92. Wang X, Lamoureux E, Zheng Y, et al. Health Burden Associated with Visual Im-
pairment in Singapore: The Singapore Epidemiology of Eye Disease Study. Ophthal-
mology [electronic article]. 2014;121(9):1837–1842. (http://www.sciencedirect.com/
science/article/pii/S0161642014002413). (Accessed March 23, 2020)
172
93. Tsai S-Y, Chi L-Y, Cheng C-Y, et al. The Impact of Visual Impairment and Use of
Eye Services on Health-Related Quality of Life among the Elderly in Taiwan: The
Shihpai Eye Study. Quality of Life Research [electronic article]. 2004;13(8):1415–
1424. (https://doi.org/10.1023/B:QURE.0000040791.87602.fe). (Accessed March 23,
2020)
94. Vision impairment and major eye diseases reduce vision-specific emotional well-being
in a Chinese population - ProQuest. (https://search-proquest-com.libproxy1.usc.
edu/docview/1894846003?rfr_id=info%3Axri%2Fsid%3Aprimo). (Accessed March
23, 2020)
95. Varma R, Kim JS, Burkemper BS, et al. Prevalence and Causes of Visual Impairment
andBlindnessinChineseAmericanAdults: TheChineseAmericanEyeStudy. JAMA
Ophthalmology [electronic article]. 2016;134(7):785–793. (https://jamanetwork-
com.libproxy1.usc.edu/journals/jamaophthalmology/fullarticle/2523778). (Accessed
March 21, 2020)
96. Hsu W-M, Cheng C-Y, Liu J-H, et al. Prevalence and causes of visual impairment
in an elderly Chinese population in Taiwan: The Shihpai Eye Study. Ophthalmology.
2004;111(1):62–69.
97. Huang S, Zheng Y, Foster PJ, et al. Prevalence and causes of visual impairment in
Chinese adults in urban southern China. Archives of Ophthalmology (Chicago, Ill.:
1960). 2009;127(10):1362–1367.
98. Saw S-M, Foster PJ, Gazzard G, et al. Causes of blindness, low vision, and
questionnaire-assessed poor visual function in Singaporean Chinese adults: The Tan-
jong Pagar Survey. Ophthalmology. 2004;111(6):1161–1168.
173
99. Liang YB, Friedman DS, Wong TY, et al. Prevalence and causes of low vision and
blindnessinaruralchineseadultpopulation: TheHandanEyeStudy. Ophthalmology.
2008;115(11):1965–1972.
100. Zhao J, Ellwein LB, Cui H, et al. Prevalence of vision impairment in older adults in
rural China: The China Nine-Province Survey. Ophthalmology. 2010;117(3):409–416,
416.e1.
101. Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment in urban and
rural areas in Beijing: The Beijing Eye Study. Ophthalmology. 2006;113(7):1134.e1–
11.
102. Varma R, Hsu C, Wang D, et al. The Chinese American Eye Study: Design and
Methods. Ophthalmic Epidemiology [electronic article]. 2013;20(6):335–347. (http:
//www.tandfonline.com/doi/full/10.3109/09286586.2013.823505). (Accessed March
21, 2020)
103. SuinnRM,AhunaC,KhooG.TheSuinn-LewAsianSelf-IdentityAcculturationScale:
Concurrent and factorial validation. Educational and Psychological Measurement.
1992;52(4):1041–1046.
104. Burkemper B, Torres M, Jiang X, et al. Factors Associated with Visual Impairment in
Chinese American Adults: The Chinese American Eye Study. Ophthalmic Epidemiol-
ogy [electronic article]. 2019;26(5):329–335. (https://doi.org/10.1080/09286586.2019.
1622737). (Accessed March 22, 2020)
105. Linden WJ van der. Handbook of item response theory. Chapman & Hall/CRC;
2019.
106. Wood SN. Generalized additive models: An introduction with r. CRC Press/Taylor
& Francis Group; 2017.
174
107. Rizopoulos D. Ltm: An r package for latent variable modelling and item response
theory analyses. Journal of Statistical Software [electronic article]. 2006;17(5):1–25.
(http://www.jstatsoft.org/v17/i05/)
108. Fenwick EK, Ong PG, Sabanayagam C, et al. Assessment of the psychometric proper-
ties of the Chinese Impact of Vision Impairment questionnaire in a population-based
study: FindingsfromtheSingaporeChineseEyeStudy. Qualityof LifeResearch [elec-
tronic article]. 2016;25(4):871–880. (https://doi.org/10.1007/s11136-015-1141-1).
(Accessed March 23, 2020)
109. Rees G, Tee HW, Marella M, et al. Vision-specific distress and depressive symptoms
in people with vision impairment. Investigative Ophthalmology & Visual Science.
2010;51(6):2891–2896.
110. Morales LS, Varma R, Paz SH, et al. Self-reported use of eye care among Latinos:
The Los Angeles Latino Eye Study. Ophthalmology. 2010;117(2):207–215.e1.
111. Chang W, Cheng J, Allaire J, et al. Shiny: Web application framework for r.
2020.(https://CRAN.R-project.org/package=shiny)
175

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Age related macular degeneration in Latinos: risk factors and impact on quality of life

PDF

Chronic eye disease epidemiology in the multiethnic ophthalmology cohorts of California study

PDF

Visual acuity outcomes after cataract extraction in Chinese Americans: the Chinese American Eye Study (CHES)

PDF

Characterizing the genetic and environmental contributions to ocular and central nervous system health

PDF

Assessing the impact of air pollution on adverse birth outcomes in a low resource setting

PDF

Antibiotic resistance in a large ICU cohort: 1995-2002

PDF

The role of steroid hormones in the etiology of urologic diseases

PDF

Hormone therapy timing hypothesis and atherosclerosis

PDF

The impact of visual field impairment on health related quality of life: a population-based study

PDF

Genes and hormonal factors involved in the development or recurrence of breast cancer

PDF

Methodological approaches to assessing diurnal cortisol rythms in epidemiological studies: how many salivary samples are necessary?

PDF

Genetic and environmental risk factors for childhood cancer

PDF

The role of heritability and genetic variation in cancer and cancer survival

PDF

The longitudinal risk factors of diabetic retinopathy: the Los Angeles Latino Eye Study

PDF

Arm lymphedema in a multi-ethnic cohort of female breast cancer survivors

PDF

Factors that influence mammographic density: role of estrogen metabolism genes, biomarkers of inflammation, and lifestyle

PDF

Body size and the risk of prostate cancer in the multiethnic cohort

PDF

Investigating a physiological pathway for the effect of guided imagery on insulin resistance

PDF

The environmental and genetic determinants of cleft lip and palate in the global setting

PDF

Identifying genetic, environmental, and lifestyle determinants of ethnic variation in risk of pancreatic cancer

Asset Metadata

Creator Barrett, Malcolm Eugene, III (author)

Core Title Vision epidemiology and the impact of vision loss on vision-specific quality of life

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Epidemiology

Publication Date 03/04/2021

Defense Date 02/02/2021

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

Tag blindness,Epidemiology,OAI-PMH Harvest,Quality of life,vision,vision loss

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Marjoram, Paul (committee chair), Chou, Chih-Ping (committee member), Cortessis, Victoria (committee member), McKean-Cowdin, Roberta (committee member), Richter, Grace (committee member)

Creator Email malcoleb@usc.edu,malcolmbarrett@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-424861

Unique identifier UC11666610

Identifier etd-BarrettMal-9304.pdf (filename),usctheses-c89-424861 (legacy record id)

Legacy Identifier etd-BarrettMal-9304.pdf

Dmrecord 424861

Document Type Dissertation

Rights Barrett, Malcolm Eugene, III

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