Does Functional Vision Behave Differently in Low-Vision Patients with Diabetic Retinopathy?—A Case-Matched Study

Abstract

purpose. A retrospective case-matched study designed to compare patients with diabetic retinopathy (DR) and other ocular diseases, managed in a low-vision clinic, in four different types of functional vision.

methods. Reading, mobility, visual motor, and visual information processing were measured in the patients (n = 114) and compared with those in patients with other ocular diseases (n = 114) matched in sex, visual acuity (VA), general health status, and age, using the Activity Inventory as a Rasch-scaled measurement tool. Binocular distance visual acuity was categorized as normal (20/12.5–20/25), near normal (20/32–20/63), moderate (20/80–20/160), severe (20/200–20/400), profound (20/500–20/1000), and total blindness (20/1250 to no light perception). Both Wilcoxon matched pairs signed rank test and the sign test of matched pairs were used to compare estimated functional vision measures between DR cases and controls.

results. Cases ranged in age from 19 to 90 years (mean age, 67.5), and 59% were women. The mean visual acuity (logMar scale) was 0.7. Based on the Wilcoxon signed rank test analyses and after adjusting the probability for multiple comparisons, there was no statistically significant difference (P > 0.05) between patients with DR and control subjects in any of four functional visions. Furthermore, diabetic retinopathy patients did not differ (P > 0.05) from their matched counterparts in goal-level vision-related functional ability and total visual ability.

conclusions. Visual impairment in patients with DR appears to be a generic and non–disease-specific outcome that can be explained mainly by the end impact of the disease in the patients’ daily lives and not by the unique disease process that results in the visual impairment.

One of the most serious and frequent complications of diabetes mellitus (DM) is diabetic retinopathy (DR), a retinal vascular disorder that is characterized by signs of retinal ischemia (hemorrhages, cotton-wool spots, and revascularization) and/or signs of increased retinal vascular permeability. ¹ From the patients’ point of view, vision loss caused by retinopathy is one of the most feared late complications of diabetes. ² DM affects approximately 18 million U.S. adults, of whom an estimated 30% have undiagnosed diabetes, and imposes an increased risk for eye disease. ³ Approximately 4.1 million U.S. adults 40 years and older have diabetic retinopathy; 1 of every 12 persons with DM in this age group has advanced, vision-threatening retinopathy. ⁴ Future projections suggest that diabetic retinopathy will increase as a public health problem, both as the U.S. population ages and as age-specific prevalence of DM increases over time. Furthermore, diabetic retinopathy among Americans older than 40 years results in substantial costs for the U.S. economy. ⁵  

Despite advances in medical and surgical management, patients with vision loss from DR continue to constitute a significant portion of the patients being seen in low-vision rehabilitation service settings. These patients present special challenges to low-vision providers because of early onset, fluctuations in vision loss and its overall complex nature, unique visual demands of disease management, and associated multisystem losses. ⁶ People with diabetic retinopathy have sometimes been included among patient samples that contribute to questionnaire content, ⁷ but no patient-reported outcome measure has yet been developed or validated ⁸ in a population of people with diabetic retinopathy for use by this specific patient group. The individualized Audit of Diabetes-Dependent Quality of Life (ADDQoL) ⁹ measure does not specifically target the impact of diabetic eye problems on quality of life. ⁸  

The unique and complex needs of people with diabetes who experience vision loss can be well measured through a disease-specific vision-related ability measurement tool. The first step is to compare the functional vision of these patients, as measured by a generic low-vision measurement tool, with their matched counterparts who have other causes of visual impairment. The main question is whether a unique pathophysiologic mechanism for vision loss in diabetic retinopathy—combined with diabetic needs to measure medication, insulin, and blood glucose levels and to maintain skin care, diet, exercise, and transportation—translate into a specific self-perceived vision-related disability pattern that differs from other common causes of visual impairment. Should vision-related disability be differentiated by etiology? An understanding of this difference will help research professionals to customize clinical outcome measures for research according to the impact of DR on patients’ daily activities versus specific pattern of vision-related disability in these patients. 

Clinicians and researchers must know whether an estimated rate of visual function in DR reflects its unique pathophysiologic identity and daily life demand or instead measures generic vision-related functioning such as other causes of visual impairment. Diabetes alone is a debilitating chronic disease that can aggravate the impact of diabetic retinopathy on patients’ functional vision. Low-vision patients with diabetic retinopathy may have a vision-related rehabilitative potential that differs from their matched nondiabetic counterparts and should be supported by a disease-specific rehabilitation plan. 

Mangione et al. ¹⁰ found that patients with DR reported a higher number of visual problems than do patients who had age-related macular degeneration (ARMD), glaucoma, cataracts, cytomegalovirus retinitis, or low vision of any cause. One study ¹¹ has shown that visually impaired individuals with diabetes reported more disruptions to functional activities than those without diabetes at first interview. This difference disappeared at 1 year, but not as a result of any improvement in diabetes; rather, nondiabetic individuals reported more disruptions. ¹¹  

The goal of our study was to compare specific domains of functional vision in low-vision patients with DR and other ocular diseases. These comparisons were made using a Rasch-scaled outcome measurement model within a subgroup of matched patients with similar sex, visual acuity, general health status, and age. This study provides a quantitative description of the relationship between disease characteristics and functional vision impairment in patients with low-vision. 

Methods

Subjects and Study Design

The subjects in this study were low-vision patients enrolled between February 2001 and December 2006 before their first visit to the Wilmer Low Vision Service. In a health-related intake questionnaire, all subjects answered demographic questions about age, duration of vision impairment, diabetes mellitus, other health conditions, and limitations on performing daily activities attributed to other conditions. All information contained within this questionnaire became the patient’s record in the intake history database. Immediately thereafter, participants underwent eye examinations that included measurement of visual acuity and refraction as a routine part of their appointments at the low-vision service. The results of the eye examination created the second database of eye physical examination findings, including the main cause of visual impairment recorded as ICD-9 diagnostic codes. Before their first visit to the Wilmer Low Vision Rehabilitation Service, consenting participants also completed an interviewer-administered functional vision measurement tool, the Activity Inventory (AI). ^{12 13} The results of this assessment were recorded in the AI database. 

All three databases were reviewed to create a comprehensive dataset of the patients’ clinical, demographic, and functional vision characteristics. The study cohort consisted of 114 patients who were considered diabetic retinopathy (DR) cases, based on ICD-9 diagnostic codes of 362.1 (background diabetic retinopathy) and 362.2 (proliferative diabetic retinopathy) and 976 low-vision patients with other diagnostic codes. To develop a one-to-one matched set, we created all possible combinations of 114 patients with DR and nonpatients with DR without diabetes (818 individuals), for which subjects were the same sex and same visual acuity class (World Health Organization Classification). ¹⁴ Thereafter, remaining potential matched pairs were ordered by the similarity of self-reported general health status, age (within 10- to 15-year units), and visual acuity in logMAR. Finally, using this ordering, we selected potential matched pairs if neither member was previously chosen for what could be considered a more similar pair. A total of 114 patients in each of the DR and non-DR groups were selected according to this algorithm (Fig. 1) . With this matching mechanism, almost 60% of all case–control pairs had exactly the same visual acuity in logMAR. 

All subjects signed a HIPAA (Health Insurance Portability and Accountability Act) authorization and a written informed consent, both of which were read to them by clinic staff. The study was approved by the Johns Hopkins Institutional Review Board and adhered to the tenets of the Declaration of Helsinki. 

Measurement Tools

Functional vision was measured by the AI and scaled using a Rasch model. A detailed description of the AI has been published elsewhere. ^{15 16} The AI has a hierarchical structure based on a theoretical framework, the Activity Breakdown Structure (ABS). ^{15 16} The ABS views the functional vision state as a constellation of vision-related activities. At the most primitive level of the ABS, specific cognitive and motor activities (e.g., sewing a hem) are identified as Tasks. At the next level of the ABS, Tasks have Goals—that is, Tasks are performed for a reason. The list of Tasks customarily performed to achieve a Goal can vary from person to person. Therefore, in the AI, the respondent must specify whether a Task is applicable to the parent Goal before rating its difficulty. Vision-related ability to perform tasks can also be measured according to the visual function needed for that specific task. For example, reading recipes, newspaper articles, price tags, and signs are activities performed under reading function. Climbing stairs, crossing streets, hiking, and walking to the store are instances of mobility function. If any of these tasks cannot be performed, a person experiences a measurable limitation in functional vision for mobility. 

In its current configuration, the AI consists of approximately 460 Tasks that are nested under the 49 Goals. The AI was administered adaptively by a computer-assisted interview. First, the respondent was asked to choose from among four ordered-response categories and rate the importance of the first Goal (not important, slightly important, moderately important, or very important). If the Goal is rated not important, then the interviewer moves on to the next Goal; otherwise, the respondent is asked to rate the difficulty of the Goal using one of five ordered response categories (not difficult, slightly difficult, moderately difficult, very difficult, or impossible). If the Goal is rated not difficult, then the interviewer moves on to the next Goal; otherwise, the respondent is asked to rate the difficulty of each of a list of subsidiary Tasks (using the same response categories for rating Goal difficulty) or identify the Task as not applicable. Because of the response contingencies, each person receives an individualized visual function measurement tool that elicits a structured functional history. Cutting across Goals and Objectives, each of the 460 Tasks was classified as representing one of four types of functional vision: reading, visual motor (i.e., visually guided manipulation), visual information processing (i.e., seeing) or mobility. In this study, responses to the AI questionnaire items were recorded for all Goal-level items (49 items) and for four domains including mobility (49 items), reading (118 items), visual motor (173 items), and visual information processing (119 items) (Fig. 1) . 

Visual acuity of the better eye or worse eye, and weighted acuity of both eyes were obtained using the Early Treatment Diabetic Retinopathy Study (ETDRS) chart. Visual acuity ratings were converted to the logarithm of the minimum angle of resolution (logMAR) scale. ¹⁷ All visual acuities were rounded to the nearest 0.1-logMAR unit. Using the ICD-9 visual impairment scale, binocular distance visual acuity was categorized as normal (20/12.5–20/25), near normal (20/32–20/63), moderate (20/80–20/160), severe (20/200–20/400), profound (20/500–20/1000), and total blindness (20/1250 to no light perception). ¹⁴  

For the present study, a group of self-reported medical and emotional comorbidities and information about patients’ diabetic care were taken from the intake history database. The systemic comorbidity summation index was defined as the summation of a list of eight general, self-reported medical and emotional conditions (a score of 1 for each medical condition). The list included heart disease, high blood pressure, stroke, memory problem, hearing difficulty, speech disorder, depressed mood, and mobility difficulty. 

Analysis

Descriptive statistics were calculated for each variable of interest (Stata 9.2; Stata Software Corp., College Station, TX). The χ² test or the Fisher exact test was used to compare patient characteristics that were categorical. The Wilcoxon two-sample rank sum test and paired t-test were used to compare patient characteristics that were continuous. Both the Wilcoxon matched-pairs signed rank sum test and the sign test of matched pairs were used to compare estimated functional vision measures between DR cases and controls. Because of multiple independent statistical comparisons, Bonferroni adjusted P < 0.008 was considered statistically significant (α < 0.05). 

Each of the study subjects responded to selected subsets of the 509 items of the AI. All the data that were collected as part of this study were merged into a single database keyed to the subject, and Rasch analysis was performed on the combined data. Separate Rasch analyses were also performed on the responses to each of the subsets of Tasks that made up the four functional domains (reading, mobility, visual motor, and visual information processing) and of the Goal-level items. We included 228 study subjects (case and controls) and 1635 nonstudy subjects in each of these Rasch analyses. Bivariate regressions were performed on each domain item measures estimated from each domain analysis versus corresponding item measures from the merged data analysis and on AI item measures estimated from difficulty ratings of Goals versus the corresponding item measures from the merged data analysis. These regressions were then used to equate person measure estimates from each domain to person measure estimates from AI Goal and Task difficulty ratings. All the Rasch analyses were performed with commercial software (Winsteps, ver. 3.16; Winsteps, Chicago, IL). In each of these Rasch analyses, some of the cases or controls were not measured because of missed responses or extreme values. Hence, all unmeasured subjects and their matched counterparts were excluded from six comparisons. This protocol created a different number of matched pairs in six statistical comparisons related to all Goal-level items and each functional domain (Fig. 1) . Rasch measurement theory is built on the premise that the latent variable—in this case, functional vision—is unidimensional and that the person’s vision-related ability is independent of the visual ability necessary to satisfy the item content. Furthermore, Rasch measurement theory assumes that the observed variable—in this case, the subjects’ ratings for each item—will order the subjects in the same way for each item and order the items in the same way for each person. This pattern of ordering is called a Guttman scale. ^{18 19} However, because it is probabilistic, Rasch measurement theory assumes that the required ordering will be disrupted by random errors in the observed variable. This random disordering of the observed responses produces a probabilistic Guttman scale. ^{18 19} Therefore, if visual ability is not strictly unidimensional, or if other variables contribute to the pattern of the subjects’ responses, then the pattern of observed responses will differ significantly from the expected responses. The degree of departure of observed responses from expected responses is captured in a χ²-like statistic called the infit mean square. 

Results

Study protocol identified 114 case-matched pairs of patients with DR and nonpatients (Fig. 1) . Demographic and clinical characteristics of all subjects, stratified by diabetic retinopathy case and control groups, are shown in Table 1 . The majority of patients were female (59%), and the average age was 67.5 years (range, 19–90). Almost 16% of case–control pairs had severe or profound visual impairment. Only five pairs had total blindness. Most of the matched pairs had moderate (39%) or near-normal visual impairment (32%). Because of a strict matching protocol by sex, visual impairment class, visual acuity (logMAR), general health status, and age, these variable distributions did not differ across case and control groups (P > 0.8). Approximately 27% of patients had graduated only from high school, 38% reported a college degree, and 12% had a graduate degree as the highest level of education. Most patients (52%) rated their general health as fair, and only 3% reported excellent general health. The majority (72%) of patients was retired and used some type of low-vision rehabilitation device. Furthermore, 23% of patients lived alone, and 27% of them were active drivers. No significant differences between groups were detected in level of education, living arrangement, or retirement. We found a statistically significant difference (P = 0.003) in use of low-vision rehabilitation devices between cases and controls. 

Disorder diagnoses in control patients were age-related macular degeneration (53%), glaucoma (13%), retinal detachments (6%), optic atrophy (6%), CVA or brain injury (2%), inherited retinal degenerations (3%), refractive disorders (2%), corneal disorders (5%), ocular injuries (3%), and other miscellaneous disorders (7%). 

The mean duration of diabetes in patients with diabetic retinopathy was 20.5 ± 9.8 years. Almost 40% of low-vision patients with diabetic retinopathy had diabetes for more than 20 years. Most of these patients reported self-monitoring blood sugar (87.7%) and a requirement for insulin injection (61.4%). 

The most common self-reported medical conditions in both case and control groups were heart disease, mobility limitation, and memory difficulty (Table 2) . The mean ± SD of the comorbidity numbers was 3.4 ± 1.5 for diabetic retinopathy cases and 1.9 ± 1.4 for controls. There was a statistically significant difference in number of comorbidities between the two groups (P < 0.0000). In addition, a greater percentage of matched DR cases reported heart disease (P = 0.000) versus control patients (Table 2) . Almost 100% of DR cases and 85% percent of controls reported 1 or more medical conditions in addition to the main cause of their visual impairment (Fig. 2) . 

Table 3shows the results of six separate Rasch analyses performed on all items, Goal-level items, and four types of functional vision, based on the study protocol. Rasch analysis provided estimates of the visual ability of each subject (person measure), the visual ability required by each item (item measure), χ²-like fit statistics (weighted mean square residuals), and separation reliabilities (ratio of the true variance in the estimated measures to the observed variance). The column labeled Infit MNSQ contains a χ²-weighted statistic (based on observed responses relative to responses expected by the Andrich model, using the estimated step calibrations) for each response category (see additional discussion later). The expected value of the infit MNSQ is 1.0 (i.e., observed response error variance is equal to the average expected response error variance). Values <1.0 indicate that the observed response error variance was less than expected and values >1.0 indicate that the observed response error variance was greater than expected. All the infit MNSQ values were acceptably close to 1.0. 

Rasch analysis revealed the lowest functional vision estimates in reading, compared with other domains (Table 3) . The mean vision-related ability measures in reading were 0.26 ± 1.09 (SD) logit and 0.22 ± 1.19 (SD) for cases and controls, respectively (Table 3) . In all domains, the person measure separation reliability indices were more than 0.8 (i.e., 20% of the observed variance in the person measure distribution can be attributed to estimation error). The item measure separation reliability in all Rasch analyses was more than 0.95 (i.e., 5% of the observed variance in the item measure distribution can be attributed to estimation error) (Table 3) . Matched analysis (Wilcoxon signed-rank sum test and sign test of matched pairs) of low-vision patients with diabetic retinopathy versus control groups did not display a statistically significant difference between groups (Fig. 3)in functional vision ability measures in reading, visual information, visual motor, and mobility (p value > 0.05). Furthermore, diabetic retinopathy patients did not differ from their matched counterparts in total visual ability (measured by all items) and functional ability for achieving their goals (P > 0.05). 

Discussion

Using a Rasch scale measurement model in this study, we failed to reject the null hypothesis of no difference between patients with DR and other visually impaired patients in reading, mobility, visual motor, and visual information processing. Visual impairment in patients with DR may be a generic and non–disease-specific outcome that could be explained mainly by the end impact of the disease on the patients’ daily lives, not by the unique disease process resulting in visual impairment. From these data, we can also speculate that considering the unique activity of daily life in patients with low-vision and DR, a generic assessment of visual impairment alone may not be adequate for targeting the rehabilitative potential of these patients. 

The result of this study is in contrast with a view sometimes expressed in diabetes care that visual loss interacts with diabetes to render the individual more vulnerable to the impact of vision loss. ¹¹ Our different result could be explained by the medical model ²⁰ of vision-related disability that views disability as a feature of the vision loss process, directly caused by DR or any other ocular disease. Matching DR cases and controls by class of visual impairment and general health status eliminated the disease-based variability of the model. This matching created similar behavior of functional vision in both groups: cases with DR and controls with other ocular diseases. This medical model–based interpretation shifts the focus from cause to impact and places all common causes of visual impairment on equal footing, allowing them to be compared by using a common metric—the ruler of vision-related disability. ²⁰  

Based on the biopsychosocial model of vision-related disability, however, our result could be a paradox. Functional vision in patients with DR not only refers to all vision system functions, but to activities and participation related to vision. Furthermore, environmental factors that make up the physical, social, and attitudinal environment in which people conduct their lives can interact with activity and participation components. What we measure with a generic measurement tool such as the AI is a latent attribute determined by these factors. The biopsychosocial model views vision-related disability as an outcome of interaction between health conditions (ocular and nonocular diseases) and contextual factors. ²⁰ Among contextual factors are external environmental factors and internal personal factors. There was no significant difference between the two groups in personal factors, such as sex, age, education, and driving. When comparing the number of comorbidities between two groups, we found a significant difference that corresponded to results from other studies. ^{21 22} The presence of more severe retinopathy or visual impairment in diabetic patients is a risk indicator for the increased risk of morbidity from ischemic heart disease and other nonocular complication of diabetes. Compared with other common causes of visual impairment, we may expect more serious adverse impact on the quality of life in diabetic patients. ²³  

Given that all other determining factors including environmental factors (for example, social attitudes, architectural characteristics, and legal and social structures) and other personal factors (coping styles, overall behavior pattern) were the same, a difference in some of health-related measures would create a different level of functional vision in both groups, especially in domains related to daily living. We can consider two reasons for our contrasting results: First, the AI as a generic measurement tool may overlook specific aspects of the impact of diabetic retinopathy and its treatment. Second, we can argue that all patients with DR may have better environmental interventions as part of their general diabetic care, either by eliminating environmental barriers or creating environmental facilitators for expanded performance of actions and tasks in daily living. 

These interventions could balance the impact of poor health conditions in patients with DR, compared with other low-vision patients. There was no difference between two groups in some of the environmental factors, including living arrangement or using supportive services. However, we did not compare other environmental factors between the two groups. Further study is needed to examine the difference in functional vision between the two groups in a standardized environment or to measure the environmentally adjusted ability of the patients using differential item functioning analysis. ²⁴  

Retinopathy is not the only cause of visual loss in diabetes. Cataract and glaucoma are other causes of low vision in this population. ²⁵ Diabetes predisposes to presenile cataract and is also a risk factor for chronic simple glaucoma and neovascular glaucoma. The process of selecting cases and controls based on ICD-9 diagnostic codes may have introduced bias, as some patients with DR may have more than one ocular disease. Despite this weakness, there are significant strengths—most particularly, our study used a case-matched protocol for comparing the two groups. This study also accessed a large, community-based low-vision cohort and has attempted to measure vision-related disability in a Rasch-based interval scale. ²⁶  

The results of this study could advance an increased recognition among low-vision researchers and practitioners that reductions in the incidence and severity of disability in patients with DR can be achieved both by enhancing the functional capacity (reserve) of the patients and by improving their performance by modifying features of the social and physical environment. To analyze the impact of these different low-vision rehabilitative interventions, we need both a generic tool such as the AI and a diabetic retinopathy–specific scale to measure all domains of areas of life, as well as the environmental factors that are important to these patients. Therefore, a great challenge to further research in diabetes is to design and validate scientifically rigorous patient-centered measurement tools suitable specifically for patients who have become visually impaired because of diabetic retinopathy. ²⁶  

 Presented at the ARVO Summer Eye Research Conference, Providence/Warwick, Rhode Island, July 2007.

 Supported by Grant EY012045 from the National Eye Institute/National Institutes of Health, Bethesda, Maryland.

 Submitted for publication November 23, 2007; revised February 22 and April 8, 2008; accepted July 17, 2008.

 Disclosure: L. Ahmadian, None; R. Massof, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Flow chart of patients included and excluded from the analysis. ICD-9 Diagnostic Codes 362.01 (background diabetic retinopathy) or 362.02 (proliferative diabetic retinopathy) were considered to define DR cases.

Table 1.

 

Demographic and Clinical Characteristics of Participants

	Case (n = 114) Mean (SD)	Control (n = 114) Mean (SD)	P	Total (n = 228) Mean (SD)
Age (y)	67.5 (15.8)	67.8 (15)	0.88	67.6 (15.4)
VA (logMAR)	0.7 (0.52)	0.72 (0.56)	0.84	0.71 (0.54)

	Count (%)	Count (%)	P	Count (%)
Sex (female)	67 (59)	67 (59)	1	134 (59)
Education			0.6
Less than high school	23 (21)	29 (25)		52 (0.23)
High school	35 (31)	27 (24)		62 (27)
College	41 (37)	44 (39)		85 (38)
Graduate	13 (11)	14 (12)		27 (12)
VA (class)			1
Normal	9 (8)	9 (8)		18 (8)
Near normal	37 (32)	37 (32)		74 (32)
Moderate	45 (39)	45 (39)		90 (39)
Severe	11 (10)	11 (10)		22 (10)
Profound	7 (6)	7 (6)		14 (6)
Total	5 (4)	5 (4)		10 (4)
Health status			1
Excellent	3 (3)	3 (3)		6 (3)
Good	37 (32)	37 (32)		74 (32)
Fair	60 (53)	60 (53)		120 (52)
Poor	14 (12)	14 (12)		28 (12)
Living alone	25 (22)	28 (25)	0.64	53 (23)
Driving	34 (30)	28 (25)	0.37	62 (27)
Retired	72 (67)	88 (77)	0.08	160 (72)
Using rehabilitation device	92 (81)	72 (63)	0.003	164 (72)

Table 2.

 

Comorbidity Characteristics of Study Participants

Variable	Case (n = 114) Mean (SD)	Control (n = 114) Mean (SD)	P
Comorbidities (n)	3.4 (1.5)	1.9 (1.4)	<0.000

	Count (%)	Count (%)	P
Heart disease	85 (75)	29 (25)	0
Memory difficulty	50 (44)	65 (57)	0.05
Mobility limitation	52 (46)	42 (37)	0.18
Sad/depressed	32 (28)	31 (27)	0.88
Hypertension	22 (19)	21 (18)	0.87
Hearing difficulty	24 (21)	9 (8)	0.005
Stroke	20 (17)	21 (18)	0.86
Speech difficulty	3 (3)	5 (4)	0.47

Figure 2.

 

Frequency distribution of multiple nonocular medical conditions in 114 DR cases and 114 controls.

Table 3.

 

Comparison of Functional Vision Measures in Cases (Diabetic Retinopathy) and Controls

	Items (Measured)	Persons (Measured)	Pairs (Analyzed)	INFIT MNSQ (Person)* Mean (SD)	INFIT MNSQ (Item) Mean (SD)	Person Measure Reliability	Item Measure Reliability	Functional Vision Mean (SD)			Sign Test P	Wilcoxon Rank Test P	t-Test
																			Cases	Controls
All Items	508	228	114	1.03 (0.45)	1.04 (0.38)	0.96	0.97	0.46 (0.87)	0.52 (0.91)	0.92	0.51	0.51
Goals	49	228	114	1.05 (0.45)	1.01 (0.21)	0.95	0.98	0.55 (0.89)	0.62 (0.94)	0.78	0.35	0.55
Reading	118	226	112	1.17 (0.75)	1.05 (0.51)	0.97	0.99	0.26 (1.09)	0.22 (1.19)	0.77	0.88	0.74
Visual information	119	222	108	1.05 (0.57)	1.09 (0.37)	0.88	0.97	0.47 (0.89)	0.62 (0.98)	0.21	0.23	0.18
Visual motor	173	222	108	1 (0.53)	0.99 (0.34)	0.9	0.95	0.37 (1.04)	0.39 (0.93)	0.39	0.83	0.81
Mobility	49	200	89	0.97 (0.63)	1.04 (0.34)	0.83	0.99	0.28 (1.08)	0.44 (1.2)	0.52	0.42	0.33

 Functional vision measures are shown in logit.

*  Infit mean square.

Figure 3.

 

Box-and-whisker plot summaries of the equated functional vision measure (logit) as measured by the AI questionnaire for all items, goals, and domains (Mobility, Reading, Visual Motor, and Visual Information) in patients with DR and controls.