National Eye Institute Visual Function Questionnaire in the Age-Related Eye Disease Study (AREDS): Title and subTitle BreakAREDS Report No. 10 FREE

THE AGE-RELATED Eye Disease Study (AREDS) is an ongoing multicenter study of the natural history of cataract and age-related macular degeneration(AMD). The study incorporated a randomized clinical trial designed to evaluate the effect of micronutrient supplementation on the incidence and progression of these 2 conditions. The main results of this trial have been published.^{1 - 2} A secondary objective of AREDS was to search for clues about the etiology of cataract and AMD and possible strategies for intervention. At the time of enrollment, participants manifested a wide range of ocular findings, varying from no AMD and no lens opacity, to moderately severe lens opacity or advanced unilateral AMD.

Increasing attention has been given to the assessment of health-related quality-of-life (HRQL) outcome measures in clinical trials during the past 2 decades. This assessment is important to obtain a more complete understanding of the effect of treatments, aging, and the natural history of disease on individuals' daily routines.³ The inclusion of HRQL outcomes in research and patient care is appropriate because research focused on developing methods to assess HRQL has shown that patients' self-reports of functioning are reproducible.⁴ Although patients' perceptions of visual functioning and global health outcomes were rarely incorporated into early ophthalmologic research, ⁵ the inclusion of quality-of-life instruments in recent clinical studies^{6 - 10} is increasing our understanding of the day-to-day living of patients with eye disease and vision loss.

Williams and colleagues, ⁹ for example, found that older individuals diagnosed as having AMD causing legal blindness in at least 1 eye were more likely to have lower scores on the Quality of Well-being Scale, ¹¹ reflecting a greater impairment in quality of life compared with similarly aged community adults with normal vision. This same study found that subjects with AMD were more apt to report emotional distress (Profile of Mood States¹² scale) and more likely to need help with daily activities (Instrumental Activities of Daily Living¹³ scale) than a national sample of older individuals. Park¹⁰ provides a summary of the types of vision rehabilitation techniques available for individuals with AMD and emphasizes the importance of education about the disease process and the subsequent adjustments to activities of daily living. Educational tools that teach increased ability to adapt to activities of daily living to individuals with reduced vision will enhance their HRQL.

Recent ophthalmologic research has taken advantage of the development of the National Eye Institute Visual Function Questionnaire (NEI-VFQ)¹⁴ to collect information on HRQL. The NEI-VFQ was designed to measure areas of functioning and well-being that were identified as important by persons with eye disease. This instrument can be used to assess self-reported visual impairment in studies of vision. The questionnaire was constructed to evaluate the effect of visual disability on HRQL across several common eye conditions. Field test results of the instrument supported use of the NEI-VFQ in patients with age-related cataracts, AMD, diabetic retinopathy, primary open-angle glaucoma, and cytomegalovirus retinitis.¹⁵ These field test results demonstrated that the NEI-VFQ was reliable and valid and a useful tool for group-level comparisons of vision-targeted HRQL in clinical research. In addition, the type or severity of the underlying eye disease did not affect the psychometric properties of the NEI-VFQ, suggesting that the measure provides reproducible and valid data when used across multiple eye conditions. Subsequently, researchers have incorporated the NEI-VFQ into clinical studies and have reported HRQL findings for patients with conditions such as AMD, diabetic retinopathy, optic neuritis, glaucoma, uveitis, and general low vision.^{6 - 7 ,16 - 19} Following the development and preliminary testing of the NEI-VFQ, AREDS added the NEI-VFQ-25 plus a 14-item appendix to its follow-up study protocol to provide additional information regarding vision-targeted HRQL in a large cohort of participants with AMD, lens opacity, or both, and to examine the relationship between its subscale scores and clinical measures of visual function. The NEI-VFQ-25 is a short-form version of the 51-item NEI-VFQ. The appendix consists of additional questions that demonstrate adequate reliability and validity for inclusion. The addition of these items enhances the reliability of various subscales of the NEI-VFQ. In addition, the AREDS cohort was used to further assess the internal consistency and reliability of the NEI-VFQ subscales.

Details of AREDS design have been published²⁰ and are briefly described herein. Between November 13, 1992, and January 15, 1998, 4757 persons aged 55 to 80 years at the time of enrollment were entered into the study at 11 clinical centers. The ocular eligibility requirements were largely determined by the AMD component findings of AREDS. Persons were enrolled in 1 of 4 AMD categories, with macular status ranging from no macular pathology in either eye (Category 1), to advanced AMD or lesions of AMD with visual acuity worse than 20/32 in 1 eye only (Category 4). Macular status was assessed, using the AREDS²¹ system for classifying age-related macular degeneration, by trained and certified readers at a reading center, based on stereoscopic color fundus photographs taken at baseline and at regular intervals during follow-up. Except for the requirement that all participants have at least 1 eye with a visual acuity of 20/32 or better and that the media be sufficiently clear for reasonable-quality fundus photography, lens opacity status was not considered in selecting participants. The large sample-size requirements for the AMD component of the study and the expected high prevalence of lens opacities in the targeted age group made it likely that a diverse array of age-related lens opacities would be present among the cohort. Additional exclusions were persons with more than minimal diabetic retinopathy, previous ocular surgery (except for cataract surgery and unilateral photocoagulation for AMD), or presence of any other eye disease that could complicate assessing the progression of lens opacities or AMD (eg, optic atrophy or acute uveitis) or that could affect visual acuity. Finally, persons with illnesses and disorders that made long-term follow-up unlikely were ineligible to participate.

The AREDS²² system for classifying cataracts was used to assess the presence and severity of nuclear, cortical, and posterior subcapsular (PSC) lens opacities at regular intervals during AREDS. The lens photographs were evaluated at a reading center by trained and certified examiners who were masked to the participants' vision status. The extent of cortical and PSC opacities was graded by estimating the area of lens involvement in sectors of a grid overlay on the retroillumination photographs. Nuclear opacities were graded on a decimal scale, using cutpoints set by a series of standard photographs with increasingly severe nuclear sclerosis. Nuclear grades ranged from 0.9 (severity less than the lower standard) to 6.1 (severity exceeding the highest standard). Posterior subcapsular opacities are not considered further in this article because they were uncommon in AREDS (<6% of the total sample were categorized as having moderate PSC at the 5-year examination) and because moderate PSC opacity, when present, often occurred with moderate nuclear opacity (33.8% of cases) or moderate cortical opacities (30.9% of cases).

Visual acuity was measured in each eye by the Early Treatment of Diabetic Retinopathy Study method as the number of letters read correctly, with scores ranging from 0 (<20/800) to 100 (20/10). Age-related macular degeneration category, lens opacity status, and visual acuity following a protocol refraction were assessed at the time of the NEI-VFQ administration.

Participants were classified into 1 of 4 AMD categories:

Participants were classified into 1 of 3 nuclear opacity groups, irrespective of cortical opacity:

Participants were classified into 1 of 3 cortical opacity groups, irrespective of nuclear opacity:

Participants were classified into 1 of 3 cataract surgery groups:

Participants were classified into 1 of 3 visual acuity groups:

Between December 16, 1997, and April 4, 2001, the NEI-VFQ-25 plus appendix was administered at the AREDS 5-year annual clinic visit by trained interviewers to 4121 AREDS participants (86.6% of the entire cohort). Forty-four (1.1%) of the questionnaires were incomplete and were not included among the data in this study. The baseline characteristics of the 44 excluded participants were similar to those of the 4077 included participants. The NEI-VFQ used in AREDS contained 39 items and 12 domains or subscales. Subscale scores were calculated by averaging item scores within the scales. The subscale scores of the NEI-VFQ, ranging from 0 to 100, were computed for overall health, overall vision, difficulty with near-vision activities (such as reading a newspaper), pain or discomfort in or around the eyes, difficulty with distance activities(such as reading a street sign), driving difficulties, limitations with peripheral and color vision, and limitations in social functioning, role limitations, dependency, and mental health symptoms related to vision.

A mean (SEM) for the overall NEI-VFQ score and for each subscale score was computed. Level-specific means (SEs) were calculated according to sex, ethnicity, current AMD category, lens opacity status (nuclear or cortical), and cataract surgery status. We compared the mean differences in scores within the ordered categorical variables (AMD category, nuclear or cortical status, and visual acuity) using a test for trend. Dunnett's multiple comparisons test was used to identify significant differences in scores within the AMD, opacity, visual acuity, and surgery categories by comparing each successive level with the reference (no disease) level. Spearman's rank correlation coefficient ρ was used to assess the association between the NEI-VFQ scores and visual acuity. Internal consistency and reliability were assessed using Cronbach's α for each multi-item subscale. Cronbach's α is a coefficient that describes how well a group of items focuses on a single idea or construct. All analyses were carried out using commercially available statistical software (SAS version 6.12; SAS Institute, Cary, NC).

Characteristics of the 4077 participants who completed the NEI-VFQ are given in Table 1. The mean age of the population was 74 years (range, 60-86 years) at the time of the questionnaire administration. Just over 57% of the participants were female and 96.7% were white. Using letter scores, the mean visual acuity of the better eye of participants was 81.5 letters (Snellen equivalent about 20/25). The mean visual acuity of the worse eye was 69.0 letters (Snellen equivalent about 20/40). About 42% of the participants had visual acuity worse than 20/20 OU. At the time of the questionnaire administration, 22.9% of the participants were classified as having AMD disease Category 1, 23.9% had Category 2, 28.3% had Category 3, and 24.9% had Category 4. A little more than 15% of the participants had no nuclear opacity, 67.2% had moderate opacity, and 17.4% had advanced nuclear opacity. Almost 45% had no cortical opacity, 35.4% had moderate opacity, and 19.6% had advanced cortical opacity. Almost 86% of the participants had not undergone cataract surgery, 6.2% had unilateral aphakia or pseudophakia, and 8.1% had bilateral aphakia or pseudophakia.

Mean (SEM) NEI-VFQ scores, proportions of participants with a perfect score (100 points—ceiling) and with the lowest possible score (0 points—floor), and internal consistency estimates within each subscale are given in Table 2. The internal consistency estimates for the NEI-VFQ subscales, as indicated by Cronbach's α, ranged from 0.58 to 0.91. All multi-item subscales demonstrated a moderately strong internal consistency and reliability. The mean reliability over the 10 multi-item subscales was 0.82.

Rank correlations between the NEI-VFQ subscale scores and visual acuity are given in Table 3. Correlations, which were all positive, ranged from small (0.24) to moderate (0.62), except for ocular pain, which showed a negligible correlation (0.07). The correlations were moderate for subscales that reflect degree of difficulty with common visual activities. The correlation was more modest for the general health subscale compared with the general vision subscale. For each subscale, correlations between subscale scores and visual acuity were similar in magnitude in the better and worse eyes. Correlation with visual acuity was lowest for the ocular pain subscale (0.08).

The NEI-VFQ subscale mean scores by subjects' age, sex, and ethnicity are given in Table 4. Women had lower mean scores for the ocular pain (89 vs 91) and driving (75 vs 80) subscales than men. Men had a lower mean score for the color vision subscale (92 vs 95). The mean overall NEI-VFQ score did not differ significantly by sex. White participants had a lower overall mean score compared with nonwhite participants(86 vs 89). The overall mean and all subscale means (except for ocular pain) were lower for participants older than 75 than for participants younger than 70.

As given in Table 5, the NEI-VFQ subscale mean scores adjusted for age, ethnicity, and sex were progressively lower for successive AMD status categories. Age-related macular degeneration Category 3 participants and, more notably, Category 4 participants had significantly lower NEI-VFQ subscale scores than Category 1 participants. The overall and subscale scores decreased with increased macular pathology (P<.001 for trend). The results of tests for trend for nuclear opacity level are provided in Table 6. The subscale scores showed more dysfunction with increased nuclear opacity, with most of the probability values for trend at P<.001. The differences between the advanced opacity group and the group without nuclear opacity were primarily responsible for this association, as demonstrated by the significant results of the Dunnett multiple comparisons test, which compares each specific level of the characteristic under study with that of the group without nuclear opacity. Participants with advanced nuclear opacity reported the most difficulty with general vision, distance activities, mental health, and driving. This pattern was not evident for the NEI-VFQ subscale scores and cortical opacity status (Table 7). The general vision subscale score was lower with increasing opacity (P<.001 for trend), and the overall and subscale mean scores did not differ significantly with increasing cortical opacity. Participants with unilateral aphakia or pseudophakia had significantly lower NEI-VFQ scores for all subscales except general health, ocular pain, social functioning, and color and peripheral vision, as given in Table 8. Subscale mean scores of participants with bilateral aphakia or pseudophakia did not differ significantly from those of the reference group(participants with no cataract surgery). Participants with vision worse than 20/20 in at least 1 eye had moderately lower NEI-VFQ scores for all the subscales compared with participants with vision 20/20 or better in both eyes, as given in Table 9, while the scores for participants with vision worse than 20/20 in both eyes were markedly lower. Furthermore, the test for trend indicated a progressive trend of dysfunction as individuals moved from unilateral to bilateral forms of vision impairment.

The AREDS cohort comprises persons with a broad age distribution and a large number of individuals with a range of manifestation of AMD and lens opacity. Vision-targeted HRQL scores assessed with the NEI-VFQ-25 plus appendix were lower for participants with progressive manifestations of AMD, nuclear opacity, cataract surgery in only one eye, or progressive degrees of visual impairment. This demonstrates that the NEI-VFQ is sensitive to the effect of age-related nuclear cataract and macular degeneration and supports the construct validity of the questionnaire. The internal consistency was moderate to high for all the subscales. Further evidence of the validity of the questionnaire is provided by the significant correlation between most of the NEI-VFQ subscale scores and visual acuity. Although larger subscale differences may be considered clinically important, some smaller but statistically significant differences that contribute to construct validity may be of uncertain clinical significance.

Cole et al, ⁷ using the larger 51-item NEI-VFQ, assessed the relationship between subscale scores and clinical measures(including visual acuity, visual field, contrast sensitivity, and color vision) and the internal consistency and reliability (Cronbach's α) of the NEI-VFQ subscales for a cohort of 244 patients 5 to 8 years after treatment for an episode of acute optic neuritis. Most of the NEI-VFQ subscales showed more dysfunction when multiple sclerosis was present and when visual acuity was abnormal. Internal consistency and reliability ranged from 0.46 (expectation subscale) to 0.92 (dependency subscale), with a mean reliability of 0.86. Although they used the 51-item field test version of the NEI-VFQ, similar results were observed by computing scores for the abbreviated (25 item) version of the NEI-VFQ, supporting the construct validity of the NEI-VFQ for ocular disease.

Mangione et al¹⁵ reported on the reliability and validity of the 51-item NEI-VFQ for 598 patients with various chronic eye diseases (age-related cataracts, AMD, diabetic retinopathy, primary open-angle glaucoma, and cytomegalovirus retinitis). Internal consistency and reliability estimates ranged from 0.66 (expectation subscale) to 0.94 (near-vision subscale), with most greater than 0.70. They found that overall and subscale mean scores were lower for the condition-specific groups compared with a reference sample of participants with no evidence of underlying eye disease. In addition, they found that subscale scores that are likely to be affected by deficits in central acuity (near vision, distance, and driving) were lowest for those in the low-vision group (best-corrected visual acuity in the better eye 20/70 or worse, or a central visual field ≤20) and for those with AMD. These authors' internal consistency and reliability, along with lower mean NEI-VFQ subscale scores among subjects with AMD or cataract, are similar to those of AREDS using the abbreviated version of the NEI-VFQ-25 plus appendix.

In summary, the NEI-VFQ-25 plus appendix is a reliable and valid questionnaire for patients with various chronic eye conditions. The data presented extend these findings to a large, broad age-range cohort, with excellent characterization of lens and fundus conditions and a range of optical abnormalities. These findings support the use of the NEI-VFQ as a valid measure of HRQL among patients with AMD, cataract, or reduced visual acuity.

Corresponding author and reprints: AREDS Coordinating Center, The EMMES Corporation, 401 N Washington St, Suite 700, Rockville, MD 20850-1707(e-mail: aredspub@emmes.com).

Submitted for publication March 19, 2002; final revision received October 9, 2002; accepted October 15, 2002.

This study was supported by contracts from the National Eye Institute, National Institutes of Health, Bethesda, Md, with additional support from Bausch & Lomb Inc, Rochester, NY.