Abstract
purpose. To describe 9-year changes in refractive errors and estimate incidence of myopia and hyperopia in adults of African-descent, along with associated risk factors.
methods. The 9-year follow-up of the Barbados Eye Studies (1997–2003) reexamined 2793 surviving cohort members (81% participation). Refractive errors were determined by automated refraction. Myopia/hyperopia were defined as spherical equivalent < −0.5 diopters (D)/> +0.5 D, and the cutoff for moderate-high myopia/hyperopia was 3.0 D. Incidence rates of myopia/hyperopia were estimated by the product-limit approach, based on eyes without such conditions at baseline. Risk factors were evaluated by logistic regression in discrete time hazard models.
results. Nine-year refraction changes varied by age. Persons aged 40 to 49 years experienced hyperopic shifts (median, +0.38 D), whereas persons ≥60 years had myopic shifts (median, −0.75D). Overall 9-year incidence was 12.0% for myopia and 29.5% for hyperopia; rates were 3.6% and 2.0% for moderate-high myopia and hyperopia, respectively. Myopia risk increased with age, baseline nuclear lens opacities (risk ratio [RR] = 1.7; 95% confidence interval [CI]: 1.01–2.9), glaucoma (RR = 6.0, 95% CI: 3.9–9.3), and ocular hypertension (RR = 2.0, 95% CI: 1.3–3.0), while cortical lens opacities decreased risk (RR = 0.6, 95% CI: 0.4–0.9). Incidence of moderate-high myopia was also related to baseline age, nuclear opacities, glaucoma, male gender (RR = 1.7, 95% CI: 1.0–2.8), and diabetes history (RR = 1.9, 95% CI: 1.01–3.5). Hyperopia risk decreased with older age, male gender, and glaucoma diagnosis.
conclusions. Refractive errors continue to develop frequently in older adults. Nuclear lens opacities, glaucoma, and diabetes increase the risk of older-onset myopia, a result of public health relevance to this and similar African-origin populations.
Refractive errors are the most frequent ocular disorders, affecting approximately one third of adults aged 40 years or older in the United States and Western Europe.
1 Genetic and environmental factors are thought to play a role in the development of refractive errors, but the true underlying mechanisms involved remain unclear. Although refractive errors can be corrected, the economic burden incurred can be substantial, given their high prevalence. Furthermore, uncorrected or inadequately corrected refractive errors have been shown to be a major cause of visual impairment in population-based studies,
2 3 4 5 6 7 8 signifying high public health relevance. Several studies have provided data on prevalence and temporal changes in refractive errors in various adult populations.
9 10 11 12 13 14 15 16 17 18 19 20 21 22 We reported a high prevalence of myopia (22%) and hyperopia (47%) among the predominantly black adult participants in the Barbados Eye Studies (BESs), aged 40 to 84 years. These refractive errors were related to concomitant ocular conditions. To our knowledge, however, no longitudinal data on refraction are available in persons of African descent. The objectives of this study were to (1) describe changes in refractive errors over 9 years; (2) provide incidence estimates; and, (3) determine associated risk factors for incident myopia and hyperopia in African-origin participants in the Barbados Eye Studies (BESs).
The Barbados Eye Studies (BESs; 1987–2003), funded by the National Eye Institute, are epidemiologic investigations of the prevalence, incidence, and risk factors for major eye diseases in a population predominantly of African descent, with the same ancestral origin as African-Americans. The baseline prevalence study, the Barbados Eye Study (BES, 1987–1992),
23 was based on a simple random sample of Barbadian-born citizens, 40 to 84 years of age, with 84% participation. The demographic composition of BES participants closely resembled the census population. Of the 4631 persons examined at the study site, 4314 (93%) were black, 184 (4%) were mixed (black and white), and 133 (3%) were white/other by self-report. The surviving cohort was invited for reexaminations to determine the incidence and progression of these eye diseases after 4 years (
n = 3427, 85% of those eligible)
24 and 9 years (
n = 2793, 81% of those eligible).
25 As described previously, the latter participants were younger and less likely to have hypertension than were nonparticipants.
25
The study protocols at baseline
23 and both follow-up visits
24 25 were standardized and included various ophthalmic and other measurements, such as the determination of best corrected visual acuity with a Ferris-Bailey chart, automated perimetry (Humphrey; Carl Zeiss Meditec, Dublin, CA), applanation tonometry, lens grading, and color stereo photography of the disc and macula. A comprehensive interview on demographics, medical and family history, and environmental and other factors was completed during dilation. Noncycloplegic refractive error was measured with the same refractor (Humphrey Automated Refractor model 530; Carl Zeiss Meditec) throughout the three study visits. If the refractor could not be used or refraction was unreliable, the participant’s present correction, if any, was determined with a lensometer (only in 1% to 2% of the study population). In this report, refractive errors were based on Humphrey autorefraction measures. The study adhered to the tenets of the Declaration of Helsinki and was approved by the institutional review board (IRB) of participating institutions. Informed consent was obtained from all study participants after explanation of the nature and possible consequences of the study.
Refractive errors in an eye were represented by the spherical equivalent (sum of spherical power and half cylindrical power) from the Humphrey refraction. Changes in refraction in an eye at 9 years were defined as the refraction at the 9-year visit minus the refraction at baseline. Myopia was defined as a spherical equivalent < −0.5 D and hyperopia as a spherical equivalent >0.5 D. Moderate-high myopia and hyperopia were defined by cutoff of < −3 D and >3 D, respectively. Incidence of myopia or hyperopia was defined as development of myopia or hyperopia at the follow-up visit, whereas no such condition was present at baseline. The incidence of moderate-high myopia was defined as the development of moderate-high myopia by the 9-year follow-up, among eyes without that condition at baseline. The incidence of moderate-high hyperopia was similarly defined.
Lenses were graded using the Lens Opacities Classification System (LOCS) II at the slit lamp. A LOCS score of ≥2 defined nuclear (NO), PSC (P), and cortical (C) opacities.
26 Open-angle glaucoma (OAG) was defined by specific study criteria based on the presence of optic disc and visual field defects not attributable to other causes.
23 Ocular hypertension included persons with an intraocular pressure (IOP) >21 mm Hg or with a history of IOP-lowering treatment, but without glaucoma visual field defects or disc damage. For the purposes of this report, the term glaucoma included OAG and suspected OAG (with some but not all the optic disc and visual field criteria for OAG).
The distribution of 9-year changes in refraction was determined among persons with refraction data at both baseline and 9-year follow-up visits. The cumulative 9-year incidence of myopia/hyperopia was estimated by the product-limit approach,
27 which allows the use of data from persons with 4 years of follow-up only. Factors associated with incident myopia and hyperopia were evaluated by logistic regression with discrete time hazard models.
28 Potential risk factors included age, gender, education, occupation, diabetes history (predominantly type 2 diabetes), and major ocular conditions, including lens opacities and glaucoma at baseline. The correlation for refractive changes between right and left eyes was high (
r = 0.75), and the prevalence rates reported at baseline were based on data from right eyes.
15 Therefore, results presented in this report (refraction change, incidence estimates, and associated factors) were based on data from right eyes. Risk factor evaluations also explored the Generalized Estimating Equation (GEE) approach,
29 which includes data from both eyes while accounting for correlations between eyes.
At the 9-year visit, 2612 of the 2793 participants were of African descent. Of these, 2391 (91.5%) had complete refraction data both at baseline and the 9-year follow-up. Analyses were based on 2230 persons (2128 had data for right eyes and 2123 had data for left eyes) after excluding 103 persons with a history of cataract surgery and 58 with visual acuity 20/200 or worse, due to the reduced reliability of their refractions. Comparisons of baseline characteristics indicated that persons excluded were older (mean ± SD: 68.8 ± 9.5 years) than those included (mean ± SD: 53.7 ± 9.4 years) and, after adjusting for age, more likely to have diabetes, lower education, refractive errors, and other age-related ocular conditions such as lens opacities and glaucoma (P < 0.05). No significant differences were found with regard to gender, blood pressure, a history of smoking, drinking, and cardiovascular disease.
Table 1presents age- and gender-specific 9-year changes in refractive errors. The mean overall 9-year change was very small, but in the direction of a slight myopic shift (i.e., negative change: mean, −0.08 D; 95% CI: −0.13 to −0.03). The median overall change, however, had a positive sign (median, +0.13 D). On average, a hyperopic shift (i.e., positive change, was observed in persons 40 to 49 years of age at baseline: mean = +0.47 D in 9 years), whereas a myopic shift was noted in persons 50 years of age or more. The magnitude of the myopic shift was broadened in those 60 to 69 years of age, but leveled off in those aged 70 years or more. Women tended to have smaller myopic shifts than men overall, having more hyperopic shifts in the youngest age group (+0.57 D vs. +0.34 D) and less myopic shift in the oldest age group (−0.65 D vs. −0.96 D). As indicated in
Table 1 , only 6% of men and 2% of women 40 to 49 years of age had shifts toward myopia larger than −0.5 D, compared with 67% of men and 46% of women in the oldest age group. Overall, 23.5% of the study population had a shift toward myopia more than −0.5 D and 26% had a shift toward hyperopia more than 0.5 D.
Table 2shows stratification by cataract status, which revealed a general myopic shift across all age groups in persons with nuclear opacities at baseline (overall mean change ± SD: −0.85 ± 1.54 D). Among those without nuclear opacities, patterns were similar to those in the overall study population, with myopic shifts mainly seen in age groups of 60 years or more, whereas shifts toward hyperopia (mean change ± SD: 0.47 ± 0.75 D) occurred in those 40 to 49 years of age. Changes in refractive errors were also evaluated by baseline refractive status. Results showed no significant differences among emmetropic (
n = 759), myopic (
n = 322), and hyperopic eyes (
n = 1047) at baseline, with age-adjusted means (95% CI) being −0.11 D (95% CI: −0.19 to −0.03 D), −0.12 D (95% CI: −0.24 to −0.00 D), and −0.04 D (95% CI: −0.11 to 0.02 D), respectively.
Table 3presents the 9-year cumulative incidence of myopia and hyperopia among eyes without refractive errors at baseline. The overall 9-year incidence was 12% (95% CI: 10.6%–13.4%) for myopia and 29.5% (95% CI: 26.9%–32.1%) for hyperopia. Although the frequency of myopia was similar in the men (12.9%) and the women (11.5%), incidence of hyperopia appeared to be higher in the women (33.3%) than in the men (25.1%;
P < 0.05). The 9-year incidence of moderate-high myopia was low, being 3.6% (95% CI: 2.8%–4.4%); rates increased with age from 0.8% at 40 to 49 years to 12.5% at 70 years or older. The incidence of moderate-high hyperopia was very low, 2.0% (95% CI: 1.4%–2.5%), with a slightly higher (
P < 0.05) incidence in women (2.7%) than in men (0.9%).
We also examined possible cohort effects by graphic exploration of the refraction patterns at each study visit against age and year of birth.
Figure 1presents prevalence of myopia by birth cohort and age group, where each person may contribute up to three data points from all study visits, including only subgroups with
n ≥ 100. Results indicated no clear cohort effect. There was no consistent pattern when comparing the older and the younger birth cohorts for each age group. Although myopia prevalence was slightly higher in persons 50 to 59 years who were born later than younger cohorts (1940–1944 vs. 1935–1939), the increase was very small and not statistically significant.
As seen in
Table 4 , the risk of development of myopia at 50 to 59 years was 3.5 times higher than at 40 to 49 years. The risk increased to 7.5 to 9 times higher for those 60 years or older, compared with the youngest age group. No significant association with incident myopia was found for gender, education, occupation, or diabetes. The presence of nuclear opacities at baseline increased the risk of myopia (RR = 1.7, 95% CI: 1.01–2.89). Although eyes with any cortical opacities at baseline had a higher crude incidence (16.1%) than those without (10.8%), cortical opacities decreased myopia risk (RR = 0.6, 95% CI: 0.4–0.9) after adjustment for nuclear and PSC opacities and other factors. The low number of PSC cases precluded any meaningful assessment. The baseline presence of glaucoma (RR = 6.0, 95% CI: 3.9–9.3) and ocular hypertension (RR = 2.0, 95% CI: 1.3–3.0) was also associated with 9-year incidence of myopia. Additional analyses using LOCS II grades as categorical variables were conducted, although more severe opacities (LOCS II score of >2) were infrequent: only 0.4% for nuclear, 0.04% for PSC, and 9.2% for cortical opacities. Results showed that myopia risk increased (
P < 0.05) with increasing severity of nuclear opacities. Incidence of myopia was found to decrease with advancing cortical opacities. Increasing age, on the other hand, decreased the risk of hyperopia
(Table 4) . Men were less likely to have hyperopia than women (RR = 0.7, 95% CI: 0.6–0.95). No statistically significant associations were found between baseline lens opacities of any type and incident hyperopia, although nuclear opacities had an adjusted RR of 0.6. Glaucoma was negatively associated with 9-year incidence of hyperopia (RR = 0.4, 95% CI: 0.2–0.8). A similar trend toward a negative association was noted for ocular hypertension, but it was of borderline statistical significance (RR = 0.6, 95% CI: 0.4–1.03,
P = 0.07). No significant interactions were found between age and cataract, whether assessed as dichotomized or categorical variables. Additional analyses using the GEE approach yielded similar results.
Associations with incident moderate-high myopia are shown in
Table 5 . Results show an increasing risk at older ages, with ninefold relative risks for persons 70 years of age or older (RR = 9.1; 95% CI: 3.3–25.0), compared with those 40 to 49 years at baseline; in addition, the men were significantly more prone to severe myopia (RR = 1.7; 95% CI: 1.0–2.8) than were the women. Persons with diabetes history had a nearly twofold increased risk of myopia < −3 D (RR = 1.9; 95% CI: 1.01–3.5). Nuclear opacities conferred a substantially increased risk (RR = 3.6; 95% CI: 2.0–6.7), whereas cortical opacities decreased the risk. Glaucoma, but not ocular hypertension, remained significantly associated with incident moderate-high myopia. The very low incidence of moderate-high hyperopia led to limited power to detect associations. Among all factors evaluated, only female gender (RR = 3.8; 95% CI: 1.6– 8.7) was significantly associated with incident moderate-high hyperopia.
The natural history of refractive errors among older adults has not been widely studied until the past decade. Most research has focused on myopia in children, as this condition is mainly considered to occur at young ages. More recent population studies, involving predominantly European-origin participants, have shown that refractive shifts continue in older adulthood.
19 21 22 The BESs provide the first longitudinal data, to our best knowledge, on refractive error changes in a large adult African-origin population. Over the 9-year follow-up period, refraction changes varied according to age. Persons 40 to 49 years of age at baseline experienced a hyperopic shift (mean, +0.47 D; median, +0.38 D) with only 4% having changes less than −0.5 D, whereas persons 60 years of age or more had refractive error shifts toward myopia (mean = −0.88 D; median = −0.75 D), with more than 50% of participants having myopic shifts.
The pattern of shift toward hyperopia among younger adults and the shift toward myopia among older adults is consistent with longitudinal observations in other population studies in Australia and the United States.
19 21 22 The 9-year changes in refractive errors in our study were also consistent with the baseline prevalence data from the BESs, which showed an increasing prevalence of myopia and a decreasing prevalence of hyperopia after 60 years of age.
15 Similar results were found in some
17 18 but not other studies.
10 11 13 16 30 The age patterns observed in our investigation indicate that actual longitudinal changes with age are a more likely explanation than a cohort effect, which was not apparent in our study. The shifts toward hyperopia in younger adults, as they increased in age during follow-up, seemed consistent with age-related biometric data from Chinese adults in Singapore.
31 In that cross-sectional study, hyperopia was more prevalent in the older than in younger members of the 40 to 59 year age group, an increase that was principally explained by shorter axial lengths and lesser vitreous chamber depths with age.
31 An additional contributor to the hyperopic shifts in younger adults from our study may be that nuclear cataract has a later onset (after 50 years). The myopic shifts in older adults thus may be largely influenced by the presence of nuclear cataract at baseline or at follow-up. In fact, baseline nuclear opacities and advancing age both independently increased the risk of myopia
(Table 4) . The 9-year incidence of myopia continued to increase with age, even in persons without nuclear opacities at baseline, from 3.3% in persons aged 40 to 49 years to 10.6%, 24.4%, and 29.2% in the subsequent age groups. Most of the older participants without nuclear opacities at baseline had nuclear cataract develop by follow-up, with frequencies being 8% and 40% in the two younger age groups, 76% in persons aged 60 to 69 years, and 90% in persons aged 70 years or older.
The Blue Mountains Eye Study (BMES) reported a 5-year hyperopic change in persons aged 49 to 54 years (+0.42 D) and a myopic shift (−0.22 D) in those aged 75 years or more. Myopic shifts of greater than 0.5 D were seen in only 3% of the younger age group, but they continuously increased to 28% in the older group.
21 In the Beaver Dam Eye Study (BDES), only 4% of participants 59 years of age or less had a shift toward myopia more than −0.5 D during a 10-year period, whereas 33% of those 70 years or more had such myopic shifts.
22 Myopic shifts at older ages appeared to be more substantial in BESs than in these two studies, and the reversal in age pattern occurred at younger ages. In BESs, a slight myopic shift was noticeable in persons as young as 50 to 59 years of age, and a marked shift toward myopia started at 60 to 69 years. In contrast, considerable myopic shifts in the BMES and BDES were not observed until 75 years of age at the 5-year follow-up and until 70 years of age at the 10-year follow-up, respectively. This finding may be partially attributable to the lower rate of cataract surgery in our study.
25
The women in the BESs population were more prone to development of hyperopia than were the men (RR = 1.4). Our results appear to support findings from various cross-sectional evaluations that reported a positive relationship between female gender and hyperopia.
10 12 13 15 17 18 The men had a 70% higher risk of moderate-high myopia (RR = 1.7). The somewhat larger 9-year myopic shift in the men (−0.11 D) than in the women (−0.06 D) was also statistically significant (
P = 0.03), after adjustment for age, lens opacities, and glaucoma status. Longitudinal investigations of refractive error change in BMES
21 and BDES,
19 22 however, showed no gender differences. The BESs protocol did not include measurements of axial length or keratometry. However, variations in ocular dimensions have been found in other adult populations and may partially explain the age and gender differences in refraction.
31 32
Education and near-work related occupations were not related to the development of myopia or hyperopia during 9 years of follow-up, consistent with findings from BDES, where education showed no effect on the longitudinal refraction changes.
22 Additional analyses also showed no significant interactions between education and occupation, as well as between age and each of these factors. In the BESs, near-work occupational categories were based on self-reported lifetime occupation (“what kind of work have you done for
most of your life”). Such a definition may not accurately reflect the extent and intensity of close-up activities of the participants. Genetic and environmental factors, and their interactions, may contribute to refractive errors, as has been mostly discussed in children.
33 Therefore, our results may be susceptible to uncontrolled confounding from factors such as near-work activity early in life and family history of refractive errors, which were not included in our data collection.
Although no association was noted between diabetes and incident myopia/hyperopia based on the cutoff of 0.5 D, persons with a history of diabetes had a nearly twofold risk of development of moderate-high myopia (< −3 D). In an adult clinic population in Denmark, a preponderance of myopia was reported in persons with diabetes (compared with those without).
34 35 In contrast, the presence of diabetes in BDES was related to a larger shift toward hyperopia
22 ; furthermore, diabetes did not predict 5-year refractive change in BMES.
21 Transient changes of refraction in diabetes could be myopic or hyperopic, depending on whether changes occur in the refractive indices or in the curvatures of the lens.
36 The relationship between diabetes and myopia thus merits further confirmation from other studies.
Eyes with nuclear opacities at baseline had a 1.7-fold risk of myopia and a 3.6-fold risk of moderate-high myopia at follow-up. The association of refractive errors to cataract, particularly nuclear cataract, has been reported in several cross-sectional population studies of adults.
15 16 17 18 32 37 38 In one study in which the relation between myopia and cataract morphology was investigated, investigators concluded that simple myopia does not predispose to cataract, but is the product of cataract—in particular, of nuclear sclerosis.
39 When examining the cross-sectional relationship of age-related cataract to refractive errors and biometric measurements, Wong et al.
38 found that nuclear cataract was associated with myopia but not with axial ocular dimensions. These observations are consistent with the index-myopia hypothesis that increasing nuclear sclerosis of the lens with age causes a myopic shift in refraction. In addition to the association between baseline nuclear opacities and incident myopia found in our study, data from several other longitudinal studies appear to support such a causal relationship. In the BDESs, no association was found between baseline myopic refraction and 5-year incident nuclear cataract,
40 but a significant relationship was reported between baseline nuclear sclerosis and a shift toward myopia after 10 years of follow-up.
22 In the BMES, the presence of nuclear cataract at baseline was associated with a 5-year myopic shift.
21 One small study also showed a significant myopic shift among persons with nuclear cataract.
41 However, myopia could also be related to the development of nuclear opacities, as shown from the significant association of baseline myopia to incident nuclear opacities, as found in separate evaluations of BMES
42 and BESs
43 data. The complicated interrelationship between nuclear cataract and myopia therefore needs further investigation.
The association of cortical cataract and refractive errors has not been established clearly. In the BESs, the presence of cortical opacities at baseline was negatively associated with incident myopia (RR = 0.6) after a 9-year follow-up. This relationship reflects a role for the
cortical only opacity type, since it was found after adjusting for other coexisting types of opacities, especially nuclear, which sometimes accompany cortical opacities. Although results were consistent with baseline findings from BES, there is no obvious explanation for such a relationship. The polarity of spherical aberration was shown to be negative in eyes with nuclear cataract, but was positive in eyes with cortical cataract
44 ; such disparity may have different effects on refractive errors.
The cross-sectional association of glaucoma and myopia has been demonstrated in various clinic-based and case–control studies.
45 46 47 Associations between IOP and refractive errors were also found in some
47 48 49 but not all reports.
50 51 52 In the older white population of the BMES, a strong relationship between myopia and glaucoma (two- to threefold) was found, as well as a borderline association between myopia and ocular hypertension.
53 In the large population survey conducted to identify patients for the Early Manifest Glaucoma Trial, myopia was reported as a major risk factor for glaucoma—particularly for normal-tension glaucoma.
54 Investigation of our baseline data also shows a positive association of myopia to glaucoma and ocular hypertension, with a negative association with hyperopia.
15 Findings from the longitudinal data in the BESs cohort were thus consistent with results from our baseline cross-sectional evaluation, as they confirmed the relationship between myopia and glaucoma/ocular hypertension. Although glaucoma, ocular hypertension, and incident myopia appear to be linked, the temporal association of these conditions is not intuitively clear. The general impression is that myopia may predispose to glaucoma, rather than being its consequence; however, the association between baseline glaucoma/ocular hypertension and subsequent development of myopia in this adult population suggests otherwise. Because myopia and glaucoma both show changes in ocular connective tissue, they may share common pathways.
55 The linkage between myopia and glaucoma may not necessarily have a single causal explanation. Increased IOP can cause axial elongation of the globe and myopia in a previously emmetropic eye—a process more likely to occur in congenital or juvenile glaucoma.
55 Studies also have shown that reduction of IOP may prevent progressive myopia.
56 In sum, the physiological and clinical interpretation of these relationships remains in doubt.
In addition to the population-based design, major strengths of our study include the long-term follow-up of the cohort, with credible participation, as well as the standardized protocol for refractive errors and other outcomes at all study visits. However, losses to follow-up are inevitable in such a long-term study, and, as expected, nonparticipants were older.
25 The current results exclude persons with cataract surgery or very poor visual acuity, an approach similarly taken by other population-based studies on refractive errors. These exclusions could lead to selection biases, as those excluded may also be older and have a higher likelihood of ocular conditions such as cataract and glaucoma at baseline. Results may thus underestimate the incidence of myopia and overestimate the incidence of hyperopia, with possible attenuation of risk estimates. The infrequent presence of PSC opacities in this population did not allow an appropriate evaluation of this potential risk factor.
Refractive changes beyond early adulthood were observed in this older adult population of African origin. During the 9 years of follow-up, myopic shifts were observed at younger ages than in European-derived populations. Lens opacities, glaucoma, and diabetes, which are frequent in persons of African descent, increased the risk of subsequent myopia or moderate-high myopia. These prevalent factors may be involved in the development of older-onset myopia, further contributing to visual problems in this high-risk population.
Principal Investigator: M. Cristina Leske
Stony Brook University, Stony Brook, NY: M. Cristina Leske, Barbara Nemesure, Suh-Yuh Wu, Leslie Hyman, Xiaowei Li, Lixin Jiang, Ling Yang, Kasthuri Sarma, Karen Kelleher, and Melinda Santoro.
Ministry of Health, Bridgetown, Barbados, West Indies: Anthea M. S. Connell, Anselm Hennis, Ann Bannister, Muthu A. Thangaraj, Coreen Barrow, Patricia Basdeo, Kim Bayley, and Anthanette Holder.
The Johns Hopkins University, Baltimore, MD: Andrew P. Schachat, Judith A. Alexander,Cheryl J. Hiner, Noreen B. Javornik, Deborah A. Phillips, Reva W. Strozykowski, and Terry W. George.
Trevor Hassell (Department of Cardiology), Henry Fraser (Chronic Diseases Research Centre), Clive Gibbons (Department of Ophthalmology), School of Clinical Medicine and Research, University of the West Indies, Queen Elizabeth Hospital, Barbados, West Indies.
The authors thank the Barbados Eye Studies participants for taking part in the study and the Ministry of Health, Barbados, West Indies, for their role in the study.