Independent Impact of Area-Level Socioeconomic Measures on Visual Impairment

Abstract

Purpose.: It is known that a person's socioeconomic status (SES; individual-level SES) is closely correlated with his or her degree of visual impairment. Whether there is an independent relationship between area-level measures of SES (e.g., living in a lower SES environment) and visual impairment is unclear. This study describes the associations of area-level SES with visual impairment.

Methods.: The authors conducted two population-based cross-sectional studies of 3280 adult Malays and 3400 adult Indians living in Singapore. Visual impairment was defined as LogMAR visual acuity >0.30 in the better-seeing eye. Area-level SES measures (e.g., proportion of people not speaking English, proportion of people with low income) were derived from the Singapore's 2000 population census.

Results.: Increasing age and individual-level SES measures (including lower education level, lower income level, and lower occupational status) were significantly associated with increased odds of visual impairment. In analyses adjusting for age and individual-level SES measures, many area-level SES measures (e.g., higher proportion of people not using English, higher proportion of people with low income) were also significantly associated with increased odds of visual impairment. These associations were consistently observed in both Malays and Indians.

Conclusions.: These data suggest that not only is a person's SES, but the SES of his or her immediate community, is associated with visual impairment. Further research is needed to investigate the underlying causes of visual health disparities and to improve the eye health of communities with lower SES.

A person's socioeconomic (SES) status (individual-level SES), such as educational and income levels, has long been recognized as an important determinant of health, including eye health and the likelihood of visual impairment and blindness. ¹ Population-based studies in the United States, ^{2 –4} Australia, ⁵ India, ⁶ China, ^7,8 Singapore, ^9,10 Japan, ¹¹ and others ¹ have consistently shown that people with lower education or income are more likely to have poorer vision than their better-off counterparts. Despite decades of medical innovation and eye care services, the socioeconomic disparities in visual health persist and remain a major challenge for health policy making. ^12,13  

However, though the influences of individual-level SES measures on morbidity and mortality are well established, there is increasing recognition that this alone may not reflect the whole spectrum of health inequalities of a person's risk for disease. ^{14 –17} In this regard, it is now thought that the SES of the person's broader living environment, sometimes termed area-level SES measures (e.g., proportion of an area's residents who do not speak English, proportion of an area's residents who have low income), may have an independent impact on health and disease. Previous studies have shown that people living in low-SES areas are more likely to have lower life expectancy ¹⁸ and higher levels of cancer, ¹⁹ coronary disease, ²⁰ asthma, ²¹ chronic kidney disease, ²² and many other health conditions ²³ than those living in high-SES areas. The direct or indirect influences of the neighborhood environment may be attributable to geographic variations in the availability and accessibility of food, cigarettes, alcohol, recreational space, and health services. ^24,25 Area-based or geographic information may therefore have important equity and policy implications. 

To the best of our knowledge, there is limited research focused on the influence of area-level SES measures on visual impairment and on whether it is independent of and additive to individual-level SES. ^2,5 In this study, we determined the independent and joint effects of individual-based and area-based SES measures on visual impairment in Malays and Indians, aged 40 and older, living in Singapore. 

Methods

Study Design and Procedure

The Singapore Malay Eye Study (SiMES) was conducted from 2004 to 2007, and the Singapore Indian Eye Study (SINDI) was conducted from 2007 to 2009. ^{26 –29} Both are population-based and cross-sectional studies. They followed the same study design and sampling areas. The methodology details have been reported elsewhere. ^{26 –29} In brief, the Ministry of Home Affairs (MHA) provided initial computer-generated lists of 16,069 ethnic Malays and 12,000 ethnic Indians residing in 15 postal districts in southwest Singapore. These residential areas (110.4 km²) account for 15.8% of the country's total land area. They were chosen because there were sufficient numbers of ethnic Malays and Indians in these regions and because the residents were fairly representative of the Singapore population in terms of age distribution, housing type, and socioeconomic status, according to the 2000 Singapore Census. We used the criteria set by the Singapore census to define Malay and Indian ethnicity, as indicated by the National Registration Identity Card. From the MHA lists, we derived a final sample frame of 5600 ethnic Malays and of 6359 ethnic Indians based on an age-stratified random sampling strategy. Of the sampling frames, 4168 ethnic Malays and 4497 ethnic Indians were deemed eligible to participate, whereas the ineligible persons included those who had moved from the residential address, had not lived there in the past 6 months, were deceased, or were terminally ill. From 2004 to 2006, the SiMES examined 3280 Malay adults aged 40 and older living in Singapore, with a participant rate of 78.7%. Of the nonparticipants, 831 (93.6%) refused to participate and 57 (6.4%) were not contactable. From 2007 to 2009, the SINDI examined 3400 Indian adults aged 40 and older living in Singapore, giving a participant rate of 75.6%. Of the nonparticipants, 1021 (93.1%) refused to participate and 76 (6.9%) were not contactable. 

For both studies, the nonparticipants on average were slightly older than the participants (P < 0.001 in both studies), but there were no gender differences (P > 0.05 in both studies). The studies adhered to the Declaration of Helsinki, and ethics approvals were obtained from the Singapore Eye Research Institute Institutional Review Board. 

The 2000 Singapore census used the development guide plan (DGP, a detailed urban land use plan for each of the 55 areas in Singapore designated by the Urban Redevelopment Authority) for geographic classification. ³⁰ To facilitate area-level analysis, we followed the DGP criteria and reclassified the study areas into nine DGP regions: Bukit Batok, Clementi, Jurong East, Jurong West, Bukit Merah, Bukit Timah, Outram, Queenstown, and Tanglin (Table 1). To obtain stability and to reduce the sampling variability of disease prevalence, we removed the DGP areas with fewer than 50 participants. As a result, the final spatial analyses for the SiMES cohort included the Bukit Batok, Clementi, Jurong East, Jurong West, Bukit Merah, Queenstown, and Tanglin regions, whereas the final analyses for the SINDI cohort included the Bukit Batok, Clementi, Jurong East, Jurong West, Bukit Merah, and Outram regions. 

Table 1.

 

Characteristics of the Participants in the Singapore Malay Eye Study (n = 3280) and the Singapore Indian Eye Study (n = 3400)

	Singapore Malay Eye Study	Singapore Indian Eye Study
Age, y	58.7 (11.0)	57.8 (10.1)
Sex, female	1705 (52.0)	1706 (50.2)
BMI, kg/cm2	26.16 (4.75)	26.2 (4.8)
Education level
No formal education	682 (21.0)	312 (9.4)
Primary education	1761 (54.3)	1555 (46.9)
Secondary education	594 (18.3)	803 (24.2)
Polytechnic/university	207 (6.4)	648 (19.5)
Income
<S$1000	1830 (56.3)	1092 (33.0)
≥S$1000	1004 (30.9)	1748 (52.8)
Retirement income	416 (12.8)	471 (14.2)
Housing type
1–2-room flat	504 (15.4)	160 (4.7)
3–4-room flat	2253 (68.9)	2021 (58.6)
5-room flat/private housing	515 (15.7)	1212 (35.7)
Marital status
Never married	140 (4.3)	149 (4.4)
Married	2431 (74.4)	2656 (78.2)
Separated/divorced	178 (5.5)	143 (4.2)
Widowed	517 (15.8)	450 (13.2)
Occupation
Service work	696 (21.4)	144 (4.3)
Professional or office work	286 (8.8)	525 (15.8)
Factory work	361 (11.1)	48 (1.4)
Homemaking	1089 (74.9)	811 (24.4)
Retired/unemployed/other	815 (25.1)	1796 (54.0)
Current smoker	662 (20.2)	499 (14.7)
Diabetes mellitus	768 (24.2)	1110 (33.7)
Hypertension	2246 (68.5)	1368 (40.2)
Area of residence
Bukit Batok	536 (16.4)	367 (10.9)
Clementi	411 (12.6)	572 (16.9)
Jurong East	387 (11.8)	399 (11.8)
Jurong West	834 (25.5)	1311 (38.8)
Bukit Merah	546 (16.7)	568 (16.8)
Bukit Timah	0 (0)	48 (1.4)
Outram	20 (0.6)	108 (3.2)
Queenstown	405 (12.4)	10 (0.3)
Tanglin	135 (4.1)	0 (0)

 Data are mean (SD) or n (%), as appropriate.

Visual Acuity Testing and Definition of Visual Impairment

After obtaining informed consent from each participant, we conducted all examinations at the Singapore Eye Research Institute, a clinical research facility located centrally in Singapore. The examination included visual acuity (VA) testing and a detailed clinical slit-lamp examination. ^27,29 VA was measured using a logarithm of the minimum angle of resolution (LogMAR) number chart (Lighthouse International, New York, NY) at a distance of 4 m. If no numbers were read at 4 m, the participant was moved to 3, 2, or 1 m consecutively. If no numbers were identified on the chart at 1 m, VA was assessed as Counting Fingers, Hand Movements, Perception of Light, or No Perception of Light. Refraction was corrected by certified study optometrists, and the best-corrected visual acuity (BCVA) was obtained for each eye. We defined visual impairment as BCVA worse than 20/40 (LogMAR >0.30) in the better-seeing eye. 

Individual-Level Socioeconomic Measures

A detailed interviewer-administered questionnaire was used to collect relevant demographic and socioeconomic information. ^9,10 With the participant's consent, randomly selected interviews were audiotaped for periodic review by the investigators to ensure quality. The collected information included marital status (0, married; 1, never married; 2, separated or divorced; 3, widowed), education (0, polytechnic/university; 1, secondary education; 2, primary education; 3, no formal education), occupation (0, service work; 1, professional/office work; 2, factory work; 3, homemaking; 4, unemployed/other), income (0, earning >1000 Singapore dollar [SGD]/month; 1, retirement income; 2, earning <1000 SGD/month), and current housing status (0, 5-room flat/private house; 1, 3–4-room flat; 2, 1–2-room flat). The definition of the individual SES index was based on the coexistence of three factors (factor 1, elementary/lower education; factor 2, income <1000 SGD/retirement income; factor 3, 1–2-room flat), and it was categorized into three groups (having all three factors, having a combination of any 2 of the 3 factors, and having ≤1 factor. ³¹  

Height was measured with a wall-mounted tape and weight with a digital scale (SECA, model 782 2321009; Vogel & Halke, Hamburg, Germany). Body mass index (BMI) was calculated by dividing weight (in kg) by the square of height (in meters). ³² Blood pressure was determined by a digital blood pressure monitor, and hypertension was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or self-reported previously diagnosed hypertension. Diabetes mellitus was defined as nonfasting blood glucose level ≥11.1 mmol/L, self-reported physician-diagnosed diabetes, or use of diabetic medications. ³³  

Area-Level Socioeconomic Measures

We obtained 2000 census data of the DGP regions from the Department of Statistics Singapore. ³⁰ For each DGP region, the proportions of residents' SES characteristics (e.g., proportion of people speaking English, proportion of people having incomes of ≤1000 SGD/month) were used as area-level SES measures (listed in Supplementary Tables S1 and S2). ³⁰ The SiMES and SINDI participants' addresses were geo-coded into several DGP regions so that every participant was given a set of DGP-specific SES measures. We also summarized the area-level SES measures into factor scores (Factor 1 and Factor 2) by using factor analysis, in which a maximum likelihood approach was used to determine which area-level variable should be retained in the model. This was followed by orthogonal (varimax) rotation. After rotation, loadings with absolute values of ≥0.6 were used to interpret the factors (listed in Supplementary Table S3). 

Statistical Analysis

Logistic regression was performed to calculate odds ratios and 95% confidence intervals for the association between potential risk factors (e.g., age, sex, marital status, presence of diabetes, blood pressure, BMI, and SES factors) and visual impairment. Variables with P < 0.05 at univariate regression were retained in multiple logistic regression models. Statistical analyses were performed using the software R (http://www.r-project.org). 

Results

Table 1 describes the baseline characteristics of the SiMES and SINDI participants. Comparison of the two study cohorts showed that the SiMES participants were more likely to have lower incomes, lower education levels, and hypertension, whereas the SINDI participants tended to be retired or unemployed and to have diabetes. 

Table 2 shows the associations between individual-level SES measures and bilateral visual impairment in SiMES and SINDI. In the multivariate regression analyses, increasing age, lower education level, lower income, being a homemaker, and being unemployed were significantly associated with visual impairment in the SiMES cohort, after further adjusting for area-level score of factor 1, BMI, current smoking status, and presence of diabetes. Similarly, in the SINDI cohort, our multivariate regression analyses showed that increasing age, lower education level, lower income, and being a factory worker were significantly associated with visual impairment. 

Table 2.

 

Associations between Individual-Level Socioeconomic Measures and Bilateral Visual Impairment

	Singapore Malay Eye Study		Singapore Indian Eye Study
	Age-Adjusted OR (95% CI)	Multivariate-Adjusted OR (95% CI)*	Age-Adjusted OR (95% CI)	Multivariate-Adjusted OR (95% CI)*
Age, y	1.19 (1.16–1.21)	1.14 (1.11–1.16)	1.12 (1.10–1.14)	1.08 (1.06–1.11)
Sex, female	2.93 (2.19–3.92)	1.31 (0.76–2.28)	2.17 (1.52–3.10)	1.18 (0.62–2.23)
Education level
Polytechnic/university	Reference	Reference	Reference	Reference
Secondary education	2.35 (0.28–19.55)	2.14 (0.25–18.64)	2.79 (1.03–7.53)	2.53 (0.76–5.98)
Primary education	4.37 (0.59–32.55)	3.00 (0.39–23.36)	3.96 (1.57–9.93)	2.51 (0.95–6.65)
No formal education	13.54 (1.82–100.92)	6.56 (1.03–51.75)	12.02 (4.65–31.07)	6.52 (2.31–18.43)
Income
≥S$1000	Reference	Reference	Reference	Reference
Retirement income	5.97 (2.58–13.80)	2.53 (1.05–6.13)	2.02 (1.06–3.83)	0.33 (0.04–2.67)
<S$1000	7.75 (3.23–18.61)	2.62 (1.02–6.72)	1.61 (1.08–2.41)	2.52 (1.56–4.05)
Housing type
5-room flat/private housing	Reference	Reference	Reference	Reference
3–4-room flat	0.92 (0.59–1.43)	0.96 (0.59–1.56)	2.02 (1.06–3.83)	1.30 (0.62–2.71)
1–2-room flat	1.39 (0.85–2.27)	1.22 (0.70–2.12)	1.61 (1.08–2.41)	1.28 (0.81–2.03)
Marital status
Married	Reference	Reference	Reference	Reference
Never married	1.53 (0.52–4.52)	1.60 (0.52–4.91)	2.00 (0.83–4.80)	2.14 (0.83–5.52)
Separated/divorced	1.26 (0.59–2.68)	0.90 (0.38–2.17)	0.82 (0.25–2.69)	0.77 (0.23–2.60)
Widowed	2.22 (1.65–2.99)	1.41 (0.99–2.00)	1.71 (1.16–2.52)	0.87 (0.53–1.44)
Occupation
Service work	Reference	Reference	Reference	Reference
Professional/office work	†	†	0.18 (0.03–0.98)	0.20 (0.04–1.22)
Factory work	1.33 (0.55–3.23)	1.09 (0.43–2.77)	4.40 (1.02–19.00)	4.32 (1.04–19.2)
Homemaking	4.30 (2.39–7.73)	2.14 (1.01–4.53)	1.76 (0.62–5.01)	0.88 (0.29–2.72)
Unemployed/other	2.19 (1.20–4.00)	2.02 (1.03–3.96)	0.85 (0.30–2.42)	0.64 (0.22–1.88)

*  All individual-level socioeconomic measures were simultaneously adjusted for each other and were further adjusted for area-based score of factor 1, body mass index, current smoking status, and presence of diabetes.

†  No person with visual impairment fell into this category.

Table 3 shows the associations between area-level SES measures and visual impairment. In the SiMES cohort, many area-level SES measures (including major ethnicity, poor house, not speaking English, no university diploma, low income, family nucleus, low occupation attainment) had significant associations with visual impairment in the age-adjusted regression analysis. In multivariate regression analyses adjusting for age, gender, current smoking status, BMI, presence of diabetes, marital status, and education level, major ethnicity, poor house, not speaking English, no university diploma, low income, low occupation attainment, and Factor 1 score were significantly associated with visual impairment. These associations persisted after adjusting for different combinations of individual-level SES measures. 

Table 3.

 

Associations of Area-Level Socioeconomic Measures with Bilateral Visual Impairment

Area-Level Socioeconomic Measures*	OR of Bilateral Visual Impairment
	Age-Adjusted*		Multivariate-Adjusted Analysis 1†		Multivariate-Adjusted Analysis 2‡		Multivariate-Adjusted Analysis 3§
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Singapore Malay Eye Study
Major ethnicity (C2)	1.11	1.03–1.19	1.10	1.02–1.19	1.11	1.03–1.21	1.11	1.03–1.21
Poor house (C4)	1.08	1.01–1.17	1.07	1.02–1.16	1.08	1.01–1.17	1.07	1.02–1.16
Not speaking English (C5)	1.11	1.03–1.20	1.09	1.01–1.19	1.09	1.02–1.20	1.11	1.02–1.21
No university diploma (C11)	1.11	1.02–1.20	1.08	1.03–1.18	1.11	1.02–1.21	1.08	1.04–1.18
Low income (C14)	1.09	1.02–1.18	1.08	1.05–1.16	1.09	1.04–1.17	1.08	1.03–1.16
Family nucleus (C16)	0.97	0.90–1.04	0.94	0.88–1.02	0.90	0.83–1.08	0.91	0.84–1.12
Low occupational attainment (C17)	1.10	1.02–1.19	1.09	1.01–1.17	1.09	1.01–1.18	1.09	1.01–1.17
Score of factor 1	1.23	0.99–1.52	1.24	1.08–1.53	1.23	1.04–1.53	1.27	1.02–1.58
Singapore Indian Eye Study
Major ethnicity (C2)	1.12	1.00–1.24	1.09	1.03–1.22	1.10	1.03–1.24	1.11	1.03–1.25
Poor house (C4)	1.17	1.06–1.31	1.14	1.01–1.28	1.14	1.02–1.29	1.15	1.03–1.29
Not speaking English (C5)	1.12	1.10–1.14	1.09	0.97–1.22	1.10	0.98–1.24	1.11	1.04–1.25
No university diploma (C11)	1.12	1.10–1.14	1.11	1.04–1.25	1.12	1.05–1.26	1.13	1.00–1.27
Low income (C14)	1.12	1.10–1.14	1.11	1.04–1.25	1.12	1.06–1.26	1.13	1.00–1.27
Family nucleus (C16)	1.13	1.11–1.15	0.93	0.83–1.04	0.93	0.83–1.05	0.92	0.83–1.04
Low occupational attainment (C17)	1.13	1.11–1.15	1.11	1.00–1.22	1.11	1.00–1.23	1.12	1.02–1.24
Score of factor 1	3.26	1.26–8.42	2.66	1.03–7.39	2.77	1.02–7.92	3.01	1.07–8.39

*  For the meaning and details of the area-level socioeconomic measures, see Supplementary Table S1 DCSupplemental). Every area-level measure was separately analyzed in different logistic regression models.

†  Multivariate-adjusted analysis 1: OR adjusted for age, sex, current smoking, BMI, presence of diabetes, marital status, and individual education level.

‡  Multivariate-adjusted analysis 2: OR adjusted for covariates in model 1 plus individual occupation, individual monthly income, and housing type.

§  Multivariate-adjusted analysis 3: OR adjusted for covariates in model 1 plus individual socioeconomic status index.

Similar associations were found in the SINDI cohort, in which major ethnicity, poor house, no university diploma, low income, low occupation attainment, and Factor 1 score were significantly associated with visual impairment in the multivariate regression analyses. The proportions of persons with visual impairment, stratified by individual education categories and area-level SES categories, are shown in Figure 1. Persons with both low individual educational level and low area-level SES were more likely to have visual impairment than were those with either low individual-level SES or low area-level SES alone. Our statistical analyses showed that there was no evidence of statistical interaction between individual-level SES and area-level SES (P > 0.05). 

Figure 1.

 

Prevalence of visual impairment by individual educational level and area-level socioeconomic status (area-level SES). Individual educational level was divided into four categories: very low, no formal education; low, primary education; high, high school education; very high, polytechnic or university education. Area-level SES was divided into two categories, with the three lower SES districts classified as low SES and the other districts classified as high SES.

Discussion

This is the first population-based study to report the independent associations of area-level SES measures with visual impairment. Singapore is a city-state country of 712.4 km² with a population of 5 million people. Even within this small area, geographic variation exists among people aged 40 years and older. Specifically, people living in socioeconomically disadvantaged areas were more likely to have poorer visual acuity. The associations persisted even after adjustment of individual-based socioeconomic measures (e.g., education, income, and housing type) and other risk factors (e.g., presence of diabetes). To the best of our knowledge, only the Baltimore Eye Study (BES) and the Melbourne Visual Impairment Project (VIP) have previously reported the influences of area-level SES surrogates on visual impairment, but their statistical analyses were limited by the lack of adjustment for major confounding factors (i.e., individual-level SES measures). Using the 1980 US census tracts, the BES found that every US$1000 decrease in area-based median household income was associated with a 0.32% increase in the prevalence of visual impairment (P < 0.005). ² Based on the 1991 Australia census, the VIP found that people living in an area with a median household income AU$35,000 or more were 68% less likely to have visual impairment, though the difference did not reach statistical significance. ⁵  

The underlying mechanisms linking area-based socioeconomic measures and visual impairment remain unclear. It has been argued that area-based measures are indeed surrogates of individual-based socioeconomic status, but this is not supported by our multivariate analysis that the effects of area-level measures were not confounded by individual-level measures. From a public health perspective, many geography-specific characteristics may affect visual health in ways that are independent of individual socioeconomic circumstances. ^{14 –17} For example, it has been suggested that accessibility to neighborhood resources (e.g., recreation facilities, food stores) can shape people's lifestyle preferences and health behaviors over long periods of time and thus affect a range of health measures and chronic diseases. ^24,25 Population density, social support, and environmental pollution may also indirectly influence health conditions through psychosocial mechanisms. ³⁴ It is unlikely that the provision of and accessibility to health facilities play a significant role given that Singapore offers universal health care coverage to its citizens. Patients are free to choose among providers in the private or government clinics, and there is no transportation barrier between the districts. The use of geographic information system (GIS) technology has now provided a unique opportunity to quantify geographic accessibility to neighborhood resources. ³⁵ Further epidemiologic studies based on GIS may provide valuable insight into the influences of residential environment (or the so-called built environment) on visual health. 

The influence of individual educational level on visual impairment is not a surprise because it has been consistently reported across different populations. ^{1 –11} People without formal education are six times more likely to have visual impairment than those with formal education. Although the impact of education has been recognized for decades, appropriate health policies to improve health among less educated people has never been clear. ¹³ There is a need for continued investigation to explain why less educated people are not seeking/receiving eye care, even if they know that this could improve their vision. Unlike education, income is more vulnerable to reporting error. Income varies in different stages of life and is generally difficult to quantify among retired persons because their earnings are not as high as they were before retirement. Therefore, we might have underestimated the influence of individual income. Interestingly, we found that the effect of occupational status differs between ethnic Malays and Indians. In the Malay cohort, the homemaker and the unemployed were more likely to have visual impairment, whereas in the Indian cohort, factory workers were at increased risk for visual loss. There is no apparent explanation for the discrepancy, but our finding is consistent with the view of Braveman et al. ³⁶ that socioeconomic measurement must be context specific. A particular socioeconomic effect identified in one population may not always be applicable to other ethnic groups, places, and social conditions, even if all are under the same health care system. 

The strengths of this study include the large and representative samples of ethnic Malay and Indian cohorts, the involvement of both individual-level and area-level socioeconomic measures, and the use of multilevel analysis. There are many limitations as well. First, area-based measures are susceptible to ecological fallacy, in which all persons living in the same area are assumed to have same or similar characteristics. ^37,38 However, ecological characteristics can be measured only in an aggregate manner, not a per person manner. ^{14 –17} Second, the area-level measures used in this study were based on the availability of the data; these measures may not be the most appropriate scales for measuring area-level SES. Third, we cannot exclude the possibility of self-selection bias, which may account for the district variations. Fourth, owing to the cross-sectional nature of our studies, we were unable to determine whether the development of visual impairment was secondary to the exposure of adverse neighborhood characteristics. Finally, our study population was confined to a small number of geographic districts, after arbitrarily excluding the regions with fewer than 50 participants. Thus, our districts may not be nationally representative of all the regions in Singapore. 

Our study shows that both individual-level and area-level factors are important determinants of bilateral visual impairment. Individual-level and area-level indicators may reflect different aspects of social stratification. The use of either measure alone may not fully account for the degree of health inequalities. The problem of socioeconomic differences in visual impairment is not simply an issue of inadequate access to health education. Our data support the need for eye care policies that incorporate individual, environmental, and geographic information. 

Supplementary Materials

Table st1, DOC - Table st1, DOC 

Footnotes

 Supported by Biomedical Research Council Grant 08/1/35/19/550 and National Medical Research Council Grant STaR/0003/2008, Singapore.

Footnotes

 Disclosure: Y. Zheng, None; E. Lamoureux, None; E. Finkelstein, None; R. Wu, None; R. Lavanya, None; D. Chua, None; T. Aung, None; S.-M. Saw, None; T.Y. Wong, None