Diabetes Mellitus and Visual Impairment: National Health and Nutrition Examination Survey, 1999-2004

Objective  To examine the prevalence and correlates of visual impairment (VI) among US adults with and without diabetes mellitus.

Methods  Using National Health and Nutrition Examination Surveys conducted during 1999-2004, we estimated the prevalence of presenting (correctable or uncorrectable), correctable, and uncorrectable VI among Americans 20 years or older with and without diabetes. Data were weighted to make estimates representative of the US civilian noninstitutionalized population. We used multivariate logistic regression to calculate odds ratios and corresponding 95% confidence intervals.

Results  Approximately 11.0% of US adults with diabetes had some form of VI (3.8% uncorrectable and 7.2% correctable). Among those without diabetes, 5.9% had some form of VI (1.4% uncorrectable and 4.5% correctable). People with diabetes were more likely to have uncorrectable VI than those without diabetes, even after controlling for selected other factors (P < .05). Our findings also suggest a strong association between VI (correctable and uncorrectable) and older age, member of racial/ethnic minorities, lower income, and lack of health insurance, all independent of diabetes status (P < .05).

Conclusions  Vision loss is more common in people with diabetes than in people without diabetes. Diverse public health strategies are needed to reduce the burden of both correctable and uncorrectable VI.

In 2005, an estimated 14.6 million Americans had diagnosed diabetes mellitus, and an additional 6.2 million had undiagnosed diabetes.¹ By 2050, it is projected that the number of individuals with diagnosed diabetes in the United States will increase to 48.3 million.² In the meantime, a previous study, based on National Health Interview Survey data, suggested that visual impairment (VI) continued to be a major public health issue for the US population.³ Diabetic retinopathy, one of the most common microvascular complications of diabetes, is considered to be one of the major causes of blindness and low vision.^4,5 Even though some evidence suggests that intense glucose and blood pressure control has resulted in a reduction in the cumulative incidence of proliferative retinopathy,⁶ diabetes may increase the risk of VI due to other ocular conditions, such as cataract^7-9 and glaucoma.^10,11 In addition, VI due to refractive error is common among people with diabetes.^12,13

We analyzed data from the National Health and Nutrition Examination Survey (NHANES), a population-based national sample that was collected during 1999-2004, to describe the level of VI among people with and without diabetes and to examine the relationship of VI to diabetes mellitus, as well as to other risk factors. Previous studies on VI and diabetes have either used data from selected populations, limited to small areas,^14,15 or self-reported VI outcomes.¹⁶

Detailed information on NHANES data and its vision component has been shown earlier.¹³ In brief, NHANES is an ongoing series of cross-sectional surveys on health and nutrition that are conducted by the Centers for Disease Control and Prevention and is designed to be nationally representative of the noninstitutionalized US civilian population.¹⁷ All of the surveys include a household interview followed by a detailed physical examination. The survey was institutional review board approved and the sampled population was interviewed and examined with informed consent. For the 1999-2004 NHANES surveys, the physical examination also included a vision examination. Distance visual acuity (VA) was measured before and after an objective autorefraction test (optical correction), using the ARK-760 (Nidek Co Ltd, Tokyo, Japan) autorefractor containing built-in VA charts. Visual acuity was measured with whatever forms of current correction (eg, glasses or contact lenses) that the participant might have worn at the time of the examination. Only those eyes with a presenting VA of 20/30 or worse were administered the autorefraction test.

In the NHANES surveys conducted during 1999-2004, the combined household interview response rate was approximately 82% and the medical examination response rate was approximately 77%. Persons who were completely blind, who were unable to see in both eyes, or who had a severe infection in one or both eyes were excluded from the vision examination. Of the 15 332 sampled adults 20 years and older who received vision examination, 22 were excluded because of lack of diabetes information or because their diabetes was diagnosed only during pregnancy. Another 2306 adults for whom presenting VA values were missing were excluded from the study. We compared VA before and after an objective autorefraction test among 1237 adults with diabetes and 11 767 without diabetes.

The definitions of VI that we used, similar to those that are used in most domestic and international population-based studies,^18-22 appear in Table 1. We considered 3 classes of VI: presenting VI (any VI, correctable or uncorrectable), uncorrectable VI, and correctable VI. Presenting VI was defined as presenting VA worse than 20/40 in the better-seeing eye before an objective autorefraction test. Uncorrectable VI was defined as VA worse than 20/40 in the better-seeing eye after an objective autorefraction test. Correctable VI (ie, VI due to refractive error) was defined as VA worse than 20/40 in the better-seeing eye before an objective autorefraction test that could be improved to normal (VA≥20/40) after an objective autorefraction test. We also stratified each class into moderate impairment (20/200<VA<20/40) and severe impairment (VA≤20/200).

A respondent was considered to have diabetes if he or she had a self-reported previous diagnosis of the disease (excluding gestational diabetes mellitus). Age was categorized as 20 to 64 years and 65 years and older. Race/ethnicity was categorized as non-Hispanic white, non-Hispanic black, Mexican American, and other (including those who selected multiple races and non–Mexican American Hispanic). Other factors included in the analysis were sex, marital status, educational level (<high school, high school, and >high school), income (poverty income ratio<1 [ie, below the poverty threshold], 1 to <2, and ≥2), health insurance, high blood pressure, smoking, and body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared). High blood pressure affects the development of age-related macular degeneration,²³ diabetic retinopathy,^24,25 and other retinal vascular diseases.²⁶ We defined high blood pressure as an average (based on 4 readings) systolic blood pressure of 140 mm Hg or higher or an average diastolic blood pressure of 90 mm Hg or higher. Smoking is associated with age-related macular degeneration,^23,27-29 cataract,^30-32 and diabetic retinopathy.^15,33 We defined smoking status as current smoker, previous smoker, and those who never smoked. We controlled for BMI because some of the studies suggested a possible association between BMI and several ocular conditions, such as age-related maculopathy^34-37 and cataract.^38,39 We divided respondents into 3 groups based on their BMI: normal/underweight (<25), overweight (25 to <30), and obese (≥30).

We estimated the prevalence of presenting, uncorrectable, and correctable VI among the US noninstitutionalized population with and without diabetes mellitus aged 20 years and older. All of the analyses were weighted to make estimates that were representative of the population. We used SAS version 9.1 (SAS Institute, Inc, Cary, North Carolina) for data management and SUDAAN version 9.0 (Research Triangle Institute, Research Triangle Park, North Carolina) to adjust for the complex design of the sample and for nonresponse. The relationship of VI to diabetes, as well as to other risk factors, was explored using multivariate logistic regression; we calculated odds ratios (ORs) and corresponding 95% confidence intervals (CIs). Associations between VI and diabetes were considered to be significant if the P value for testing the null hypothesis of no association was <.05.

Table 2 shows the characteristics of the respondents with and without diabetes. The average age among the respondents with diabetes was 59 years. The average age among those without diabetes was 45 years. Non-Hispanic black individuals were more prevalent in the populations with diabetes than among those without diabetes. Adults with diabetes were more likely to have a lower educational level and less income than those without diabetes. There were fewer current smokers and more former smokers among people with diabetes. In addition, people with diabetes were more likely to be obese and to have higher blood pressure.

The overall prevalence of presenting VI among the participants with diabetes was 11.0% (moderate, 9.7% and severe, 1.4%); among those without diabetes, the prevalence was 5.9% (5.1% and 0.9%, respectively) (Figure 1). After optical correction (objective autorefraction test), the prevalence of uncorrectable VI was 3.8% among the adults with diabetes (moderate, 2.9% and severe, 1.0%) and 1.4% among those without diabetes (1.2% and 0.3%, respectively). Overall, the prevalence of correctable VI among the respondents with diabetes was 7.2% (95% CI, 5.5%-9.4%) and 4.5% (95% CI, 4.2%-4.9%) among those without diabetes.

People with diabetes were more likely to have presenting VI across all age groups, except among the oldest of the old group (≥80 years); in this age group, the prevalence was similar in both groups (Figure 2). After the age of 40 years, the prevalence of uncorrectable VI increased with age and people with diabetes were more likely to have uncorrectable VI than those without diabetes. The difference between people with and without diabetes increased as people became older until they reached the oldest of the old group. People with diabetes also had a higher prevalence of correctable VI across different age groups except the oldest of old group. The difference between people with and without diabetes of correctable VI was largest among the younger age groups.

After controlling for age, people with diabetes were more likely to have correctable and uncorrectable VI than people without diabetes (Table 3). If we added into the model other confounders, especially socioeconomic factors and ocular risk factors, diabetes was no longer associated significantly with correctable VI (OR, 1.18; 95% CI, 0.77-1.81). However, adults with diabetes were still more likely to have uncorrectable VI than those without diabetes, even after controlling all other factors (OR, 1.49; 95% CI, 1.01-2.20).

As shown in Table 3, both uncorrectable and correctable VIs were associated independently with older age, lower income, and no health insurance after controlling for all other factors. The impact of aging on uncorrectable VI was much greater than that on correctable VI. People with less education were more likely to have uncorrectable VI, especially those with less than a high school education (OR, 1.87; 95% CI, 1.30-2.68). Non-Hispanic black individuals were more likely to have uncorrectable VI than non-Hispanic white individuals (OR, 1.57; 95% CI, 1.08-2.29). In addition, non-Hispanic black individuals and Mexican American individuals had higher odds of having correctable VI than non-Hispanic white individuals (OR, 1.82; 95% CI, 1.41-2.33 and OR, 1.80; 95% CI, 1.36-2.39, respectively). There were no significant differences in correctable and uncorrectable VI by sex or marital status. We also tested interaction effects including interaction of age and diabetes on VI but found no significant effects.

Correctable, as well as uncorrectable, VI can increase the likelihood of injury and lead to lower quality of life.^40-47 Our findings suggest that people with diabetes had a 60% and 170% higher prevalence of correctable and uncorrectable VI, respectively, than those without diabetes. The risk of uncorrectable VI remained associated significantly with diabetes, even after controlling for age, sex, marital status, race/ethnicity, educational level, income, health insurance, high blood pressure, smoking, and BMI. However, we did not find a statistically significant association between correctable VI and diabetes after adjusting for all other selected factors.

Based on the 2000 US Census and population-based studies in the United States, Australia, and Europe, the Eye Disease Prevalence Research Group (EDPRG), a group of principal investigators of several population-based vision studies, estimated that 2.76% (moderate, 1.98% and severe, 0.78%) of Americans 40 years and older had uncorrectable VI, including blindness and low vision.⁵ In our study, we found that uncorrectable VI increased with age, especially after the age of 40 years. The prevalence of uncorrectable VI among US adults with diabetes 20 years and older was 3.8% (moderate, 2.9% and severe, 1.0%) and was 1.4% among those without diabetes (moderate, 1.2% and severe, 0.3%). Our estimates are lower than the EDPRG estimate, even if we only include those 40 years and older (2.1% [moderate, 1.7% and severe, 0.4%]); this is possibly because we only included the community-dwelling population and excluded those who were unable to see.

Diabetic retinopathy affects approximately 3.4% of US adults 40 years and older and 0.75% have vision-threatening retinopathy.⁴⁸ Moreover, many cases of diabetic retinopathy are asymptomatic in their early stages. Results from one study have indicated that retinopathy may, under current diagnostic criteria, occur in the prediabetic stage.⁴⁹ Our findings indicated an independent association between diabetes and uncorrectable VI. As a result, early diagnosis and the prompt treatment of diabetes mellitus is extremely important to reduce the incidence of uncorrectable VI due to diabetic retinopathy and other ocular conditions that are associated with diabetes. To prevent and reduce uncorrectable VI among people with diabetes, good glycemic control and good blood pressure control are necessary.⁵⁰ Moreover, early detection and the treatment of diabetic retinopathy through dilated eye examination and photocoagulation could prevent and delay visual loss from this microvascular complication.

While most of the population-based studies emphasize uncorrectable VI, a few have focused on correctable VI among children and adults.^{18,21,22,51-54} The Baltimore Eye Survey²² reported that more than 50% of the subjects improved their presenting vision after refractive correction. The Proyecto VER (Vision Evaluation and Research) Study⁵³ found that uncorrected refractive error accounted for 73% of the impaired VA among Mexican American individuals 40 years and older. Similar findings were reported in the United Kingdom,⁴⁵ India,⁵⁵ and Australia.^56-58 Recently, the National Eye Institute estimated the prevalence and cost of correctable VI of the general population and emphasized the importance of correcting VI due to refractive error in improving safety and quality of life.^13,59 The Blue Mountains Eye Study⁵⁶ also found a lower prevalence of presenting VI in their second cross-sectional phase, which indicated a possible impact of public education and community awareness. Consistent with previous studies, our findings suggest that a large proportion of people with diabetes (65%) can improve their presenting vision through the provision of a simple accurate spectacle prescription (eg, glasses or contact lenses). In particular, we found that the ratio of correctable VI vs uncorrectable VI was much higher among the younger population with diabetes (<40 years), which highlights the importance of reducing correctable VI through diverse public health interventions that are focused on different groups in need.

Our results showed that the prevalence difference of correctable VI between adults with and without diabetes is not mainly due to age. The strength of the association between correctable VI and diabetes was reduced by adjusting for other socioeconomic and ocular risk factors. These findings are consistent with previous research from Proyecto VER⁵³ and the Blue Mountains Eye Study,⁵⁶ which suggested that the socioeconomic factors, probable markers of limited access to health care services, were associated with correctable VI. The 1994 Robert Wood Johnson Foundation National Access to Care Survey⁶⁰ suggested that the financial barrier (could not afford glasses or no insurance) was the main reason that people reported an unmet need for eyeglasses.

The addition of VA and refractive error assessment, in concert with dilated fundus examination, may further contribute to improved vision outcomes for individuals with diabetes. Our findings underscore the importance of the use of an annual comprehensive eye examination (CEE) for people with diabetes as recommended by the American Diabetes Association,⁶¹ the American Optometric Association,⁶² and the American Academy of Ophthalmology.⁶³ A regular CEE will include medical and ocular history, VA measurement (refraction if needed), intraocular pressure measurement, and the examination of the vitreous humor, retina, and optic nerve head. An annual CEE will help to identify the risk factors for uncorrectable VI, detect diabetic retinopathy and other ocular conditions at their early stages, ensure appropriate and continuous eye care, and thus prevent or delay permanent vision loss; a CEE will also help to find out and correct VI due to refractive error and, therefore, improve the individual's visual ability and quality of life. However, given that a high prevalence of VI and low use of recommended eye care services^64,65 exist among people with diabetes, it is important to enhance vision screening^66,67 and public health education programs⁶⁸ to identify high-risk populations and educate the public in the need for ocular correction and eye care.

Our study is subject to several limitations. First, the institutionalized population (eg, nursing home) was not included in NHANES. This may result in possible underestimates of VI prevalence, as suggested by the Blue Mountains Eye Study.⁶⁹ Second, because the participants who were completely blind, unable to see in both eyes (estimates not available), or had a severe infection in one or both eyes (n = 8) were excluded from vision examination, there is further underestimation in the prevalence estimates of VI. Third, only objective refraction (objective autorefraction test) was measured and used to support final visual improvement. Because there is no subjective refinement, VA estimates after an objective autorefraction test may not be considered as “best corrected,”⁷⁰ resulting in a possible underestimate to correctable VI. Fourth, although presenting VA was tested with current usual correction, if available, some people may not have worn their usual correction. As a result, the prevalence of correctable VI could be overestimated. Fifth, although we used data that were collected in 6 years, the sample size in some age groups was small. Estimates having a relative standard error higher than 30% were considered to be statistically unreliable (eg, prevalence of correctable severe VI among people with diabetes) and should be interpreted with extreme caution. Finally, although previous studies have suggested that a self-report of physician-diagnosed diabetes is accurate and valid,^71-73 our reliance on self-reported diabetes status may underestimate the impact of diabetes on VI because many individuals with undetected hyperglycemia are missed.⁷⁴

As the US population ages and changes demographically, the social and economic burden of VI may increase dramatically. The high prevalence of VI among people with diabetes indicates a need for diverse public health strategies to reduce the burden of both correctable and uncorrectable VI. It is important to identify and pursue ways to increase access to eye care for everyone and to correct VI, where possible, to diminish morbidity and mortality due to impaired vision.

Correspondence: Xinzhi Zhang, MD, PhD, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, 4770 Buford Hwy, NE (K-10), Atlanta, GA 30341-3727 (xzhang4@cdc.gov).

Submitted for Publication: October 4, 2007; final revision received December 4, 2007; accepted January 16, 2008.

Financial Disclosure: None reported.

Disclaimer: The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.