Approximately 21 million individuals in the US have diagnosed diabetes.¹ Diabetic retinopathy, a progressive condition that can ultimately lead to blindness, affects approximately 29% of adults aged 40 years and older with diabetes.² Diabetes is also associated with an increased likelihood of other conditions that may affect vision, such as cataracts and glaucoma.³ We assessed the association between diagnosed diabetes and self-reported trouble seeing while controlling for other covariates that may affect vision.

The present study used 2011–13 data from the National Health Interview Survey (NHIS).⁴ The NHIS is a nationally representative sample of US households, with one adult member of each family selected to complete a more in-depth survey. Final 2011–13 sample adult response rates ranged from 61% to 66%. Respondents were asked if they had trouble seeing even when wearing usual vision correction. Possible responses included “yes”, “no”, and “don’t know”, and were used to create a binary variable equal to one for those who did and zero for those who did not report trouble seeing (responses of “don’t know” and refusals [accounting for <0.01% of responses] were considered missing). Those who reported receiving a diabetes diagnosis from a healthcare provider were classified as having diabetes. The sample consisted of adults aged 25 years and older with complete demographic, comorbidity, vision, and diabetes information.⁵

Multivariate logistic regression analysis was used to describe the association between diagnosed diabetes and trouble seeing. We first adjusted for age and sex only, successively controlling for select demographic characteristics and diabetes-related comorbidities to investigate the possible reasons for any association. Sampling weights were used to produce nationally representative estimates and standard errors accounted for the complex design of the NHIS.

Before adjusting for covariates, approximately 19% of individuals with diagnosed diabetes had trouble seeing compared with approximately 8% of those without diabetes, a significant difference of over 10 percentage points (Table 1). Those with diagnosed diabetes were older, less educated, and more likely to have a history of comorbid conditions, characteristics that may also be associated with trouble seeing. Table 2 presents results for the multivariate logistic analysis of the correlates of trouble seeing. Model 1 includes an indicator for diagnosed diabetes, age group, and sex. Models 2–4 further control for race/ethnicity (Model 2), for marital status, health insurance status, education level, and census region (Model 3), and a history of comorbid conditions (Model 4). Before controlling for a history of comorbid conditions, having diagnosed diabetes was associated with approximately double the odds of self-reported trouble seeing (P < 0.01; Table 2 [Models 1–3]). After controlling for a history of comorbid conditions, the estimated odds ratio (OR) for having diagnosed diabetes decreased to 1.59 (P < 0.01;Table 2 [Model 4]).

In addition to the positive correlation between diagnosed diabetes and trouble seeing, we also found that older age, less education, and a history of comorbid conditions were positively associated with self-reported trouble seeing. Conversely, male sex, being married, and having health insurance were negatively associated with trouble seeing. For example, we estimated that having less than a high school degree compared with a bachelors degree or higher was associated with an OR of approximately 2 (P < 0.01). A history of high blood pressure, stroke, and any heart disease were associated with ORs of 1.42 (P < 0.01), 1.79 (P < 0.01), and 1.87 (P < 0.01), respectively, compared with no history.

The most recent analytical studies similar to ours using nationally representative US data to examine the association between diabetes and visual impairment used 1999–2004 data.^3,5 Herein, we provide up-to-date estimates of the association of diagnosed diabetes and other covariates with trouble seeing. Compared with the most similar previous study,⁵ we found a larger association between diagnosed diabetes and trouble seeing. For example, our Model 1 estimates implied an OR of 2.13, whereas comparable estimates from the previous study suggested an OR of 1.59.⁵

These findings are subject limitations. First, approximately 28% of all diabetes is undiagnosed,¹ suggesting that self-reported diabetes may underestimate true prevalence. To the extent that individuals in the present analysis were incorrectly classified as having or not having diabetes, the actual association between diabetes and trouble seeing is likely to be larger than the estimates we presented. Furthermore, trouble seeing was self-reported and it was not possible to distinguish more severe from milder forms of vision problems. Analysis of the association between diabetes and severe trouble seeing may have yielded different results than those reported here.

Reducing visual impairment due to common eye health problems that are responsive to treatment is a US national health objective.⁶ Most visual impairment is correctable or treatable.⁷ Establishing the factors associated with trouble seeing may help identify those individuals most at risk of vision loss.