Diagnostic performance of glycated hemoglobin for diabetic retinopathy in non-diabetic older overweight/obese African-Americans.

OBJECTIVES

Although clinicians do not routinely screen for diabetic retinopathy in non-diabetic patients, previous studies have shown that diabetic retinopathy can occur in patients with prediabetes. However, due to the limitations of glycated hemoglobin (HbA1c) in overweight/obese subjects, African-Americans and older adults, little is known about the correlation between HbA1c and diabetic retinopathy in non-diabetic older overweight/obese African-Americans. The aims of this study were to determine the association between HbA1c and diabetic retinopathy, and the optimal diagnostic threshold of HbA1c that predicts diabetic retinopathy in non-diabetic older overweight/obese African-Americans.

METHODS

The 2005-2012 data from the U.S. National Health and Nutrition Examination Surveys (NHANES) were utilized for this study. Prevalence odds ratios from logistic regression analyses were used to estimate risks of diabetic retinopathy across HbA1c categories, adjusting for age, sex, and hypertension. Receiver operating characteristic curve was used to determine diagnostic cutoff point of HbA1c for prevalent diabetic retinopathy.

RESULTS

There were gradients of increasing prevalence and odds of diabetic retinopathy with increasing HbA1c in non-diabetic overweight/obese African-Americans 50years of age and older. HbA1c cut-off point of 5.2% (AUC=.726, 95% CI=0.696-0.756) was found to maximize sensitivity [93.5%; 95% CI: 83.2-95.7] for diabetic retinopathy, though specificity [22.1%; 95% CI 19.9-32.8] was low.

CONCLUSION

Current criteria for diagnosis of prediabetes are effective in identifying many older overweight/obese African Americans with diabetic retinopathy. Based on our analysis, a lower HbA1c of 5.2% could serve as a more sensitive cutoff point for defining prediabetes in this population subgroup.