Should the hemoglobin A1c diagnostic cutoff differ between blacks and whites? A cross-sectional study

Abstract

Background: Hemoglobin A1c (HbA1c) levels are known to be consistently higher in black persons than in white persons at any given glycemic level. Whether the optimal diagnostic threshold of HbA1c should differ between blacks and whites is unclear.

Objective: To compare the relationships between HbA1c level and the prevalence of retinopathy in black and white U.S. adults.

Design: Cross-sectional study.

Setting: A nationally representative sample of the National Health and Nutrition Examination Survey from 2005 through 2008.

Patients: 2804 white persons and 1008 black persons aged 40 years or older in the United States.

Measurements: Prevalence of retinopathy. Logistic regression models and restricted cubic spline models were constructed separately for white and black populations to test the HbA1c levels at which risk for retinopathy begins to increase.

Results: After adjustment for age, sex, hypertension, body mass index, family history of diabetes, and use of antidiabetes medications or insulin, the lowest HbA1c category for which the prevalence of retinopathy was significantly higher than the reference category (<5.5%) was 6.0% to 6.4% for white persons (risk difference, 4.8% [95% CI, 0.5% to 9.1%]) and 5.5% to 5.9% for black persons (risk difference, 5.3% [CI, 1.0% to 9.5%]). The restricted cubic spline models indicated that the risk for retinopathy increased at lower HbA1c levels in black persons than in white persons.

Limitation: The cross-sectional design of the study precluded examining the effect of the duration at each HbA1c level.

Conclusion: The prevalence of retinopathy begins to increase at a lower HbA1c level in black Americans than in white Americans. The findings do not support increasing the diagnostic threshold of HbA1c in black persons.

Primary funding source: None.

Figure 1

Results of restricted cubic spline models showing the association between HbA_1c value and the probability of retinopathy in U.S. adults aged 40 years or older not treated for diabetes. The lines represent the probability determined from restricted cubic spline models, with knots specified at HbA_1c levels of 5.5%, 6.0%, 6.5%, and 7.0%.

Figure 2

Receiver-operating characteristics curve of HbA_1c for detecting the prevalence of retinopathy at each possible threshold in U.S. adults aged 40 years or older not treated for diabetes. Adjusted for age, sex, hypertension, BMI, and family history of diabetes.