The Role of Hemoglobin A1C in Diabetes Screening and Diabetic Retinopathy

Abstract: Hemoglobin A1C (A1C) is used in various settings. Its performance has not been evaluated
systemically. We compared A1C in diagnosis of diabetes with fasting plasma glucose (FPG) and 2-h
postchallenged plasma glucose (2hPG) parameters in a cross-sectional cohort in the United Stated.
Adult subjects (≥20 years) were identified from the National Health and Nutrition Examination
Survey 2005–2016 without a history of diabetes who had BMI, A1C, FPG, and 2hPG (n = 10,416). For
comparisons, we calculated the sample weighted prevalence, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) with subgroup analyses. For the retinopathy
study, diabetic subjects with established diabetes who responded to the question of diabetic retinopathy were evaluated (n = 3907). Compared to the FPG/2hPG criteria, A1C ≥ 48 mmol/mol (6.5%) had
a low sensitivity at 25.90%, with specificity 99.70%, PPV 84.70%, and NPV 95.70%. Subgroup analyses
revealed a lower sensitivity in males (24.52%); the lowest in non-Hispanic White (21.35%), in the
third decade (14.32%), and in the BMI < 22.50 kg/m2 group (7.21%). The prevalence of self-reported
diabetic retinopathy increased drastically with an inflection point at A1C 48 mmol/mol (6.5%) from
11.52% to 18.32% (p < 0.0001). A1C ≥ 48 mmol/mol (6.5%) should be cautiously used to diagnose
diabetes in certain subgroups due to very low sensitivity in certain groups. With the confirmation of
the association of increasing self-reported diabetic retinopathy with A1C ≥ 48 mmol/mol (6.5%),
the current A1C cutoff is an acceptable value with the understanding of especially low sensitivity in
certain subgroups.
Keywords: diabetes mellitus; diagnosis; sensitivity; specificity; retinopathy