SUN-616 Poor Diagnostic Concordance Between Fasting Plasma Glucose and Glycosylated Hemoglobin in a Black South African Population

Background: While elevations in fasting plasma glucose (FPG) and glycosylated hemoglobin (HbA(1c)) are both recognized by the American Diabetes Association (ADA) as diagnostic of hyperglycemia, previous comparisons of these tests have demonstrated discordant individual classifications and population estimates. This may be due to additional postprandial glycemia reflected by HbA(1c) and, in African-descent populations, to non-glycemic factors that contribute to higher HbA(1c) at any given level of glycemia. We hypothesized that glycemic classifications based on FPG or HbA(1c) would differ in a Black South African population and investigated factors associated with discordance. Methods: 889 Black adults with previously undiagnosed diabetes, aged 40-79 years, from the population-based Health and Ageing in Africa: a Longitudinal Study of an INDEPTH Community in South Africa (HAALSI) cohort were included. Concordance between ADA FPG (normoglycemia [NG] <100 mg/dl, prediabetes [pre-DM] 100-125 mg/dl, diabetes [DM] ≥ 126 mg/dl) and HbA(1c) (NG <5.7%, pre-DM 5.7-6.4%, DM ≥ 6.5%) classifications was assessed using Cohen’s kappa statistic and logistic regression models were used to identify predictors of discordance. Results: Median age was 55 years (IQR 49-62) and 49.3% of the sample was male. Median glucose was 86.4 mg/dl and median HbA(1c) was 5.4%. Pre-DM, as defined by HbA(1c), was present in 204 participants (22.9%), while FPG-defined pre-DM was present in 122 (13.7%). DM defined by HbA(1c) was present in 146 (16.4%), while FPG-defined DM was present in 36 (4.0%). Concordance between the two tests was poor (kappa statistic 0.18; 95%CI 0.13-0.24). Self-reported history of tuberculosis (OR 1.90, p=0.026) and higher HbA(1c) (OR 4.70, p<0.001) were associated with increased likelihood of discordance, whereas higher fasting glucose was associated with decreased likelihood of discordance (OR 0.58, p<0.001). There was no association between discordance and hemoglobin, HIV status, BMI, waist circumference or hip circumference. Conclusion: FPG and HbA(1c) exhibit poor concordance in classifying hyperglycemia in this Black South African population, with HbA(1c)-based definitions identifying higher prevalences of pre-DM and DM. Further work is needed to confirm whether these discrepancies are due solely to elevations in postprandial glucose. In the interim, clinicians should consider confirming elevated HbA(1c) concentrations with oral glucose tolerance testing, particularly in those with a history of tuberculosis, prior to making a diagnosis of DM in this population.