Abstract

Aim: To determine glycosylated hemoglobin (HbA1c) cut-points for glucose intolerance in Asian Indians.

Methods: 2188 participants without known diabetes were randomly selected from the Chennai Urban Rural Epidemiology Study. All had fasting plasma glucose [FPG] and 2hr post glucose [2hr PG] measurements after 75g load and were classified as impaired fasting glucose [IFG] (ADA criteria: FPG ≥5.5 mmol/L and < 7 mmol/L and WHO criteria: FPG ≥6.1mmol/L and <7 mmol/L; impaired glucose tolerance [IGT]: 2hr PG ≥ 7.8 mmol/L and < 11.1 mmol/L, or diabetes: FPG ≥7mmol/L and/or 2hr PG ≥ 11.1 mmol/L. HbA1c was measured using the Biorad Variant machine. Based on Receiver Operating Characteristic curves, optimum sensitivity and specificity were derived for defining HbA1c cut points for diabetes, IGT and IFG.

Results: Mean values of HbA1c among NGT, IGT and diabetes subjects were 5.5±0.4%, 5.9±0.6% and 8.3±2.0% respectively [p for trend <0.001] with considerable overlap. To identify diabetes based on 2hr PG, HbA1c cut point of 6.1% had area under the curve (AUC) of 0.941 with 88.0% sensitivity and 87.9% specificity. When diabetes was defined as FPG ≥ 7.0mmol/L, the HbA1c cut point was 6.4% (AUC=0.966, sensitivity 93.3%, specificity 92.3%). For IGT, AUC = 0.708; IFG – WHO, AUC = 0.632 and IFG – ADA, AUC=0.708, the HbA1c cut point was 5.6%.

Conclusion: In Asian Indians, HbA1c cut point of 6.1% and 6.4% defined diabetes by 2hr PG or FPG criteria respectively. A value of 5.6% optimally identified IGT or IFG, but was less than 70% accurate.