Accuracy of glycated haemoglobin in screening for pre-diabetes in Asian Indians--a community survey: the Chandigarh Urban Diabetes Study (CUDS).

Aim: To compare American Diabetes Association and International Expert Committee recommended cut-off values of HbA(1c) for detecting the presence of pre-diabetes against plasma glucose values obtained from oral glucose tolerance tests in Asian Indians.

Methods: A cross-sectional randomly sampled population survey involving 2368 adults, aged ≥ 20 years. HbA(1c) was measured on a Bio-Rad 10 system in 1972 subjects.

Results: Of the 1972 subjects studied, 329 were detected to have pre-diabetes based on isolated impaired fasting glucose in 125 subjects (6.3%), isolated impaired glucose tolerance in 141 subjects (7.1%) and the presence of both in 63 subjects (3.2%). The HbA(1c) cut-off of 34 mmol/mol (5.7%), as recommended by the American Diabetes Association for detecting the presence of pre-diabetes, showed sensitivity of 62%, specificity 77%, with a positive predictive value of 34.7%, a negative predictive value of 89.5% and accuracy of 67.8%; whereas the HbA(1c) cut-off recommended by the International Expert Committee of 42 mmol/mol (6%) had a sensitivity of 36%, specificity of 90%, positive predictive value of 42.7%, negative predictive of 85.4% and an accuracy of 77%. However, both these HbA(1c) cut-offs underdiagnosed the presence of pre-diabetes in 38 and 64% of these subjects, respectively.

Conclusions: The American Diabetes Association and the International Expert Committee recommended HbA(1c) cut-off values and oral glucose tolerance tests identify different pre-diabetes cohorts. Long-term prospective studies are required to define the usefulness of one over the other.