To the Editor.
—In the Review entitled "A Clinical Approach for the Diagnosis of Diabetes Mellitus," Dr Peters and colleagues1 argue for the use of glycosylated hemoglobin A1c (HbA1c) in the diagnosis of diabetes mellitus. They highlight the practical and methodological difficulties in oral glucose tolerance testing and emphasize the benefit of using a diagnostic test that provides therapeutic information. They bolster their clinical arguments with a meta-analysis of studies measuring both oral glucose tolerance (the current "gold standard") and HbA1c. They use the meta-analysis to determine the sensitivity and specificity of an HbA1c with respect to the gold standard, but these data could be used in an additional way to help support their argument.One established method to evaluate a diagnostic test for binary classification (eg, diabetic or not diabetic) is to calculate the receiver operating characteristic (ROC) curve and to estimate the area under the