Accuracy of HbA1c for diagnosing Diabetes Mellitus among Hong Kong Chinese

Abstract

Introduction: Since 2010, HbA1c ≥6.5% was recommended as a diagnostic criterion for the detection of diabetes mellitus (DM) by the American Diabetes Association (ADA). This new diagnostic criterion has quickly gained popularity among clinicians in Hong Kong for detecting DM among patients with impaired fasting glucose (IFG) in place of the 75-g oral glucose tolerance test (OGTT) for its convenience and stability, because HbA1c is not affected by fasting, acute stress or short-term lifestyle changes. However, its accuracy compared to the conventional plasma glucose tests remained controversial in the Chinese population. This study evaluated the accuracy of HbA1c by calculating its sensitivity and specificity for diagnosing DM among Hong Kong Chinese patients with IFG compared to the diagnosis by OGTT as the gold standard. Methods: This was a cross-sectional study piloted in two public outpatient clinics in Hong Kong from 30 May to 30 September 2013.Convenience sampling was employed. 238 non-diabetic Chinese adults with IFG (i.e., fasting blood glucose level between 5.6 and 6.9 mmol/L) who attended the two participating clinics within the study period were recruited to receive a paired 75-g OGTT and HbA1c test to determine their glycaemic status. Results: An HbA1c cut-off of ≥6.5% had a sensitivity of 22.9% and a specificity of 94.4% in predicting the presence of DM diagnosed by OGTT among Hong Kong Chinese adults who had IFG. The optimal cut-off threshold of HbA1c was 6.1%, with a sensitivity and specificity of 70.8% and 76.3%, respectively, and an area under the curve of 0.76. Conclusion: Among Hong Kong Chinese adults who had IFG, HbA1c ≥6.5% was highly specific for detecting undiagnosed DM, but its accuracy as a diagnostic test was sub-optimal because it missed 77.1% of the cases. An HbA1c threshold of ≥6.1% might be more suitable than ≥6.5% for predicting the presence of DM.