Abstract
AIMS: This article performed a retrospective cohort study to estimate the annual incidence rates of diabetes and to assess the utility of HbA1c as a predictor for progression to diabetes in Chinese community adults aged 40 years or older. METHODS: In all, 2778 nondiabetic subjects (including 1901 women) underwent HbA1c testing and oral glucose tolerance test (OGTT) measurements at baseline and after 3 years. Diabetes and prediabetes were defined using the WHO criteria. The HbA1c cutoff points were evaluated to predict the future risks of diabetes. Relative risk (RR) was calculated using the chi-square test. The area under the receiver operating characteristic (ROC) curve (AUC) was used to evaluate the predictive efficiency of fasting plasma glucose (FPG), 2 hr postprandial plasma glucose (2hPG), and HbA1c for progression to diabetes. A superior cutoff point was defined as the point on the ROC curve with a larger Youden index. RESULTS: Overall, 7.5% (210/2778) of the subjects progressed to diabetes, yielding an annual 2.5% diabetes incidence rate. Additionally, 4.5% (100/2227) of the subjects with normal glucose tolerance (NGT) and 19.6% (110/561) of the subjects with prediabetes progressed to diabetes, and the relative risk of progression to diabetes was 5.188 times higher in subjects with prediabetes than in subjects with NGT (p < 0.001). Compared to subjects with HbA1c values ≤ 5.6%, the RRs of progression to diabetes in subjects whose HbA1c ranged from 5.7 to 5.8%, 5.9 to 6.2%, 6.3 to 6.4%, and ≥6.5% were 1.165, 2.582, 5.732, and 16.619, respectively. However, the RRs for subjects with HbA1c ranging from 5.7 to 5.8% and those with HbA1c ≤ 5.6% did not differ significantly (p = 0.615). The AUCs for predicting diabetes after 3 years by FPG, 2hPG, and HbA1c were 0.752 (95% confidence interval 0.718-0.787), 0.710 (95% confidence interval 0.671-0.748), and 0.756 (95% confidence interval 0.720-0.793), respectively. The HbA1c cutoff point of 5.9% (sensitivity of 0.771 and specificity of 0.580) may better identify individuals at high risk of progression to diabetes than the 5.7% value (sensitivity of 0.862 and specificity of 0.371) due to the former's larger Youden index of 0.351, which exceeded the indices for FPG and 2hPG. CONCLUSIONS: The use of HbA1c values ≥ 5.9% may provide greater accuracy in evaluating the risk of progression to diabetes and identify individuals with prediabetes with greater reliability among Chinese adults.