Abstract: Objective To assess the validity of using fasting plasma glucose(FPG)and HbA1c for the screening of diabetes and pre-diabetes。Methods 457 subjects，（ 297men,160 women，average age 61.4士13.1years） in our hospital underwent an oral glucose tolerance test(OGTT) for screening diabetes．HbA1c was examined at the same time.Based on 1999 WHO criteria，81 had normal glucose tolerance (NGT)，7 had impaired fasting glucose(IFG)，76 had impaired glucose tolerance (IGT),13 had IGT and IFG，280 had diabetes．Using a receiver oerating characteristic curve(ROC curve)to judge the optimal cut-point.Result 1.the optimal cut-point of HbA1c for DM was 6.45% that was associated with a sensitivity and specificity of 0.794 and 0.825 respectively.Area under the curve was 0.871(95%CI0.839～0.903).2. The cut-point of HbAlc for pre-diabetes was 5.85% ，which was associated with a sensitivity and specificity of 0.583 and 0.531 respectively．Area under the curve was 0.587（95% C1 0.504～0.671).3 When only used HbA1c to screening DM, the sensitivity and specificity are 0.794 and 0.825 respectively, the rate of misdiagnosis is 21.3%;When only used FPG to screening DM, the sensitivity is 0.646,the rate of misdiagnos is 35.3%;But when using FPG or HbA1c to screening DM,the sensitivitiy and specificity are 0.871 and 0.623 respectively,the rate of misdiagnosis is 12.9%。 Conclusions 1.Suggest the optimal cut-point of HbA1c for diabetes was 6.45% ,because it have the high sensitivity and specificity.2.The cut-point of HbA1c for pre-diabetes is 5.85%.But the sensitivity and specificity are lower,so it have little diagnose meaning.3.The detection of FPG and HbA1c may reduce erroneous misdiagnosis rate.

Key words: Fasting plasma glucose, HbAlc, Oral glucose tolerance test, Diabetes mellitus, Receiver operator characteristic curve(R0C curve)