Combined use of a fasting plasma glucose concentration and HbA1c or fructosamine predicts the likelihood of having diabetes in high-risk subjects

OBJECTIVE - To assess the validity of using fasting plasma glucose (FPG) concentrations in conjunction with HbA(1c) or fructosamine for the screening of diabetes in high-risk individuals. RESEARCH DESIGN AND METHODS - In this study, 2,877 Hong Kong Chinese (565 [19.6%] men; 2,312 [80.4%] women) with various risk factors for glucose intolerance underwent a 75-g oral glucose tolerance test (OGTT) for screening of diabetes. The risk factors included a family history positive for diabetes, a history of gestational diabetes or impaired glucose tolerance, and obesity. RESULTS - Using World Health Organization (WHO) criteria, 1,593 (55.4%) had normal glucose tolerance, 657 (22.8%) had impaired glucose tolerance, and 627 (21.8%) had diabetes. When the 1997 American Diabetes Association (ADA) criteria were applied, 394 (13.7%) had diabetes with an FPG greater than or equal to 7.0 mmol/l. Using multiple receiver operating characteristic curve analysis, the paired values of an FPG of 5.6 mmol/l and a HbA(1c) of 5.5% gave an optimal sensitivity of 83.8% and specificity of 83.6% to predict a 2-h plasma glucose (PG) greater than or equal to 11.1 mmol/l. Likewise, the paired values of an FPG of 5.4 mmol/l and a fructosamine level of 235 mu mol/l (n = 2,408) gave an optimal sensitivity of 81.5% and specificity of 83.2%. An FPG greater than or equal to 5.6 mmol/l and an HbA(1c) greater than or equal to 5.5% was 5.4-fold more likely to occur in diabetic subjects (based on the WHO criteria) compared with nondiabetic subjects. For paired parameters less than these values, the likelihood ratio of this occurring in diabetic subjects was only 0.11. Similarly, an FPG greater than or equal to 5.4 mmol/l and a fructosamine greater than or equal to 235 mu mol/l was fivefold more likely to occur in diabetic subjects than in nondiabetic subjects, with both parameters less than these values having a likelihood ratio of 0.04. Using these paired values as initial screening tests, only subjects who had an FPG greater than or equal to 5.6 mmol/l and <7.8 mmol/l and an HbA(1c) greater than or equal to 5.5% (n = 642) required an OGTT to confirm diabetes, thereby saving 77.7% [(2,877 - 642)/2,877] of the OGTTs performed. Similarly, only subjects who had an FPG greater than or equal to 5.4 mmol/l and <7.8 mmol/l and a fructosamine greater than or equal to 235 mu mol/l (n = 526) required OGTT to confirm diabetes, meaning that 78.2% [(2,408 - 526)/2,408] of the OGTTs could have been saved. Based on the 1997 ADA criterion of an FPG cutoff value of 7.0 mmol/l, the corresponding numbers of OGTTs to be saved were 82.6% and 85.5%, respectively. CONCLUSIONS - The paired values of FPG and HbA(1c) or FPG and fructosamine helped to identify potentially diabetic subjects, the diagnosis of which could be further confirmed by the 75-g OGTT. Using this approach, similar to 80% of OGTTs could have been saved, depending on the diagnostic cutoff value of FPG.