Objectives: To determine the usefulness of a single, fasting blood glucose (FBG) value in measuring the prevalence of diabetes mellitus in a large, homogeneous population.
Design: Fasting blood glucose and 2-h oral glucose tolerance test (OGTT) values were determined. Based on the results of the OGTT, the sensitivity and specificity of different cut-off levels of FBG for the diagnosis of diabetes were assessed. ROC (receiver operating characteristic) analysis was performed on the data.
Setting: A health screening unit at the University Hospital in Malmö, Sweden.
Subjects: A total of 1843 Caucasian women without known diabetes aged 55-57 years.
Results: The prevalence of previously undiagnosed diabetes was 3.9% and the prevalence of impaired glucose tolerance (IGT) was 27.9% using the WHO cut-off values for 2-h blood glucose values after an OGTT. With an FBG cut-off value of 6.7 mmol L-1, the sensitivity of a single FBG value was 36.6%. Reducing the cut-off value to 6.0 mmol L-1 increased the sensitivity to 53.4%. At a cut-off level of 4.8 mmol L-1, the sensitivity reached the high value of 85.9%, but the specificity was only 45% and the predictive value of a positive test as low as 5.9%. ROC analysis showed that the optimal cut-off value for FBG in this population was 5.3 mmol L-1, giving a sensitivity and specificity of 77% but a positive predictive value of only 11.9%.
Conclusions: This study has shown that in a large and homogeneous Caucasian population of women aged 55-57 years with a high prevalence of IGT, a single FBG value is not useful as a screening tool for diabetes mellitus.