Poor specificity of fasting plasma glucose cut-off values in ruling out glucose intolerance: the complementary...

With the objective to assess the diagnostic power of the ADA criteria in detecting glucose intolerance, we studied 654 patients by performing an oral glucose tolerance test (OGTT).

The design required computing sensitivity, specificity and predictive values for different cut-off levels of fasting plasma glucose (FPG). The patients were recruited in an outpatient facility of a General Internal Medicine department affiliated to a Medical School.

Lowering the threshold from 7.0 to 6.1, 5.6, 5.0 and 4.4 mmol/l (126 to 110, 100, 90 and 80 mg/dl) respectively, resulted in a progressive fall in specificity (99.6, 91.6, 76.0, 45.3 and 15.8%) combined to a rise in sensitivity (8.6, 30.7, 56.4, 76.4 and 92.9%). Two subjects with FPG <5.0 mmol/l were diabetic. At the lowest FPG of 4.4 mmol/l the percentage of glucose intolerant patients by OGTT was still 11% while at 6.1 mmol/l the false positive rate was 50%.

The simple screening based on FPG alone, as suggested by ADA, is ineffective, not detecting an unacceptable number of subjects with glucose intolerance, and conveys the high cost of working up an excessively large number of false positives. Only the OGTT represents the effective screening procedure.