Effectiveness of screening for diabetes.

We examined the patient records of 6445 patient visits to an executive health surveillance program to evaluate the diagnostic yield from screening for diabetes mellitus by measurement of fasting serum glucose and hemoglobin A1c. We found increased fasting serum glucose levels (greater than or equal to 6.6 mmol/L) in 3% (197/6445), of whom only half received further confirmatory testing. Increased screening values for glucose were associated with a 70% incidence of diabetes mellitus and impaired glucose tolerance when subsequent oral glucose tolerance tests were performed. Confirmatory testing with the oral glucose tolerance test was equivalent in cost and superior in diagnostic yield to repeated fasting glucose determination. Nonetheless, preferential use by clinicians of repeated glucose determination was found to contribute to underdiagnosis. Hemoglobin A1c could not reliably predict impaired or diabetic glucose tolerance. We conclude that maximal value of screening for diabetes mellitus is achieved only by obligatory confirmatory oral glucose tolerance testing.