Natural history of glucose tolerance in relatives of diabetic patients: low prognostic value of the oral glucose tolerance test.

Abstract

Seven hundred and fifty-one initially nondiabetic relatives of diabetic persons, in whom oral glucose tolerance tests (OGTT) have been performed, were followed for over 10 yr. Of these, 11.6% have died and no information was available from 13.7%. All other subjects were either retested (N = 369) or answered by questionnaire whether they are being treated for diabetes (N = 32) or not (N = 155). For test evaluation the sum of the 1- and the 2-h values during OGTT was used. The diagnosis of “diabetes” after 10 yr was made if the sum exceeded 500 mg/dl, which in 88% of cases coincides with fasting blood glucose values above 120 mg/dl. Among the retested individuals the results of OGTT showed a high degree of variation in both directions, which seems to represent just random variance rather than improvement or deterioration of biologic significance. These random changes in glucose tolerance do not correlate with changes in body weight. Sixty-three individuals have developed overt diabetes within the 10 yr. To optimize the discriminating value of the initial tests, the criteria dividing impaired glucose tolerance (IGT) from “normal” have been varied by a computer program. Sensitivity and specificity of the diagnosis IGT with respect to later development of diabetes depend on the criteria used and are inversely related. At the point where sensitivity and specificity are equal, both amount to 45%, and their sum never exceeds 90%. Using the criteria for IGT and diabetes recently proposed by members of an expert committee at the NIH, which are mainly based on the 2-h value and the fasting value, a sensitivity of 66% and a specificity of 19% are found. Thus, although a higher tendency toward the development of overt diabetes can be observed in the group of subjects with IGT compared with normal subjects, the prognostic value of the OGTT is very low when applied to individuals.