Interpretation of the glucose tolerance test

The oral glucose tolerance test: an objective method of interpretation

Oral glucose tolerance test results from a positively-skewed, unimodal distribution were analyzed in such a way as to uncover a natural division within the set of results for each time point. The division in results was obtained from a break in the curve formed when plasma glucose concentration was plotted vs the percentile of the population described by those concentrations. The percentile at which the break occurs separated normal from abnormal glucose concentrations objectively. Previously natural divisions between normal and abnormal glucose concentrations had been found only in bimodal distributions from atypical populations such as the Pima Indians and Nauruans of Micronesia. The glucose concentrations at which separations in our unimodal distribution occur compare well with the available data from the atypical populations. According to one measure of reliability, the 3-h time point was more effective than the more commonly utilized 2-h point in distinguishing normal from abnormal plasma glucose concentration.

Imprecision of new criteria for the oral glucose tolerance test

Subjects selected from a centre specializing in diabetes detection have been classified using nine different methods of diagnosis including the recent criteria of the National Diabetes Data Group and the World Health Organization. The applicability of these two new criteria have been evaluated and compared with seven other previous criteria. The agreements and discrepancies between these criteria have been assessed. Application of the new criteria result in a major redistribution of subjects with abnormal glucose tolerance into a rare category of diabetes mellitus and a large category of 'impaired glucose tolerance'. An important percentage of our population (33%) is excluded from the three clinical classes circumscribed by the criteria of the National Diabetes Data Group. Among these subjects, two-thirds correspond to the subjects named 'non-diagnostic' by these authors and one-third are not classifiable. The need to allow for all possible oral glucose tolerance test responses is emphasized.