Reproducibility of oral glucose tolerance data in normal and mildly diabetic subjects

SUMMARY Plasma glucose, plasma insulin and blood pyruvate levels were measured during two identical oral glucose tolerance tests carried out a few days apart on fifty-two normal or mildly diabetic subjects. No systematic differences could be detected between the first and second test. Several different functions derived from the tests were examined both for reproducibility and sensitivity in detecting departures from normal: a ‘clinical value index’ as calculated for each of these functions taking both reproducibility and sensitivity into account. This index was higher for the area under the glucose tolerance curve than for the glucose value at any single selected point during the test or for the ‘H’ index. The coefficient of variation for the glucose area was found to increase with increasing impairment of glucose tolerance, and confidence limits for this function were defined, enabling a significance of a single abnormal test to be assessed. Day-to-day variations in glucose tolerance were found not to be the result of fluctuations in the insulin response, nor to be related to the fasting pyruvate level before the test. It seems most likely that these variations result from fluctuations in glucose absorption and distribution.