A validation study of the glycemic index

In this study we examined the influence of selected variables on the variance in GI values obtained for bread and cooked barley. Twenty normal-weight men completed seven test sessions. During each session they consumed a food and provided blood samples obtained by finger prick. On three occasions the test food was a glucose drink. During the other four test sessions, the food was cooked barley, white bread, protein bread and peanut bread. The results showed that the GI of white bread was 69.5 and the GI of barley was 34.7 indicating that the GI of barley was different than the published value of 25 and the GI of white bread was not different than the published value of 70. The GI of barley was not affected by choice of standard (glucose versus white bread). However, contrary to previous studies, the two GI values were not correlated and could not be inter-converted. The results are showed considerable bias in the estimates with CVs of 37 and 53% for the GI estimates for barley and white bread respectively using the mean of three glucose test as the standard. Use of the median and/or removed of outliers had little effect on the estimates and the variance. The results showed no significance effect of ethnic group or interaction of food and ethnic group for the GI values. The results indicate a great degree of bias in the GI estimates obtained using standardized methods. This suggests that the GI is likely to have limited utility in explaining the relationship between postprandial blood glucose and risk for chronic diseases. (Abstract shortened by UMI.)...