Development of a glycemic index database for dietary assessment

To address growing research interests and needs, a glycemic index (GI) database was developed for the more than 18,000 foods in the University of Minnesota's dietary data collection software, the Nutrition Data System for Research (NDSR). The primary source of data was current scientific literature with GI selected whenever possible from studies that used the following methodology: North American foods, healthy subjects, and a 2-h glucose response. Two GI numbers were included for all foods, one based on glucose as a reference, and the other, a white bread reference. Assigning indices within a large database also required imputation, because there are relatively few data available from the literature. For database foods that were not a match to foods in the literature, GI was either estimated from similar foods or calculated from available carbohydrate amounts and the GI of ingredients within the food. To evaluate the calculation procedure, GIs were calculated and then compared to known literature values for 102 multi-ingredient foods. A wide range of foods had comparable GIs, while some sweetened dairy products and unsweetened breakfast cereals showed larger GI differences. The GI database provides researchers with a tool to identify low- and high-GI foods and to investigate whether GI or glycemic load (GL) in the diet will influence disease risk factors.