Predicting the Glycaemic Index of Mixed Meals

Abstract

The Glycaemic Index (GI) provides a measure of the rise in blood glucose following consumption of a test food relative to a reference food. Various associations have been found between certain dietary GIs and a number of diseases. GI has also been used to assist people with diabetes to choose foods that help to stabilise their blood glucose levels. The overall GI of a whole meal or diet has been estimated using a summation model of the individual components. The GI of each component is weighted according to its available carbohydrate contribution. Meal GI = Σ GIn×AvailCHOn (g) AvailCHOMeal (g) The validity of this model has not been thoroughly tested. Two simple meals of bread and beans have been used to test the model, with one study finding good agreement, while the other found little agreement between predicted and observed GI. Therefore the aim of this study was to robustly assess how well the summation model predicted the GI of typical mixed meals. A secondary aim was to compare the GI of three meals containing meat, vegetables and sauce that differed in their major carbohydrate source.

Thirty healthy participants aged between 21-49 years old were recruited from the public via posters and e-mails throughout the University of Otago. Fifteen people from each sex including ten from three age brackets (18-30yr, 30-40yr, 40-50yr) were recruited. Four reference glucose beverages (two 50g and two 25g), seven test foods (potato, rice, pasta, kumara, peas, carrots, sauce) and three meals containing potato, rice or pasta plus the other vegetables, sauce and 50g pan-fried chicken were tested by all participants. Nutrient analysis to determine carbohydrate content of the seven test foods was performed. Capillary blood glucose was measured before eating and over two hours post-prandially. Incremental areas under the blood glucose curve were calculated, and a mean GI was obtained for all foods and meals. Meal GI was predicted by inserting the observed GI for each food into the summation model and this was compared to the observed GI for each meal.

Mean (95% CI) GI values for the foods were: potato 72 (62, 85), rice 48 (41, 62), spaghetti 56 (48, 66), kumara 84 (72, 98), peas 29 (25, 34), carrots 31 (27, 36), and sauce 35 (30, 41). Observed and predicted mean (95% CI) GI values for the potato, pasta and rice based meals were: 53 (46, 62) cf 63 (56, 70), 38 (33, 45) cf 51 (45, 56) and 38 (33, 44) cf 55 (49, 61) respectively. The predicted meal GIs were greater than the observed meal GIs in all three cases (p<0.001). The summation model overestimated GI by 19-45% when applied to mixed meals.

The present study provides reliable information regarding the ability of the summation model to predict a composite GI. Using measured and published GI values for foods resulted in significant and variable overestimation of measured meal GIs. Researchers using this model to predict meal or dietary GI should be aware of the limitations associated with the model.