Editorial How Important is Prediction

The most common approach to the diabetes diet is to provide exchanges based on the amount of carbohy-drate, fat, and protein present in foods. This type of exchange system allows total energy intake to be con-trolled. However, recent studies have demonstrated that different types of protein have different effects on blood glucose and insulin responses, and that different foods, classified by glycemic index (Gl) or relative glycemic area (RGA), produce widely different glycemic re-sponses (1-3). Therefore, it has been suggested that a more rational approach to the diabetes diet would take into account the biologic effects of specific foods and nutrients (3). This concept has generated vigorous debate about 7) the reproducibility of glycemic responses of foods in different individuals and different centers, 2) the ability of Gl to predict the glycemic responses of mixed meals, and 3) whether differences between foods are large enough to be clinically meaningful (2,4). The scope of debate has been enlarged by an article by Gannon et al. on p. 544 describing the acute metabolic responses to different doses of oral glucose in patients with dia-betes. It has been assumed that the glycemic response to a food is related linearly to the amount of carbohy-drate ingested. However, Gannon et al. conclude that the net glucose response area to the oral glucose dose is best described by a second- or third-order polynomia