Glycemic Index 3

The strength of any measure, such as the glycemic index (GI), is the accuracy and precision in which the values reflect the real world. Call up the GI of foods on Google and you’ll get a host of websites with a variety of values as if those values were written on stone tablets and delivered from the mountain top. So, where do most of those internet and diet book values come from? I’ll get to that.

Focusing for a second on the GI of foods presented in the last blog, a reader asked,

“Why the difference between the GI for Coca Cola in Australia and Atlanta?”

The answer is one of the keys to understanding why I contend the GI--as a clinical tool--is useless. I’ll soften that a bit: it’s a very crude tool at best, and may be useless. There are a number of reasons for that contention which I’ll touch on. However, criticism of the GI hasn’t stopped research interests from using it.

Case in point, a recent study published in the NEJM titled, Low-carbohydrate-diet score and the risk of coronary heart disease in women, the authors contended low carbohydrate diets (similar to Atkins and South Beach) were not associated with an increased risk of heart disease in a group of 88,000 nurses (the Nurses Health Study). Based upon a food frequency questionnaire, individual carbohydrate intake was coded for all study participants, and based upon the GI values of the foods reported from that questionnaire (Harvard has their own database for GI values), the Glycemic Load (that’s the GI multiplied by the amount of a particular food eaten) was also reported. The absolute carbohydrate intake and the calculated Glycemic Load was used to categorize women into both carbohydrate intake groupings, low to high, and Glycemic Load values. The authors found no relationship between absolute carbohydrate intake and heart disease, but that those with higher glycemic loads were found to be more likely to develop heart disease.

The reference I listed in yesterdays blog: Kaye Foster-Powell, et al. International table of glycemic index and glycemic load values is a compilation of numerous studies on a variety of foods which gives a crude database for the GI values of a number of foods. Pulling that reference and looking at something simple like rice, reveals that the values are all over the board depending upon the kind of rice, the method of testing, the country of origin, the study subjects used to arrive at the value—and potentially, the method of cooking the rice. As an example, Kellogg’s Corn Flakes is reported as follows:

Kellogg’s Corn Flakes ( USA) Study subjects: Diabetics GI = 92
Kellogg’s Corn Flakes (Canada) Study subjects: Diabetics GI = 86
Kellogg’s Corn Flakes (Australia) Study subjects: Healthy GI = 77
Kellogg’s Corn Flakes (Auckland) Study Subjects: Healthy GI = 72

Therefore, conducting a clinical research study, which of those values do you use to calculate the Glycemic Load? Usually the values are averaged to come up with some middle-of-the-road average value. But what does that do to accuracy? In addition to the database issue, and most problematic, foods are not eaten alone (as are the test foods listed). Most foods are part of what we call a “mixed meal” where a number of different foods are eaten simultaneously. Are the GI values valid for mixed meals? The answer is an unequivocal no. Eat Kellogg’s Corn Flakes from anywhere, put milk and part of a banana on it with some table sugar and all bets are off. The milk fat slows gastric emptying, thus changing the GI for corn flakes, the banana has fiber which again changes the value. And not to mention, what if you eat in a hurry, will that change the GI of the food? The answer is, it most likely will.

The bottom line is trust. Visit any old website and read an essay on the GI with a table of typical values. Find out where the author(s) derived the numbers. If they were from the International Table of Glycemic Index and Glycemic Load Values, trust them even less. Most values are averages from either healthy or diabetic subjects, some with glucose as a reference and some with white bread as a reference. Some evaluated test subjects for two hours and some for three hours. Conclusions in science drawn from such scattered and inhomogeneous sources build a foundation for junk science. The answer in the context of nutritional and food science, and in my humble opinion, is to remain skeptical.