A Glycemic Index Expert Responds to the Tufts Research


The take-home message from the Tufts study is that the GI value of white breadis 70. That's nothing new: The same value has been found in dozens of otherstudies around the world (1).

What does a GI value of 70 mean? It means that onany given day, an exchange of white bread will have approximately 70 percent ofthe glycemic impact of an exchange of glucose sugar.

The authors of the Tufts study erroneously concluded that the GI value of thevery same bread varied from 44 to 132. That's not correct. The bread didn'tvary: Its GI value remained 70. The GI is a property of the food; the person issimply the instrument of measurement. The Tufts researchers did not follow GItesting protocol, which is carefully designed to compensate for the variabilityinherent in using a person as a test instrument. As a result, all the Tuftsstudy measured was the natural variation in individual responses to food atdifferent times. Unfortunately, they mistook that for variation in the breaditself.

It's true that any one person's glycemic response varies from day to day andthat glycemic response also varies from person to person. Nevertheless, therelative glycemic differences between foods are maintained. A person withdiabetes can be confident that a high GI food (GI value of 70 or more) willproduce a significantly higher glycemic response than a low GI food (GI value of55 or less) 95 percent of the time. Just as the height of high tide and lowtide varies from day to day and place to place, we still know that high tidewill be higher than low tide on any one day at any one place. It's not theabsolute level, but the difference, that's important.

Any one person's glucose tolerance may vary from day to day by as much astwo-fold. (In people with diabetes, this variability is actually less than innormal subjects.) Why this occurs is not clear, but we can point to varyingbeta-cell responsiveness and insulin sensitivity, factors that are beyondanyone's control. The beta cells just work better on some days than on others.This variability among and within people must be managed carefully in order todetect true differences in the glycemic potential of the carbohydrates indifferent foods.

That's why GI testing has such a strict protocol. Ten subjects are used, each ofwhom is given the reference food (glucose) on three separate days. Each time,the overall fluctuation in their blood sugar is determined by testing theirblood glucose eight times over a period of two hours. The findings from thosethree days of testing are averaged to find each person's usual response to thereference food, glucose. Next, his or her glycemic response to the test food ismeasured once, using the same two-hour testing protocol. Then each person'sresponse to the test food is expressed as a percentage of their average responseto the reference food. Finally, the relative responses of all ten subjects tothe test food are averaged. This is the published GI value. The GI value ofbread (70) means that the overall fluctuation in blood glucose after eating anexchange of white bread will be about 70 per cent of the effect of pure glucose(GI value of 100).

When it's done properly, there's nothing crude about GI testing. By taking theaverage of ten subjects, each of whom has undergone this painstaking process, weare simultaneously compensating for both within-subject and between-subjectvariability. If we were to test white bread over and over again using thisprotocol, we'd get the same result: a GI of 70.

The Tufts researchers did not satisfy the strict GI testing protocol. Theyrepeatedly compared one test of the reference food (glucose) with one test ofthe test food (white bread). Inexplicably, they did this three times with threedifferent groups of people. In a small sub-group of subjects who participated inall three groups (that is, who received the reference food on three occasions),the Tufts authors were able to calculate the GI according to the standardprotocol. And lo and behold, they arrived at 70!

The Tufts study's within-subject variability was also heightened by the factthat it used venous sampling, not fingertip capillary sampling. Capillarytesting is associated with much less variability than venous testing (2). We insiston capillary testing in the Australian Standard for GI testing (which is underreview by the International Standards Organization).

The Tufts study uses the variability inherent in individual glycemic responsesto criticize the GI. Were that a legitimate criticism, then that samevariability could be used to denigrate carbohydrate counting as well. Can we besure that fifteen grams of carbohydrate in white bread will always give half theglycemic response of thirty grams? No, we can't. It will also vary for the samereason: day-to-day within-subject variability. Yet carb counting is considered acornerstone of good diabetes self-care.

Meta-analyses show that a diet based on low GI carbohydrate foods (compared to aconventional low fat diet) will reduce A1c's by an average of 0.6 of a point(3), about the same amount as a serious exercise program. Other meta-analysesshow that low GI diets improve blood lipids (4) and weight control (5). Longerstudies are still required, but long term prospective cohort studies indicatethat diets with a low GI are likely to reduce the risk of chronic disease (6).


  1. Foster-Powell K, Holt SH, Brand-Miller JC. International table of glycemic index and glycemic load values: 2002. Am J Clin Nutr 2002;76:5-56.
  2. Wolever T, Bjorck I, Brand-Miller J, et al. Determination of the glycaemic index of foods: interlaboratory study. Br J Nutr 2003;57:475-482.
  3. Brand-Miller J, Petocz P, Colagiuri S. Meta-analysis of low glycemic index diets in the management of diabetes. Diabetes Care 2003;26:3363.
  4. Opperman A, Venter C, Oosthuizen W, Thompson R, Voster H. Meta-analysis of the health effects of using the glycaemic index in meal-planning. Br J Nutr 2004;92:367-81.
  5. Thomas D, Elliott E, Baur L. Low glycaemic index or low glycaemic load diets for overweight and obesity. Cochrane Database of Systematic Reviews 2007;Issue 3:Art. No.: CD005105. DOI: 10.1002/14651858.CD005105.pub2.
  6. Barclay A, Petocz P, McMillan-Price J, et al. Glycemic index, glycemic load and chronic disease risk – a meta-analysis of observational studies. Am J Clin Nutr, in press. 2007.

Professor Jennie Brand-Miller holds a Personal Chair in Human Nutrition at theUniversity of Sydney. Her research focuses on all aspects of carbohydrates,including diet and diabetes, the glycemic index, and insulin resistance. She isChair of the Nutrition Committee of the Australian Academy of Science, pastpresident of the Nutrition Society of Australia, the Director of SydneyUniversity Glycemic Index Research Service (a GI testing service for the foodindustry), and Chair of the Board of Directors of a non-profit company, GlycemicIndex Limited, which administers a food symbol program for consumers incollaboration with Diabetes Australia and the Juvenile Diabetes ResearchFoundation. She has published over 200 journal articles and twenty books,including The New Glucose Revolution series, an international bestseller.

Click here to view/write comments
Diabetes Health Medical Disclaimer
The information on this site is not intended or implied to be a substitute for professional medical advice, diagnosis or treatment. All content, including text, graphics, images, and information, contained on or available through this website is for general information purposes only. Opinions expressed here are the opinions of writers, contributors, and commentators, and are not necessarily those of Diabetes Health. Never disregard professional medical advice or delay seeking medical treatment because of something you have read on or accessed through this website.