Methodology

Equivalent glycemic load (EGL): a method for quantifying the glycemic responses elicited by low carbohydrate foods

Thomas MS Wolever¹ , Alison L Gibbs² , Matt Spolar³ , Elinor V Hitchner^3,4 and Colette Heimowitz³

¹Glycemic Index Laboratories, Inc., 36 Lombard Street, Suite 100, Toronto, Ontario, M5C 2X3, Canada

²Department of Statistics, University of Toronto, Toronto, Ontario, Canada

³Atkins Nutritionals, Inc., New York, NY, USA

⁴Cadbury Schweppes Science and Technology Center, Whippany, NJ, USA

author email corresponding author email

Nutrition & Metabolism 2006, 3:33doi:10.1186/1743-7075-3-33

Published:	24 August 2006

Abstract

Background

Glycemic load (GL) is used to quantify the glycemic impact of high-carbohydrate (CHO) foods, but cannot be used for low-CHO foods. Therefore, we evaluated the accuracy of equivalent-glycemic-load (EGL), a measure of the glycemic impact of low-CHO foods defined as the amount of CHO from white-bread (WB) with the same glycemic impact as one serving of food.

Methods

Several randomized, cross-over trials were performed by a contract research organization using overnight-fasted healthy subjects drawn from a pool of 63 recruited from the general population by newspaper advertisement. Incremental blood-glucose response area-under-the-curve (AUC) elicited by 0, 5, 10, 20, 35 and 50 g CHO portions of WB (WB-CHO) and 3, 5, 10 and 20 g glucose were measured. EGL values of the different doses of glucose and WB and 4 low-CHO foods were determined as: EGL = (F-B)/M, where F is AUC after food and B is y-intercept and M slope of the regression of AUC on grams WB-CHO. The dose-response curves of WB and glucose were used to derive an equation to estimate GL from EGL, and the resulting values compared to GL calculated from the glucose dose-response curve. The accuracy of EGL was assessed by comparing the GL (estimated from EGL) values of the 4 doses of oral-glucose with the amounts actually consumed.

Results

Over 0–50 g WB-CHO (n = 10), the dose-response curve was non-linear, but over the range 0–20 g the curve was indistinguishable from linear, with AUC after 0, 5, 10 and 20 g WB-CHO, 10 ± 1, 28 ± 2, 58 ± 5 and 100 ± 6 mmol × min/L, differing significantly from each other (n = 48). The difference between GL values estimated from EGL and those calculated from the dose-response curve was 0 g (95% confidence-interval, ± 0.5 g). The difference between the GL values of the 4 doses of glucose estimated from EGL, and the amounts of glucose actually consumed was 0.2 g (95% confidence-interval, ± 1 g).

Conclusion

EGL, a measure of the glycemic impact of low-carbohydrate foods, is valid across the range of 0–20 g CHO, accurate to within 1 g, and at least sensitive enough to detect a glycemic response equivalent to that produced by 3 g oral-glucose in 10 subjects.