Methodology for Adding Glycemic Index to the National Health and Nutrition Examination Survey Nutrient Database
ABSTRACT Generating valid estimates of dietary glycemic index (GI) and glycemic load (GL) has been a challenge in nutritional epidemiology. The methodologic issues may have contributed to the wide variation of GI/GL associations with health outcomes observed in existing literature. We describe a standardized methodology for assigning GI values to items in the National Health and Nutrition Examination Survey (NHANES) nutrient database using the new International Tables to develop research-driven, systematic procedures and strategies to estimate dietary GI/GL exposures of a nationally representative population sample. Nutrient databases for NHANES 2003-2006 contain information on 3,155 unique foods derived from the US Department of Agriculture National Nutrient Database for Standard Reference versions 18 and 20. Assignment of GI values were made to a subset of 2,078 carbohydrate-containing foods using systematic food item matching procedures applied to 2008 international GI tables and online data sources. Matching protocols indicated that 45.4% of foods had identical matches with existing data sources, 31.9% had similar matches, 2.5% derived GI values calculated with the formula for combination foods, 13.6% were assigned a default GI value based on low carbohydrate content, and 6.7% of GI values were based on data extrapolation. Most GI values were derived from international sources; 36.1% were from North American product information. To confirm data assignments, dietary GI and GL intakes of the NHANES 2003-2006 adult participants were estimated from two 24-hour recalls and compared with published studies. Among the 3,689 men and 4,112 women studied, mean dietary GI was 56.2 (men 56.9, women 55.5), mean dietary GL was 138.1 (men 162.1, women 116.4); the distribution of dietary GI was approximately normal. Estimates of population GI and GL compare favorably with other published literature. This methodology of adding GI values to an existing population nutrient database utilized systematic matching protocols and the latest comprehensive data sources on food composition. The database can be applied in clinical and survey research settings where there is interest in estimating individual and population dietary exposures and relating them to health outcomes.
- Journal of the American Academy of Nutrition and Dietetics 06/2013; 113(6):767-768. DOI:10.1016/j.jand.2013.03.021 · 2.44 Impact Factor
Article: Authors' Response.Journal of the American Academy of Nutrition and Dietetics 06/2013; 113(6):768-769. DOI:10.1016/j.jand.2013.03.022 · 2.44 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Research studies have suggested that chronic consumption of high glycemic index foods may lead to chronically high oxidative stress. This is important because oxidative stress is suspected to be an early event in the etiology of many disease processes. We hypothesized that dietary glycemic index and glycemic load were positively associated with oxidative stress assessed by plasma F2-isoprostanes in healthy, premenopausal women (BMI =24.7 ± 4.8 kg/m2 and age 25.3 ± 3.5 years, mean ± SD). We measured plasma F2-isoprostanes in 306 healthy premenopausal women at the baseline visit for the Women In Steady Exercise Research (WISER) study, using gas chromatography-mass spectrometry (GC-MS). Dietary glycemic index and load were calculated from the National Cancer Institute Diet History Questionnaire, and participants were divided into quartiles of dietary glycemic index and of glycemic load. Plasma F2-isoprostanes were compared across quartile groups of dietary glycemic index and glycemic load using linear regression models. Plasma F2-isoprostanes (pg/mL) increased with quartile of glycemic load (test for linear trend, p = .033), and also increased with quartile of glycemic index in participants with BMI ≥25 (p = .035) but not in those with BMI <25 (p = .924). After adjustment for BMI, alcohol consumption and total energy intake, both these positive trends remained marginally significant (p = .123 for quartiles of glycemic index and p = .065 for quartiles of glycemic load).Nutrition Research 10/2014; 35(1). DOI:10.1016/j.nutres.2014.10.005 · 2.59 Impact Factor