Content area
Abstract
Objective: The USDA standard release databases use food weight to determine nutrient content; many dietary self-assessment methods and emerging image-assisted technologies utilize estimated food volumes to assess dietary intake. Data are needed to understand the comparability of assessment using volumetric vs. weight measures. This study evaluates the accuracy of macronutrient content for foods estimated by volume compared to weight.
Materials and Methods: Weights and volumes of 37 food portions from 6 groups were measured. Each trial was comprised of 10 replicates; additional quality control trials were conducted for 10% of randomly-chosen foods. Commonly consumed foods were selected to include variation in water content and shape. Foods were prepared and measured in an experimental nutrition laboratory. Nutrient information was extracted from the USDA SR 28 database for each food’s weight and volume; differences in weight and macronutrients were computed for each trial.
Results: Significant differences in weight determined via volume by the USDA SR 28 (USDA weight) and experimental weight were found in 76% of trials. For 24% of all foods, calories estimated by USDA weight were significantly more than estimates by experimental weight; for 46% of foods, calories by USDA weight were significantly less than estimates by experimental weight. Protein content estimates by USDA and experimental weight differed significantly from each other for 100% of dairy and 69% of protein foods. Carbohydrate estimates differed significantly for 65% of foods; the highest was white rice, where estimates by USDA weight were 8.7±0.48g/serving lower than experimental weight estimates. Fiber content estimates differed significantly for 75% of fruit and vegetables. Lipid content estimates were significantly different for fat-dense foods and were as large as 4.85g/serving.
Significance: Many researchers are reliant on nutrient databases to accurately determine the nutrient content of various foods. It is important to ensure the accuracy of these databases to estimate nutrient content based on food volumes. Funding disclosure This work was supported by P30 CA15704 and U01-CA135133.
