USDA’s Nutrient Databank System – a tool for handling data from diverse sources

Authors: Haytowitz, David; Lemar, Linda; Pehrsson, Pamela

Submitted to: Journal of Food Composition and Analysis
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/28/2008
Publication Date: 1/24/2009
Citation: Haytowitz, D.B., Lemar, L.E., Pehrsson, P.R. 2009. USDA’s Nutrient Databank System – a tool for handling data from diverse sources. Journal of Food Composition and Analysis. http://dx.doi.org/10.1016/j.jfca.2009.01.003.

Interpretive Summary: The Nutrient Databank System (NDBS) is USDA's repository for food composition data and is the source of publicly available datasets, such as the USDA National Nutrient Database for Standard Reference. The NDBS handles the processing of data from diverse sources, including USDA’s National Food and Nutrient Analysis Program, the food industry, scientific literature, and food labels. NDL designed the NDBS as a 3-tiered data management system. Full nutrient profiles are developed and any missing nutrient data is imputed according to scientific principles from similar foods. A formulation module, employing linear regression techniques, estimates formulations and full nutrient profiles, based on the nutrient content of package or recipe ingredients and target values derived from label information. A recipe module calculates nutrient profiles based on ingredients and their proportions, allowing for the application of yield and retention factors. The NDBS incorporates quality control checks at all levels. Data points and documentation are managed and maintained in one place providing an “audit trail” for each data point. The NDBS permits the annual release of reliable, current data for a comprehensive suite of components for a wide variety of foods on NDL's Web site: http://www.ars.usda.gov/nutrientdata. Nutrient data are provided for the "What We Eat in America" component of the National Food and Nutrition Examination Survey (NHANES) and other food surveys. Through these releases, NDL provides critical data for researchers, diet and health professionals, and consumers.

Technical Abstract: This manuscript describes key features of USDA’s Nutrient Databank System (NDBS), which allows processing of food composition data from diverse sources, including USDA’s National Food and Nutrient Analysis Program, the food industry, scientific literature, and food labels. NDL designed the NDBS as a 3-tiered data management system ("Initial", "Aggregate", and “Compile”) with modules to facilitate handling of data. Raw data and documentation (data source, sample description, sample handling, and analytical methods) are migrated into the Initial module. NDL scientists compare new data with old values and decide how to combine the Initial data in Aggregate. Data can be grouped and weighted by parameters such as study, source, and market share. Depending on the type of data, various statistical algorithms are used to generate statistics, such as mean, standard error, number of data points, and error bounds. In the Compile module, food names are finalized and common measures selected. Full nutrient profiles are developed and any missing components are imputed according to standardized scientific principles. A formulation module, employing linear regression techniques, estimates formulations and full nutrient profiles, based on the nutrient content of ingredients and target values derived from label information. A recipe module calculates nutrient profiles based on ingredients and their proportions, allowing for the application of food yield and nutrient retention factors. The NDBS automatically documents how each value was derived and incorporates quality control checks at all levels. Prior to release, the completed nutrient profiles are sent for review and, if approved, disseminated. The NDBS brings together a number of standalone applications into one integrated system allowing the management of ~7,500 food items for up to 140 nutrients/food components. Data points and documentation are managed and maintained in one place providing an "audit trail" for each data point. The NDBS contains modules to assign confidence codes using NDL's expert system. The NDBS permits the annual release of reliable, current data for a comprehensive suite of components for a wide variety of foods on NDL's Web site: http://www.ars.usda.gov/nutrientdata. Nutrient data are provided for the "What We Eat in America" component of the National Food and Nutrition Examination Survey (NHANES) and other food surveys. Through these releases, NDL provides critical data for researchers, diet and health professionals, and consumers.