Skip to main content
ARS Home » Northeast Area » Beltsville, Maryland (BHNRC) » Beltsville Human Nutrition Research Center » Nutrient Data Laboratory » Research » Publications at this Location » Publication #239545
Title: Evaluation of Data Quality

Author
item Holden, Joanne
item Bhagwat, Seema

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 4/15/2009
Publication Date: 6/21/2009
Citation: Holden, J.M., Bhagwat, S.A. 2009. Evaluation of data quality. Evaluation of Data Quality, June 13-21, 2009, Tucuman, Argentina.

Interpretive Summary:

Technical Abstract: Interest in assigning data quality indicators to food composition databases is expressed at a global level. The quality of analytical data needs to be determined as part of the data acquisition process. Data quality evaluation procedures developed by the USDA (Exler 1983) and recently refined and modified (Holden et al. 2002) have been used to develop USDA’s Special Interest Databases (e.g., Isoflavones, Flavonoids). In addition, analysts and database compilers have adapted specific facets of the USDA system to develop procedures for evaluating other datasets. (Wenzel-Menezes, 2003) The present USDA system regards all the five evaluation categories (i.e., sampling plan, sample handling, analytical quality control, analytical method, number of samples) as equally important and gives equal maximum points (i.e., 20) to each category. Further considerations regarding the distribution of rating points are required to emphasize the importance of analytical method and sampling plan categories. Analytical method information required for evaluation is nutrient specific. Modifications for the sampling plan category are being made to address diverse populations and localities and the size and distribution of population, particularly in the countries of smaller size. A new category for “food description” will be added to the existing five categories as suggested by Møller et al. (2007). A manual for rating each category, deriving quality indices (QIs), and assigning confidence codes (CCs) will be prepared and used to train compilers, analysts, and policy experts to increase understanding of basic concepts related to the evaluation of data quality with international collaboration. The process of applying more broadly the assessment of data quality to data in National Nutrient Databank (NDBS) is under consideration. A plan will be developed to enhance the NDBS system to incorporate all the necessary pieces of information to evaluate the data in the National Nutrient Database for Standard Reference (SR) to assign data quality indicators (CCs). Therefore, nutrients of public health priority will be identified for the evaluation. The provision of quantitative information on the quality of values for specific dietary compounds and food will assist the data user to determine the reliability of the estimates under consideration for other database applications.