HANDLING POST PROCESSING OF A LARGE, COMPLEX, BLAISE INSTRUMENT IN A CONTINUING NATIONAL SURVEY

Authors: Clemens, John; Steinfeldt, Lois

Submitted to: International Blaise Users Conference
Publication Type: Abstract Only
Publication Acceptance Date: 4/15/2004
Publication Date: 4/14/2004
Citation: Clemens, J., Steinfeldt, L. 2004. Handling post processing of a large, complex, Blaise instrument in a continuing national survey [abstract]. 9th International Blaise Users Conference Abstracts. Available: http://www.stacan.ca/english/conferences/blaise2004/abstracts.htm

Interpretive Summary:

Technical Abstract: The Automated Multiple Pass Method (AMPM) an instrument developed by the USDA Food Surveys Research Group, is used to collect 24-hour dietary recall data in the National Health and Nutrition Examination Survey (NHANES), a continuous nationwide survey of approximately 5,000 individuals per year. The AMPM is a very large and complex Blaise instrument containing more than 2,500 questions, 21,000 responses, and 140,000 defined data fields. Although nearly all these are needed to encompass the vast range of possible food details, typically only 400 - 800 fields are needed for any given intake record. The Post Interview Processing System (PIPS), a Visual Basic application utilizing the Blaise Application Programming Interface (API), was developed by FSRG to extract food details from the Blaise database, reorganize, sort, and export them in a format required for food coding and subsequent data analysis. The ongoing nature of the NHANES requires the PIPS system: (1) be quick and easy to use, in order to facilitate frequent extractions needed to continually provide food coders with data; (2) permit ongoing updating of the automated food coding and food detail reorganization paths; and (3) be independent to Blaise data model changes resulting from food question changes. Ease of operation is achieved through the PIPS system user-friendly Windows interface coupled with the use of configuration files that save custom processing settings and manage batch number assignment. Utilization of input maps to identify branches of food coding fields that should be skipped while PIPS is traversing the network tree of a Blaise data file provides speedy data extraction. PIPS uses a Microsoft Access database to store, reformat, reorganize, and automatically code the data. As a result, automatic food coding and reorganization paths can be maintained and updated simply by updating Access tables and does not require any reprogramming. By utilizing the Blaise API and taking advantage of structures in the dietary intake data model, the PIPS system is transparent to all but the most radical of changes to the Blaise instrument data model. Consequently, PIPS has been successfully run without programming changes on three versions of the dietary intake data model, involving hundreds of revisions, deletions, and additions to the food detail questions, response categories, edits, and skip patterns. In conclusion, PIPS meets the requirements for use in a large ongoing survey and has successfully been used to extract and process dietary data from the 2004 NHANES to date.