Defining and Measuring Primary Dimensions of Self-reported Health: An IRT Approach to Item Banking and Computerized Adaptive Testing

• Activity Number: 169
• Type: Topic Contributed
• Date/Time: Tuesday, August 13, 2002 : 8:30 AM to 10:20 AM
• Sponsor: Section on Health Policy Statistics*
• Author(s): Chih-Hung Chang*+ and David Cella
• Affiliation(s): Evanston Northwestern Healthcare and Evanston Northwestern Healthcare
• Keywords: health status assessment ; item response theory ; rating scale model ; partial credit model ; item bank ; computerized adaptive testing

Abstract

OBJECTIVE: This study illustrates a method for understanding the extent to which a large pool of health status questions can be examined for their primary dimensionality. SAMPLE: We simultaneously administered five questionnaires to 1,714 people with cancer and HIV disease. METHODS: We used item response theory (IRT)-based approaches that combined rating scale model and partial credit model. ANALYSES: Item responses were co-calibrated onto one presumed underlying construct. To identify significant, primary sub-structures in the data, we then conducted a factor analysis of the data residuals to determine the dimensional pattern of any deviation from unidimensionality. RESULTS: Two definable primary dimensions were derived: Physical well-being and mental well-being. Item banks were then established and computerized adaptive testing platform developed. CONCLUSIONS: These results provide direction for further defining these dimensions, so they can be measured more precisely and consistently. The delivery of precise and ready-to-use health status information for both patients and clinicians during office visits allows for better-informed treatment planning in clinical practice.