Psychometric Evaluation of PROMIS: IRT Calibration, DIF, Scaling, and Validity

• Activity Number: 89
• Type: Invited
• Date/Time: Monday, August 3, 2009 : 8:30 AM to 10:20 AM
• Sponsor: Section on Health Policy Statistics
• Author(s): Wen-Hung Chen*+ and Ron D. Hays and Bryce B. Reeve
• Companies: United BioSource Corporation and University of California, Los Angeles and National Cancer Institute
• Keywords: item response theory ; health-related quality of life ; item banking ; patient-reported outcome ; differential item functioning

Abstract

The Patient-Reported Outcomes Measurement Information System (PROMIS) project is a unique effort to develop health-related quality of life (HRQOL) item banks. This paper presents an overview of the psychometric evaluation process used in PROMIS including evaluation of item response theory (IRT) assumptions, IRT item parameter calibration, assessing model fit, assessing differential item functioning (DIF), and standardizing the score metric. The evaluation of the validity of the PROMIS domain scores relative to existing health outcome measures (e.g., SF-36; CESD) is also reported. The existing health outcome measures were strongly correlated with PROMIS scores. In addition, PROMIS domain scores tended to provide greater information than existing legacy measures. The PROMIS item banks and domain scores are psychometrically sound and provide the basis for precise health outcomes assessment.