Using Rasch analysis to compare the psychometric properties of the Short Form 36 physical function score and the Health Assessment Questionnaire disability index in patients with psoriatic arthritis and rheumatoid arthritis
Article first published online: 25 MAY 2007
Copyright © 2007 by the American College of Rheumatology
Arthritis Care & Research
Volume 57, Issue 5, pages 723–729, 15 June 2007
How to Cite
Taylor, W. J. and McPherson, K. M. (2007), Using Rasch analysis to compare the psychometric properties of the Short Form 36 physical function score and the Health Assessment Questionnaire disability index in patients with psoriatic arthritis and rheumatoid arthritis. Arthritis & Rheumatism, 57: 723–729. doi: 10.1002/art.22770
- Issue published online: 25 MAY 2007
- Article first published online: 25 MAY 2007
- Manuscript Accepted: 16 OCT 2006
- Manuscript Received: 21 JUN 2006
- Wellington Medical Research Foundation
- Otago University Research grant
- Psoriatic arthritis;
- Item-response theory;
- Health status measures
Item-response theory is increasingly used in the development of robust measurement tools. The extent to which the Health Assessment Questionnaire (HAQ) disability index (DI) and Short Form 36 (SF-36) physical functioning subscale (PF) fit a Rasch model in psoriatic arthritis (PsA) is uncertain. Our objective was to compare the psychometric properties of the HAQ DI and SF-36 PF in PsA and rheumatoid arthritis (RA) using Rasch analysis.
Patients with RA (n = 142) and PsA (n = 134) were identified from a disease register based at a regional rheumatology service that serves a population of ∼400,000 individuals. Responses to the HAQ DI and SF-36 PF were analyzed for item fit, differential item functioning (DIF), scale length (item separation), floor effects, and item difficulty by fitting the data to a Rasch model. The extent to which each instrument measured the same concept (disability) was also assessed in the PsA cohort using the Rasch model.
Item separation was much better for the SF-36 PF than the HAQ DI in PsA (9.12 logits versus 2.06 logits). There was evidence of marked DIF for the HAQ DI items activities, grip, and rising and relatively minor DIF for 4 items of the SF-36 PF. The distribution of SF-36 PF was better than HAQ DI in PsA, with floor effects of 3.1% versus 30.4%. Common person equating demonstrated that the 2 instruments measure the same construct in PsA.
The SF-36 PF has significant psychometric advantages over the HAQ DI in PsA.