Validation and assessment of minimally clinically important difference of the unadjusted Health Assessment Questionnaire in a Danish cohort: uncovering ordinal bias

• Published in: Scandinavian journal of rheumatology 2020-01, Vol.49 (1), p.1-7
• Main Authors: Ørnbjerg, LM, Christensen, KB, Tennant, A, Hetland, ML
• Format: Article
• Language: English
• Subjects: Aged; Arthritis, Rheumatoid - diagnosis; Arthritis, Rheumatoid - rehabilitation; Disability Evaluation; Female; Follow-Up Studies; Health Status; Humans; Male; Middle Aged; Outcome Assessment, Health Care - methods; Patient Reported Outcome Measures; Quality of Life; Severity of Illness Index; Surveys and Questionnaires - standards
• ISSN: 0300-9742; 1502-7732
• DOI: 10.1080/03009742.2019.1637932

Objectives: The Stanford Health Assessment Questionnaire Disability Index (HAQ-DI) is a widely used patient-reported outcome for functional disability in rheumatoid arthritis (RA). Minimal clinically important differences (MCIDs) in the HAQ-DI were previously calculated based on device-corrected ordinal HAQ-DI scores, leading to limited generalizability and validity for today's patients. Our objectives were to examine the internal construct validity of the unadjusted HAQ-DI and to determine an MCID in a cohort of Danish RA patients based on the transformed linear logit scale of the HAQ-DI. Method: The study included 362 RA patients registered in the DANBIO registry. The Rasch model was fitted to HAQ-DI data at baseline and after 3 months' follow-up. MCID was calculated as the median changes in the original HAQ-DI score and logit HAQ-DI score in those patients who had experienced minimal improvement (15-30 mm on a 0-100 mm Patient Global scale). Results: HAQ-DI data showed acceptable fit to the Rasch model at both time-points, and consistent item ranking across time indicated instrument invariance. Sixty-one patients (16.8%, ~1/6) had an improvement above the MCID on the logit scale but improvement below the MCID on the original scale, while the opposite was not the case for any patients. Conclusions: The Danish unadjusted version of the HAQ-DI showed acceptable internal construct validity. Application of the logit MCID classified an additional one in six patients as having achieved an MCID compared to the MCID calculated on the ordinal scale, which may have potential implications for the powering of future studies.