Integrating patient reported outcome measures and computerized adaptive test estimates on the same common metrics: an example from the assessment of activities in rheumatoid arthritis


DOĞANAY ERDOĞAN B., ELHAN A. H. , Kaskatı O. T. , Öztuna D., KÜÇÜKDEVECİ A. A. , KUTLAY Ş., ...More

International Journal of Rheumatic Diseases, vol.20, no.10, pp.1413-1425, 2017 (Journal Indexed in SCI) identifier identifier identifier

  • Publication Type: Article / Article
  • Volume: 20 Issue: 10
  • Publication Date: 2017
  • Doi Number: 10.1111/1756-185x.12571
  • Title of Journal : International Journal of Rheumatic Diseases
  • Page Numbers: pp.1413-1425
  • Keywords: computerized adaptive testing, item bank, Rasch analysis, rheumatoid arthritis, RASCH, DISABILITY, SCALES, MODEL

Abstract

© 2015 Asia Pacific League of Associations for Rheumatology and Wiley Publishing Asia Pty LtdAim: This study aimed to explore the potential of an inclusive and fully integrated measurement system for the Activities component of the International Classification of Functioning, Disability and Health (ICF), incorporating four classical scales, including the Health Assessment Questionnaire (HAQ), and a Computerized Adaptive Testing (CAT). Methods: Three hundred patients with rheumatoid arthritis (RA) answered relevant questions from four questionnaires. Rasch analysis was performed to create an item bank using this item pool. A further 100 RA patients were recruited for a CAT application. Both real and simulated CATs were applied and the agreement between these CAT-based scores and ‘paper-pencil’ scores was evaluated with intraclass correlation coefficient (ICC). Anchoring strategies were used to obtain a direct translation from the item bank common metric to the HAQ score. Results: Mean age of 300 patients was 52.3 ± 11.7 years; disease duration was 11.3 ± 8.0 years; 74.7% were women. After testing for the assumptions of Rasch analysis, a 28-item Activities item bank was created. The agreement between CAT-based scores and paper-pencil scores were high (ICC = 0.993). Using those HAQ items in the item bank as anchoring items, another Rasch analysis was performed with HAQ-8 scores as separate items together with anchoring items. Finally a conversion table of the item bank common metric to the HAQ scores was created. Conclusion: A fully integrated and inclusive health assessment system, illustrating the Activities component of the ICF, was built to assess RA patients. Raw score to metric conversions and vice versa were available, giving access to the metric by a simple look-up table.