Selection of items for a computer-adaptive test to measure fatigue in patients with rheumatoid arthritis: a Delphi approach

Purpose Computer-adaptive tests (CATs) can measure precisely at individual level with few items selected from an item bank. Our aim was to select fatigue items to develop a CAT for rheumatoid arthritis (RA) and include expert opinions that are important for content validity of measurement instruments. Methods Items were included from existing fatigue questionnaires and generated from interview material. In a Delphi procedure, rheumatologists, nurses, and patients evaluated the initial pool of 294 items. Items were selected for the CAT development if rated as adequate by at least 80% of the participants (when 50% or less agreed, they were excluded). Remaining items were adjusted based on participants' comments and re-evaluated in the next round. The procedure stopped when all items were selected or rejected. Results A total of 10 rheumatologists, 20 nurses, and 15 rheumatoid arthritis patients participated. After the first round, 96 of 294 items were directly selected. Nine items were directly excluded, and remaining items were adjusted. In the second round, 124 items were presented for re-evaluation. Ultimately, 245 items were selected. Conclusion This study revealed a qualitatively evaluated item pool to be used for the item bank/CAT development. The Delphi procedure is a beneficial approach to select adequate items for measuring fatigue in RA.