P035 Patient-reported outcome measures for rheumatoid arthritis disease activity: using Rasch measurement theory to achieve more meaningful measurement

Background/Aims Disease activity (DA) monitoring is a standard of care in rheumatoid arthritis (RA), and there is demand for achieving this through patient-reported outcome measures (PROMs). A systematic review of PROMs for RA DA following COSMIN guidelines demonstrated a lack of sufficient evidence for content validity for the 10 existing PROMS, thus none could be recommended for use. The aim of this study was to use Rasch measurement theory (RMT) to develop a valid item pool for measurement of DA in RA. Methods Paper questionnaires were sent to people aged 18 or over with RA from four South Wales University Health Boards between September 2020 and November 2021. The questionnaire included 268 individual RA DA items extracted from the 10 PROMs identified by the systematic review, another four PROMs and a non-measurement group of items. Further items suggested by patient feedback were incorporated, including a Pain Activity Scale, discomfort when walking, standing, and exercising, plus fear of falling when walking. Items were grouped into domains established by OMERACT: tenderness and swelling, patient global, pain, fatigue, physical functioning; and additionally stiffness, swelling, discomfort/fear and mood. Exploratory factor analyses (EFA) were used to indicate which items from which domains loaded onto factors together. For each separate domain, psychometric properties were assessed by RMT analyses. This provided results on targeting, model fit, internal consistency, local dependency, unidimensionality and item threshold ordering. Results A test dataset of n = 398 and a validation dataset of n = 293 were available. EFA of the test dataset showed that 30 items from across the tenderness and swelling, patient global, pain, fatigue, physical functioning and stiffness domains loaded together. RMT analyses of these items in the test dataset indicated that the patient global domain comprised two distinct domains, which followed a clear content difference of general health and disease activity items. In assessing the best items to measure the now 7 domains, 12 items were discarded. These 7 domains were confirmed in the validation data. Subtest analyses indicated that patient global general health and fatigue did not measure the same overall construct as the other 5 domains, and instead worked together to measure a separate construct. This structure was confirmed using bi-factor model confirmatory factor analyses in the validation data. Conclusion Patient global items relating to general health and disease activity were two separate domains. RA DA can be measured using tenderness and swelling, patient global disease activity, pain, physical functioning and stiffness items, whilst fatigue and patient global general health items measure a health-related quality of life construct. The next steps are to undertake cognitive interviews to establish content validity and to develop a computer adaptive test based on anchored locations calculated using these data