VALIDATION DATA

Cronbach's alpha for the complete 19-item questionnaire was 0.96.

Reproducibility was evaluated by a test-retest within patients declaring no change of state (either flare or no-flare) between the two measures.  103 patients responded twice within the predefined time period, 81 had only knee or hip osteoarthritis, and 66 among those were in the same state of flare (37 patients) or no flare (29 patients) at both times. The ICC [95%CI] was 0.85 [0.78-0.90] and the associated SEM was 10.2.

258 patients reporting a flare in the past 4 weeks were compared to 140 patients reporting no flare. The mean [95%CI] score was 48.7 [45.7-51.7] for patients reporting a flare in contrast to 16.9 [13.6-20.7] for those with no flare, with a difference of 31.8 (P < 0.0001) largely exceeding 2 SEM.

Twelve OA patients underwent unstructured interviews that helped create a guide for conducting semi-structured interviews. Therefore, items were generated from semi-structured interviews with 29 patients and 16 HCPs and one focus group with 10 patients.

CFA was used to establish the most appropriate factor structure of the questionnaire by assessing the model fit indices. The initial model with 33 items had poor fit indices (RMSEA  0.09; SRMR  0.05; CFI  0.85 and TLI  0.82). All items had loadings of >0.4 in their respective domains. Refinement of the model took place in 25 iterative steps following the procedure described in the method section, problematic items were either deleted or moved to another dimension. After content consideration and index changes for the CFA, 14 items were removed. In its final form, the Flare-OA questionnaire included 19 items, resulting in a significant improvement in the fit indices as compared with the initial model (RMSEA 0.06; SRMR 0.04; CFI 0.96 and TLI 0.94)