Objective: To develop preliminary criteria for defining disease flare in patients with polyarticular-course juvenile rheumatoid arthritis (JRA).
Methods: Data from a randomized clinical trial of etanercept in JRA (51 patients) and the 6 core response variables (CRV) for JRA were used to derive flare definitions. The criterion standard of flare was treatment with placebo. Candidate flare definitions were assessed by receiver-operator characteristic (ROC) curve properties and other statistics for diagnostic tests.
Results: Of the possible flare definitions tested with acceptable statistical properties, the one that seemed to be the most useful was worsening in any 2/6 CRV by > or = 40% without improvement in more than 1 of the remaining CRV by > or = 30%. Two other superior flare definitions were (1) worsening in 3/6 CRV by > or = 30% and (2) any worsening of the Childhood Health Assessment Questionnaire, worsening of erythrocyte sedimentation rate by > or = 30% and worsening of the active joint count by > or = 10%.
Conclusions: CRV are useful for defining flare in JRA. Worsening in any 2/6 CRV by > or = 40% without concomitant improvement of more than one of the remaining CRV by > or = 30% appears to be the most suitable preliminary flare definition. Because the proposed flare criteria were derived from a small number of patients, it is essential to perform more definitive testing of this and several alternative flare definitions in larger patient populations.