Relative Responsiveness of Physician/Assessor-Derived and Patient-Derived Core Set Measures in Rheumatoid Arthritis Trials

Outcomes of rheumatoid arthritis (RA) clinical trials are assessed by means of composite indices such as the American College of Rheumatology (ACR) core data set1 and the Disease Activity Score (DAS)2. The ACR core set measures include tender joint count (TJC), swollen joint count (SJC), physician global assessment, patient-reported pain, patient-reported function, patient global assessment, and an inflammatory marker, either erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP). The Disease Activity Score (DAS) includes TJC, SJC, patient global assessment, and an inflammatory marker (ESR or CRP). Thus, these composite measures include physician/assessor-derived assessments, patient-derived assessments, and a laboratory measure. Of the core set measures, those that are patient-derived have been advocated as better predictors of longterm outcomes like disability and death3-7, and as a better tool to document the longterm course of RA in the clinic8. One study has suggested that single measures assessed by patients, such as severity of pain and global assessment, are at least as sensitive to change in trials as such physician/ assessor measures as joint counts9. However, composite measures are not the same as individual core set items, and the relative sensitivity to change of such composite measures may not parallel those of their specific building-block items. Composite measures combining patient-derived and physician/assessor-derived elements from the core set have rarely been compared for their sensitivity to change10,11. These reports, one limited to 3 trials testing anakinra and the other to one trial of methotrexate and leflunomide, suggested equivalence or superiority of the patient-derived measures compared to physician/assessor-derived measures. The implication of this finding would be to place a greater emphasis on composite patient-derived outcomes not only in clinical practice, but also in clinical trials. However, the relative performance of the composite patient-derived versus composite physician/assessor-derived core set measures has not been evaluated in a larger number of patients from trials, or in a comprehensive study that includes trials of a wide spectrum of agents. Nor have the implications for clinical trial design with respect to required sample size been evaluated with such a strategy. In a recent reevaluation of response definitions for RA, we assembled a large dataset of recently completed multicenter RA trials; this gave us the opportunity to compare the sensitivity to change of patient-derived versus physician/assessor-derived composite measures in RA. We investigated whether a composite of patient-derived (CPD) core set measures performed similarly to a composite of physician/assessor-derived (CMD) core set measures in its ability to distinguish efficacious RA therapies.