To use individual patient data from rheumatoid arthritis (RA) clinical trials to identify factors that affect the response to treatment as defined by the American College of Rheumatology (ACR) criteria for improvement (the “ACR response”).

Primary trial data from 14 diverse, randomized, controlled trials of second‐line drugs or devices in RA were analyzed. The trials included 11 methotrexate (MTX) trials (5 placebo controlled and 6 comparative, of which 2 were unpublished), 1 combination trial of cyclosporine plus MTX, 1 induction trial of a combination treatment in early RA (the COBRA trial), and 1 placebo‐controlled trial of a new device (Prosorba). Both patient factors and disease activity measures (primarily, items from the ACR core criteria set) were available.

A total of 1,435 patients (549 in placebo‐controlled trials, 886 in comparative trials) were studied. In both active treatment and placebo groups, disease duration had a strong effect on the likelihood of patient response (e.g., with any active treatment, the response rate was 53% for patients with ≤1 year of disease, 43% for 1–2 years' disease duration, 44% for 2–5 years, 38% for 5–10 years, and 35% for >10 years; P = 0.001). Decreasing response with greater disease duration was seen during treatment with most of the individual active drugs, as well as with placebo. Other factors decreasing the rate of response to treatment included any prior use of a disease‐modifying antirheumatic drug (DMARD), higher disease functional class (according to the Steinbrocker criteria), low disease activity (according to patient's global assessment), and female sex. Each ACR core set variable exhibited a diminished response to treatment in patients with long‐standing disease. The difference between active treatment and placebo response rates was not affected by disease duration nor by other factors associated with the ACR response.

RA patients with longer disease duration do not respond as well to treatment compared with patients with early disease, and female sex, prior DMARD use, disease functional class, and disease activity also have effects on the likelihood of patient response to treatment. This has implications for trial interpretation and for the clinical expectations of RA patients.