Studies of the efficacy of hydroxychloroquine in rheumatoid arthritis have had methodological flaws and have failed to produce definitive results. The benefits and toxicity of hydroxychloroquine sulfate in 120 patients with rheumatoid arthritis of less than 2 years duration are assessed.

A 36-week randomized double-blind, placebo-controlled trial was conducted at two university centers and four community rheumatology private practices. Patients had to have had their disease for less than 2 years and to have never received a second-line drug. Patients were randomly assigned to receive hydroxychloroquine or an equivalent number of placebo tablets in a dose of up to 7 mg/kg per day (maximum 400 mg/day). The initial dose was half the maximum dose and, if after 2 weeks of treatment there had been no side effects, the full dose was prescribed. There were four a priori primary outcomes: (1) a joint index composed of the tender joint count, the swollen joint count, the grip strength, and the duration of morning stiffness; (2) a pain index composed of the pain dimension of the Arthritis Impact Measurement Scales (AIMS) and the visual analog pain scale of the Health Assessment Questionnaire (HAQ); (3) a physical function index composed of the HAQ, the physical disability dimension of the AIMS, and the McMaster-Toronto Arthritis Patient Performance Disability Questionnaire; (4) the psychological function subscale of the AIMS. Secondary outcomes included adverse events, patient and physician global assessments, hematocrit, erythrocyte sedimentation rate (ESR) and corticosteroid usage. An intent-to-treat analysis assessed improvement at 36 weeks by Student's t-test and average improvement over the course of the study by analysis of variance for repeated measures.

Of 148 eligible patients, 120 were randomized. The characteristics of those randomized to hydroxychloroquine compared to placebo were similar at the study onset. At 36 weeks and over the course of the study there was statistically significant improvement in the joint index (P = 0.004, P = 0.034, respectively), the pain index (P = 0.007, P = 0.001, respectively), and the physical function index (P = 0.020, P = 0.011, respectively) in the group receiving hydroxychloroquine compared to the placebo group. There was no improvement in psychological function for hydroxychloroquine compared to placebo (P = 0.837 at 36 weeks, P = 0.89 over the course of the study). Among the secondary outcomes there was significant improvement only in the patient's (P = 0.01) and the outcome assessor's (P = 0.03) assessment of change and a trend towards a fewer number of intra-articular corticosteroid injections (P = 0.05) in the hydroxychloroquine-treated group. There were no important differences in the side effects between hydroxychloroquine or placebo.

Over 36 weeks, hydroxychloroquine had a significant benefit on synovitis, pain, and physical disability of recent-onset rheumatoid arthritis, but did not benefit psychological function.