Meta-analysis of second-line antirheumatic drugs: sample size bias and uncertain benefit

Placebo controlled trials of methotrexate, auranofin, penicillamine, azathioprine, sulphasalazine, gold sodium thiomalate and chloroquines were subjected to meta-analysis. The difference between drugs and placebo in the erythrocyte sedimentation rate was 8.8 mm/hr [95% confidence interval (CI), 6.4-11.3]. In multiple linear regression analyses, with the physician's global evaluation and relative change in joint tenderness count as outcome variables, a substantial sample size bias was demonstrated. The effect decreased with increasing sample size. The risk of dropping out from any cause was larger on drug than on placebo (odds ratio, 1.17; CI, 0.99-1.38). No evidence of a worthwhile effect on radiological changes was found. Of the 3439 patients, 4 went into complete remission on drug. We conclude that the benefit of second-line drugs is uncertain.