Seventy RCTs were included in the review (n=18,846; 112 comparisons). The five comparisons that were meta-analysed were similar in the frequency of reporting sequence generation. However, trials with biological agents reported allocation concealment less frequently.
Placebo was associated with a statistically significantly larger percentage of the annual radiographic progression rate than single disease-modifying antirheumatic drug (DMARD; MD -0.65, 95% CI -1.05 to -0.25; n=762 patients) and glucocorticoid treatment (MD -0.54, 95% CI -0.71 to -0.38; n=1,141 patients). Single DMARD was associated with a significantly larger percentage of the annual radiographic progression rate than combination DMARDs (MD -0.62, 95% CI -1.00 to -0.24; n=1,384 patients) and biological agent plus methotrexate (MD -0.61, 95% CI -0.72 to 0.51; n=4,965 patients). The difference in percentage of the annual radiographic progression rate between two DMARDs combined plus step-down glucocorticoids was not statistically significantly different from biological agents plus methotrexate. The analyses for single DMARD versus placebo (I2=93%) and combination DMARDs versus single DMARD (I2 not reported) were associated with substantial heterogeneity.
Subgroup analyses showed similar results.
The funnel plots were asymmetric for the comparison single DMARD versus placebo, suggesting that publication bias may be present.