Twenty-seven RCTs (11,049 patients) were included in the review. The number of patients recieving biological DMARDs were: 847 for abatacept (three RCTs), 1,027 for adalimumab (four RCTs), 250 for anakinra (one RCT), 750 for certolizumab (three RCTs), 795 for etanercept (four RCTs), 277 for golimumab (three RCTs), 611 for infliximab (three RCTs), 473 for rituximab (three RCTs), and 1,473 for tocilizumab (four RCTs). In all trials were randomised, no important differences in baseline characteristics were observed, and patients, providers and outcome assessors were blinded.
All drugs except anakinra and golimumab demonstrated a significant advantage compared with control for ACR20 (American College of Rheumatology 20% response). All drugs except anakinra demonstrated a significant advantage over placebo for ACR50 and ACR70. Preliminary analyses indicated that swollen joint count and disease duration were statistically significant covariates, so the analyses were performed with these as covariates. The results suggested that the mean effectiveness of drugs compared with placebo decreased as swollen joint count increased, and the relative effectiveness of biological DMARDs improved for those with a longer duration since diagnosis. No other covariate was found to be statistically significant.
Between drug comparisons demonstrated a significant advantage for certolizumab compared with all other individual drugs for ACR20 and ACR50, and for etanercept versus anakinra and adalimumab for ACR50 and ACR70. A sensitivity analysis excluding one trial (that did not require treatment failure of previous DMARDs) did not change the outcomes of the meta-analyses.