Objectives.: To determine possible differences in serious adverse effects among the 10 currently approved biological and targeted synthetic DMARDs (b/ts-DMARDs) for RA.

Methods.: Systematic review in bibliographic databases, trial registries and websites of regulatory agencies identified randomized trials of approved b/ts-DMARDs for RA. Network meta-analyses using mixed-effects Poisson regression models were conducted to calculate rate ratios for serious adverse events (SAEs) and deaths between each of the 10 drugs and control (i.e. no b/ts-DMARD treatment), based on subjects experiencing an event in relation to person-years. Confidence in the estimates was assessed by applying the Grading of Recommendations Assessment, Development and Evaluation approach (GRADE).

Results.: A total of 117 trials (47 615 patients) were included. SAEs were more common with certolizumab compared with abatacept (rate ratio = 1.58, 95% CI: 1.18, 2.14), adalimumab (1.36, 95% CI: 1.02, 1.81), etanercept (1.60, 95% CI: 1.18, 2.17), golimumab (1.45, 95% CI: 1.00, 2.08), rituximab (1.63, 95% CI: 1.16, 2.30), tofacitinib (1.44, 95% CI: 1.03, 2.02) and control (1.45, 95% CI: 1.13, 1.87); and tocilizumab compared with abatacept (1.30, 95% CI: 1.03, 1.65), etanercept (1.31, 95% CI: 1.04, 1.67) and rituximab (1.34, 95% CI: 1.01, 1.78). No other comparisons were statistically significant. Accounting for study duration confirmed our findings for up to 6 months' treatment but not for longer-term treatment (6-24 months). No differences in mortality between b/ts-DMARDs and control were found. Based on the GRADE approach, confidence in the estimates was low due to lack of head-to-head comparison trials and imprecision in indirect estimates.

Conclusion.: Despite low confidence in the estimates, our analysis found potential differences in rates of SAEs. Our data suggest caution should be taken when deciding among available drugs.

Systematic review registration number.: PROSPERO CRD42014014842.