We searched for randomized controlled trials (RCTs) in the Cochrane Central Register of Controlled Trials (CENTRAL) (via The Cochrane Library Issue 6, June 2015), MEDLINE (via OVID 1946 to June 2015), and EMBASE (via OVID 1947 to June 2015). Data extraction, risk of bias and GRADE assessments were done in duplicate. We calculated both direct estimates using standard meta-analysis and used Bayesian mixed treatment comparisons approach for NMA estimates to calculate odds ratios (OR) and 95% credible intervals (CrI). We converted OR to risk ratios (RR) which are reported in the abstract for the ease of interpretation.
MAIN RESULTS: This update included 73 new RCTs for a total of 90 RCTs; 79 RCTs with 32,874 participants provided usable data. Few trials were at high risk of bias for blinding of assessors/participants (13% to 21%), selective reporting (4%) or major baseline imbalance (8%); a large number had unclear risk of bias for random sequence generation (68%) or allocation concealment (74%).Based on direct evidence of moderate quality (downgraded for inconsistency), biologic+MTX/
DMARD was associated with a statistically significant and clinically meaningful improvement in ACR50 versus comparator (RR 2.71 (95% confidence interval (CI) 2.36 to 3.10); absolute benefit 24% more patients (95% CI 19% to 29%), number needed to treat for an additional beneficial outcome (NNTB) = 5 (4 to 6). NMA estimates for ACR50 in
tumor necrosis factor (TNF) biologic+MTX/
DMARD (RR 3.23 (95% credible interval (Crl) 2.75 to 3.79), non-TNF biologic+MTX/
DMARD (RR 2.99; 95% Crl 2.36 to 3.74), and
anakinra + MTX/
DMARD (RR 2.37 (95% Crl 1.00 to 4.70) were similar to the direct estimates.Based on direct evidence of moderate quality (downgraded for inconsistency), biologic+MTX/
DMARD was associated with a clinically and statistically important improvement in function measured by the Health Assessment Questionnaire (0 to 3 scale, higher = worse function) with a mean difference (MD) based on direct evidence of -0.25 (95% CI -0.28 to -0.22); absolute benefit of -8.3% (95% CI -9.3% to -7.3%), NNTB = 3 (95% CI 2 to 4). NMA estimates for TNF biologic+MTX/
DMARD (absolute benefit, -10.3% (95% Crl -14% to -6.7%) and non-TNF biologic+MTX/
DMARD (absolute benefit, -7.3% (95% Crl -13.6% to -0.67%) were similar to respective direct estimates.Based on direct evidence of moderate quality (downgraded for inconsistency), biologic+MTX/
DMARD was associated with clinically and statistically significantly greater proportion of participants achieving remission in RA (defined by disease activity score
DAS < 1.6 or DAS28 < 2.6) versus comparator (RR 2.81 (95% CI, 2.23 to 3.53); absolute benefit 18% more patients (95% CI 12% to 25%), NNTB = 6 (4 to 9)). NMA estimates for TNF biologic+MTX/
DMARD (absolute improvement 17% (95% Crl 11% to 23%)) and non-TNF biologic+MTX/
DMARD (absolute improvement 19% (95% Crl 12% to 28%) were similar to respective direct estimates.Based on direct evidence of moderate quality (downgraded for inconsistency), radiographic progression (scale 0 to 448) was statistically significantly reduced in those on biologics + MTX/DMARDs versus comparator, MD -2.61 (95% CI -4.08 to -1.14). The absolute reduction was small, -0.58% (95% CI -0.91% to -0.25%) and we are unsure of the clinical relevance of this reduction. NMA estimates of TNF biologic+MTX/
DMARD (absolute reduction -0.67% (95% Crl -1.4% to -0.12%) and non-TNF biologic+MTX/
DMARD (absolute reduction, -0.68% (95% Crl -2.36% to 0.92%)) were similar to respective direct estimates.Based on direct evidence of moderate quality (downgraded for imprecision), results for withdrawals due to adverse events were inconclusive, with wide confidence intervals encompassing the null effect and evidence of an important increase in withdrawals, RR 1.11 (95% CI 0.96 to 1.30). The NMA estimates of TNF biologic+MTX/
DMARD (RR 1.24 (95% Crl 0.99 to 1.57)) and non-TNF biologic+MTX/
DMARD (RR 1.20 (95% Crl 0.87 to 1.67)) were similarly inconclusive and downgraded to low for both imprecision and indirectness.Based on direct evidence of high quality, biologic+MTX/
DMARD was associated with clinically significantly increased risk (statistically borderline significant) of serious adverse events on biologic+MTX/
DMARD (Peto OR [can be interpreted as RR due to low event rate] 1.12 (95% CI 0.99 to 1.27); absolute risk 1% (0% to 2%), As well, the NMA estimate for TNF biologic+MTX/
DMARD (Peto OR 1.20 (95% Crl 1.01 to 1.43)) showed moderate quality evidence of an increase in the risk of serious adverse events. The other two NMA estimates were downgraded to low quality due to imprecision and indirectness and had wide confidence intervals resulting in uncertainty around the estimates: non-TNF biologics + MTX/
DMARD: 1.07 (95% Crl 0.89 to 1.29) and
anakinra: RR 1.06 (95% Crl 0.65 to 1.75).Based on direct evidence of low quality (downgraded for serious imprecision), results were inconclusive for
cancer (Peto OR 1.07 (95% CI 0.68 to 1.68) for all biologic+MTX/
DMARD combinations. The NMA estimates of TNF biologic+MTX/
DMARD (Peto OR 1.21 (95% Crl 0.63 to 2.38) and non-TNF biologic+MTX/
DMARD (Peto OR 0.99 (95% Crl 0.58 to 1.78)) were similarly inconclusive and downgraded to low quality for both imprecision and indirectness.Main results text shows the results for
tofacitinib and differences between medications.
AUTHORS' CONCLUSIONS: