Six RCTs were identified (3,501 participants, range 127 to 1,220). Most were phase III trials and one was a late phase II trial. All the studies were double-blind RCTs; one was a double-dummy parallel group RCT. The authors concluded that all the studies were of adequate quality as each had accounted for the sources of error that were considered important. Follow-up ranged from 20 to 24 weeks. Loss to follow-up ranged from 0.1% to 31.5%.
Combined therapy with tocilizumab and methotrexate: Risk of at least one adverse event was significantly higher for high-dose tocilizumab than for controls (OR 1.53, 95% CI 1.26 to 1.86; four studies) but not for low-dose versus controls (three studies, Ι²=30%) and high-does versus low-dose tocilizumab (three studies, Ι²=35%). For at least one serious adverse event, there was no significant difference in risk for high-dose versus low-dose tocilizumab (three studies, Ι²=46%) and for low-dose versus controls (three studies). Results for high-dose versus controls (four studies) was not reported due to high heterogeneity (Ι²=59%).
For at least one infection, there was a significant increase in risk for high-dose tocilizumab versus controls (OR 1.30, 95% CI 1.07 to 1.58; three studies). There was no significant increase in risk for low-dose versus controls (two studies) and high-dose versus low-dose tocilizumab (two studies). There was no significant increase in risk of at least one serious infection for high-dose versus controls (four studies), low-dose versus controls (two studies) and high-does versus low-dose tocilizumab (three studies).
Monotherapy with 8mg/kg tocilizumab (high dose): There was no significant increase in risk for 8mg/kg tocilizumab versus controls for at least one serious adverse event (three studies), at least one infection (two studies) and for at least one serious infection (two studies). No results were reported for high-dose monotherapy versus controls for at least one adverse event (three studies) due to high heterogeneity (Ι²=64%).
The main adverse events reported were nasopharyngitis, respiratory tract disorder, skin and soft tissue pathology, and gastrointestinal side effects. The main infections reported were skin, subcutaneous and respiratory tract infections. Except where stated otherwise, heterogeneity was very low (Ι²<1%). Results of the sensitivity analysis were reported. The authors considered that the funnel plots showed no evidence of publication bias.