Eight trials met the inclusion criteria (n=2,385; range 26 to 682). Follow-up ranged from eight weeks to three years. Trials scored 3, 4 or 5 on the Jadad scale.
Efficacy up to six months (four trials): With etanercept, more people achieved an ACR20 response (55% versus 19%, RR 2.94, 95% CI 2.27 to 3.81), ACR50 (26% versus 6%, RR 5.28, 95% CI 3.12 to 8.92) and ACR70 (7% versus 1%, RR 4.83, 95% CI 1.74 to 13.47). No significant heterogeneity was observed.
Efficacy at one year (five trials): With etanercept, more people achieved an ACR20 response (77% versus 67%, RR 1.14, 95% CI 1.07 to 1.23), ACR50 (59% versus 43%, RR 1.36, 95% CI 1.21 to 1.53) and ACR70 (34% versus 21%, RR 1.56, 95% CI 1.30 to 1.88). Statistical heterogeneity was observed for all these analyses (I2=59% to 66%).
Safety and withdrawals: There was no statistically significant increase in the number of serious adverse events (four trials), serious infections (three trials), malignancy (four trials), death (four trials) and withdrawal due to adverse event (eight trials). Withdrawal due to lack of efficacy was lower in the etanercept groups (5% versus 10%, RR 0.48, 95% CI 0.30 to 0.78).