Data for network meta-analyses inconsistency and SUCRA and mean rank are reported in the appendix (pp 579–85). Empirical heterogeneity variance for continuous outcomes for drug versus placebo comparisons was 0·05 (50% percentile) and 0·24 (75% percentile); for binary outcomes it was 0·12 (50% percentile) and 0·34 (75% percentile). Funnel plots are shown in the appendix (pp 630–32). We retained only a few studies—all in adults—with reported outcomes closest to 26 weeks or 52 weeks (appendix pp 586–88); therefore results for outcomes at these timepoints were deemed not informative.
