Table 1.
Meta-analysis | Cochrane review Neilson (2012) | European MA Becker et al. (2012) | American MA Potti & Berghella (2012) | Stockholm MA Salmelin et al. (2013) | IPD MA Schuit et al. (2013) |
---|---|---|---|---|---|
Type of meta-analysis | Aggregate | Aggregate | Aggregate | Aggregate | Individual participant data |
Data collection | 5 RCTs, principal authors of Swedish and French RCTs contacted for missing data | 5 RCTs, principal authors of Swedish, Finnish, French and Dutch RCTs among authors to European and IPD MAs | 5 RCTs, only data used in original articles are used | 4 RCTs, only data used in original articles are used | 4 RCTs, IPD provided by principal investigators: all randomized cases from Swedish and French RCTs included, from Finnish RCT 11 exclusions, from Dutch RCT 14 exclusions |
Measures of treatment effect | RR with 95% CI (fixed-/random-effect model as appropriate); fixed-effect when no heterogeneity | RR with 95% CI (fixed-/random-effect model as appropriate); fixed effect when no heterogeneity | RR with 95% CI (fixed-/random-effect models as appropriate); fixed-effect when no heterogeneity | RR with 95%CI; consistently random-effect, although tests for heterogeneity were performed | RR with an RR <1 indicating treatment benefit; random-effect log-binomial model; imputation of missing data |
Assessment of heterogeneity (figures indicate substantial heterogeneity) | Tau2 (>0), I2 (>30%), chi-square for heterogeneity (p < 0.10) | Tau2 (>0), I2 (>50%) | I2 (40–84%, if ≥85% no MA), chi-square for heterogeneity (p < 0.10) | Performed, but random-effects MA consistently used | I 2 (0% indicating no heterogeneity, 25% low, 50% moderate, 75% high) |
IPD, individual participant data; RR, risk ratio.