2.5 | Data analysis
Meta-analysis was performed by Review Manager 5.3 (The
Cochrane Collaboration, Copenhagen,
Denmark) and Stata 14.0 (Stata Corporation, College Station, TX, USA).
Effect sizes computed for the analyses were presented as
risk ratios (RR) with 95% confidence intervals (CI). The RR was not
estimable when the events were zero in both groups. I2<50% was indicative of no heterogeneity, and the fixed effect
model was adopted. By contrast, I2 >50%
was indicative of heterogeneity, and the random effect model was applied
after sensitivity analysis. Publication bias was assessed visually by
funnel plots and then by Egger tests.
P < .05 was considered statistically significant.