“…Sensitivity analysis for attrition bias caused a 20% increase (RR 1.20, 95% CI 1.11 to 1.29; 12 studies; 83,868 participants) due the removal of two studies [25,33]. Subgroup analysis by setting showed facility-based [24,25,27,33,51,63,75,88] and community-based [38,49,57,73,83] had significant effects without significant heterogeneity between subgroups. Subgroup analysis by timing of intervention showed postnatal [24,25,88] and prenatal/postnatal interventions [27,33,44,49,57,73,75,83] both had a significant effect, whereas prenatal interventions [38,51,63] had no effect on early initiation of breastfeeding.…”