Integrated Interventions Delivered in Health Systems, Home, and Community Have the Highest Impact on Breastfeeding Outcomes in Low- and Middle-Income Countries.

Background: Improving breastfeeding rates is critical. In low- and middle-income countries (LMICs), only subtle improvements in breastfeeding rates have been observed over the past decade, which highlights the need for accelerating breastfeeding promotion interventions.Objective: The objective of this article is to update evidence on the effect of interventions on early initiation of and exclusive (<1 and 1-5 mo) and continued (6-23 mo) breastfeeding rates in LMICs when delivered in health systems, in the home or in community environments, or in a combination of settings.Methods: A systematic literature search was conducted in PubMed, Cochrane, and CABI databases to identify new articles relevant to our current review, which were published after the search date of our earlier meta-analysis (October 2014). Nine new articles were found to be relevant and were included, in addition to the other 52 studies that were identified in our earlier meta-analysis. We reported the pooled ORs and corresponding 95% CIs as our outcome estimates. In cases of high heterogeneity, random-effects models were used and causes were explored by subgroup analysis and meta-regression.Results: Early initiation of and exclusive (<1 and 1-5 mo) and continued (6-23 mo) breastfeeding rates in LMICs improved significantly as a result of interventions delivered in health systems, in the home or community, or a combination of these. Interventions delivered concurrently in a combination of settings were found to show the largest improvements in desired breastfeeding outcomes. Counseling provided in any setting and baby-friendly support in health systems appear to be the most effective interventions to improve breastfeeding.Conclusions: Improvements in breastfeeding practices are possible in LMICs with judicious use of tested interventions, particularly when delivered in a combination of settings concurrently. The findings can be considered for inclusion in the Lives Saved Tool model.