Objective To summarise current evidence evaluating the effects of human milk on the risk of bronchopulmonary dysplasia (BPD) in preterm infants.

Design We searched for studies on human milk and BPD in English and Chinese databases on 26 July 2017. Furthermore, the references of included studies were also screened. The inclusion criteria in this meta-analysis were the following: (1) preterm infants (<37 weeks); (2) human milk; (3) comparing with formula feeding; (4) the outcome included BPD; and (5) the type of study was randomised controlled trial (RCT) or cohort study.

Result A total of 17 cohort studies and 5 RCTs involving 8661 preterm infants met our inclusion criteria. The ORs and 95% CIs of six groups were as follows: 0.78 (0.68 to 0.88) for exclusive human milk versus exclusive formula group, 0.77 (0.68 to 0.87) for exclusive human milk versus mainly formula group, 0.76 (0.68 to 0.87) for exclusive human milk versus any formula group, 0.78 (0.68 to 0.88) for mainly human milk versus exclusive formula group, 0.83 (0.69 to 0.99) for mainly human milk versus mainly formula group and 0.82 (0.73 to 0.93) for any human milk versus exclusive formula group. Notably, subgroup of RCT alone showed a trend towards protective effect of human milk on BPD but no statistical significance.

Conclusion Both exclusive human milk feeding and partial human milk feeding appear to be associated with lower risk of BPD in preterm infants. The quality of evidence is low. Therefore, more RCTs of this topic are needed.