CONCLUSION
This comprehensive NMA confirmed that corticosteroids were mostly effective for neonatal and child relevant outcomes compared with placebo or no treatment. There was no important difference between corticosteroids on neonatal death, neurodevelopmental disability, IVH and birthweight. Low to moderate-CE suggest that dexamethasone may reduce chorioamnionitis, and foetal death but may increase puerperal sepsis and RDS. However, the 95%CI indicates both beneficial and detrimental effects for these outcomes. The opposing direction of these outcomes does not allow to derive recommendations about what corticosteroid should be used and large well designed RCTs are warranted to improve the certainty of evidence. Ideally, they should represent low resource settings and also evaluate the best schemes of administration. Individual participant data meta-analysis could help to answer these questions. In the meantime, monitoring short-term and long-term health outcomes, including neurodevelopmental disability will be important.
Since there is no robust evidence on which corticosteroid should be prescribed, decisions should be based on availability, costs, opportunity, and facilities. Shared decision-making would help patients to take their choices when facing this scenario.