Institutional deliveries weakly associated with improved neonatal survival in developing countries: Evidence from 192 Demographic and Health Surveys

Background: Child and maternal mortality remain high in many developing countries. A principal strategy used in low- and middle-income countries is increasing the proportion of pregnancies delivered at facilities. Although this strategy is reasonable with high quality facilities, evidence for the protective effects of facility deliveries is mixed. Methods: We pooled 1.47 million birth records collected by the Demographic and Health Surveys to estimate the association between institutional deliveries and early neonatal mortality. Subsample analysis and instrumental variable estimation were used to assess and correct the extent to which mortality differentials are biased by an increased likelihood of facility attendance for high-risk deliveries. Results: No associations between institutional deliveries and early neonatal mortality were found in the pooled sample [adjusted odds ratio (aOR) 0.995, 95% confidence interval (CI) 0.966-1.025)]. When stratified by facility type, protective effects were found for private facilities (aOR 0.876, 95% CI 0.840-0.914), but not for public hospitals or health centres. Significant protective effects were found when past behaviour was used to eliminate selection bias generated by short-term responses to medical need (aOR 0.884, 95% CI 0.814-0.961). At the community and country levels, strong positive associations were found between early neonatal mortality among facility deliveries and the prevalence of institutional deliveries. Conclusion: Facility deliveries have the potential to reduce early neonatal mortality in developing countries. The results presented suggest that the quality, utilization and protective effects of institutional deliveries vary widely across countries; major improvements in both utilization and quality of care will be needed to achieve further improvements in maternal and child health.