A prediction model for neonatal mortality in low- and middle-income countries: an analysis of data from population surveillance sites in India, Nepal and Bangladesh

The document discusses the development and validation of prediction models for neonatal mortality in low- and middle-income countries, focusing on data from India, Nepal, and Bangladesh. The models aim to identify high-risk infants to improve public health strategies. Data from over 49,000 live births and 1,742 neonatal deaths across rural and urban settings were analyzed, incorporating maternal, delivery, and infant characteristics. These include maternal education, economic status, prematurity, birth interval, and infant condition shortly after birth.

Three models were created, based on information available at different stages: the start of pregnancy, delivery, and five minutes post-birth. Predictive accuracy was moderate early on, improving significantly as more detailed information became available. Key predictors included low maternal education, prematurity, multiple births, and poor infant condition after delivery. The study highlights that neonatal deaths are concentrated among a small, high-risk group of infants, suggesting targeted interventions could significantly reduce mortality.

The authors recommend integrating these models into community-based and facility-based care strategies, enhancing early identification and management of at-risk pregnancies and newborns. They emphasize the need for broader public health measures, such as improving maternal nutrition and healthcare access, to address underlying risk factors. These findings underscore the potential of data-driven approaches to inform policies and reduce neonatal mortality in resource-limited settings.