Clustering of neonatal deaths in Bangladesh: results from the Projahnmo studies.

BACKGROUND: Neonatal mortality (NM) tends to be clustered within a small subset of mothers, households and/or geographical areas. Knowledge of the maternal and newborn factors associated with NM can help identify high-risk mothers and guide the targeting of intervention programmes. METHOD: Data from pregnancy history surveys conducted as part of the Project for Advancing the Health of Newborns and Mothers (Projahnmo) in Sylhet and Mirzapur districts of Bangladesh were used to investigate risk factors for NM. We analysed data from babies born between 2001 and 2005 in the control clusters of the Projahnmo trials. Generalised linear mixed models were applied to quantify the heterogeneity among mothers and to investigate factors that contribute to this heterogeneity. RESULTS: There was an indication of correlation among siblings' outcomes. Neonates whose preceding sibling had died as a neonate in the mothers' lifetime pregnancy history were more likely (up to 1.9 times) to die than those with a living sibling. Factors that varied at the child and mother levels as well as the preceding siblings' outcome explained a large proportion (60% in Sylhet and 70% in Mirzapur) of the between-mother variation in NM. CONCLUSION: The preceding sibling's outcome may be a surrogate for genetic and other maternal health factors such as nutrition, infection or environmental factors that were not measured within Projahnmo. Further research into these factors is required in order to explain the variation in the risk for NM.