Risk factors for neonatal mortality in low-income population subgroups in Karachi, Pakistan

This paper assesses the risk factors for neonatal mortality (0-28 days of life) for full-term singleton live births. The study sample comprised 1,011 ever-married women who were residents of four multi-ethnic, multi-religious low-income areas of Karachi, Pakistan during 1995. The analysis sample was restricted to 4,488 live births to 912 women. Results showed that 4.8% (n = 217) of all births ended in death in the neonatal period, and 76% (n = 164) of these neonatal deaths occurred in the first week of life. While neonatal mortality rates had declined appreciably over time, a large proportion of neonatal deaths were clustered in a small group of women. The univariate analysis showed a statistically significant association between a number of maternal-level parameters (e.g., mother's age at birth, level of formal education, employment status, religious affiliation, and consanguinity) and child-level parameters (e.g., birth order, birth interval, survival status of the preceding child, sex of the neonate, year of death). At the multivariate level, consanguinity, birth order and year of death were found to be statistically significant risk factors for neonatal mortality analyses despite adjustment for death clustering. The study highlights the importance of inbreeding as a determinant of early offspring death, and points to the need for better understanding of why deaths are clustered in certain families. Given the widespread practice of close cousin marriages in Pakistan across all socio-economic groups, there is a need to validate the findings in less economically disadvantaged population subgroups within the country.