Trends in and determinants of infant mortality in rural Bangladesh

Abstract

Objective: Study the trends in and determinants of infant mortality in rural Bangladesh. Methods: About 48,000 infants born during 1984-1990 were followed up for one year in the Demographic Surveillance Systems of three ICDDR,B field sites (Matlab, Abhoynagar, and Sirajgonj). Logistic regression was used in modelling mortality risks during infancy, evaluating deaths during the first 28 days of life (neonatal mortality) and during the rest of the first year (post-neonatal mortality). Results: Neonatal and post-neonatal mortality declined in the Matlab treatment area (receiving improved health and family planning services) and the Matlab comparison area. The mortality decline was concentrated between 4 and 14 days for neonates, and 1 and 5 months for post-neonates, which were the ages when the treatment area had lower mortality than the comparison area. There was no significant decline in neonatal mortality, and a modest reduction in post-neonatal mortality was observed at Sirajgonj, a more remote and traditional area. There was significant decline in post-neonatal mortality in Abhoynagar, although neonatal mortality did not decline during this period. In Matlab, risk factors for neonatal and post-neonatal mortality include lack of mothers' formal education and a short preceding birth interval. Additional risk factors include age of mother (young and old), religion (Hindu), sex (male), and birth order (2+) for neonates; and size of dwelling house (small) for post-neonates. Conclusions: Reductions in neonatal mortality in three field sites have been modest except in the Matlab treatment area, where maternal tetanus immunization has improved. Reductions in post-neonatal mortality have, however, been more substantial and widespread