Cause-specific neonatal deaths in rural Bangladesh, 1987–2005: Levels, trends, and determinants

Abstract

This paper reports on an analysis of neonatal mortality from communicable and non-communicable diseases in Bangladesh. The competing-risks model employed incorporated both observed and unobserved heterogeneity and allowed the two heterogeneity terms to be correlated. The data used came from the Health and Demographic Surveillance System, Matlab. The results confirm the conclusions of previous studies about the levels, trends, and causes of neonatal death in the Matlab area: the education of the mother helps protect her children from death from both communicable and non-communicable diseases; the children of a father in a low-status occupation are particularly vulnerable to death from communicable diseases; and children born to mothers aged less than 20 face a particularly high risk of dying from a non-communicable disease. The risks of dying from a communicable disease and from a non-communicable disease were both found to fall significantly as the distance to the nearest health centre decreased.

Keywords:

• Millennium Development Goals
• neonatal deaths
• competing-risks models
• unobserved heterogeneity
• Bangladesh

Notes

1. Unnati Rani Saha and Arthur van Soest are at Tilburg University, Warandelaan 2, 5037 AB, 5000 LE Tilburg, the Netherlands. Unnati Rani Saha is also at ICDDR,B, Dhaka, Bangladesh. E-mail: [email protected]. Govert E. Bijwaard is at the Netherlands Interdisciplinary Demographic Institute (NIDI).

2. The authors wish to thank three anonymous reviewers, and Katie Carman, Sabu Padmadas, and Frederic Vermeulen for their valuable comments which helped to improve the paper. They are grateful to Carel van Mels for providing them with the verbal autopsy codes used in Matlab and to Nurul Alam and Mahfuzur Rahman for the clarification of verbal autopsy data.