The sisterhood method of estimating maternal mortality : a study in Matlab, Bangladesh

Abstract

The thesis is about ways to improve the technique of maternal mortality estimation known as the "sisterhood method', which entails questioning respondents on the circumstances of death for any sister who may have died. From the records of maternal deaths that occurred during 1976-90 in the Matlab Demographic Surveillance System (DSS) area, a field station of the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B), a list was prepared of deceased women's surviving brothers and sisters, aged 15 years and over, born to the same mother. One adult brother or sister of each of the deceased women was interviewed and asked whether he or she had a sister whose death was due to pregnancy, childbirth or in the 90 days after childbirth. Information on maternity-related deaths obtained in the survey was then compared with the information recorded in the DSS. Results suggest that in places similar to Matlab the sisterhood method will produce negative bias in estimates of maternal mortality mainly through misreporting of induced abortion-related deaths, particularly those associated with premarital conceptions. The analysis demonstrates that the problem giving rise to most of this negative bias is not with the method but with the sociocultural background of respondents, which encourages them to hide pregnancyrelated deaths to never-married women. These findings highlight the need for more research into ways of asking questions about pregnancy status of sisters, particularly the never-married sisters, who have died. This study not only found out that the sisterhood estimate of maternal mortality will contain a negative bias but also offered recommendations to improve reporting in future sisterhood surveys by identifying some important determinants of accurate reporting of maternal deaths. It then explored the possibilities for extending the sisterhood method. The main question asked in this regard was whether causes of maternal deaths could be studied using this method. It is concluded that though cause-of-death information obtained by the method will always be subject to some error, it can provide an indication of an overall distribution of causes of maternal deaths which can be of value to public health, particularly for the planning of programs aimed at reducing maternal mortality and for evaluation of such programs over time. Show more