Maternal mortality estimation: separating pregnancy-related and non-pregnancy-related risks.

The increased availability of survey data and improved estimation techniques have furthered our understanding of maternal mortality in developing countries. Both the indirect and direct sisterhood methods of estimation depend on time-of-death information from surveys. This report proposes a method for calculating two rates, one during the pregnancy period and one outside of it. Analysis of both rates provides more information about mortality associated with pregnancy than do methods that only produce one rate. The pregnancy-related mortality rate can be estimated by assuming that non-pregnancy-related risks are constant, irrespective of whether women are pregnant or not. An estimated 69 percent of deaths in Bolivia during pregnancy may be pregnancy related; this result is significantly lower than that obtained using the traditional sisterhood method. In certain cases, this result may be viewed as a plausible lower bound. A variety of estimates should probably be used for policy purposes.