Using census data to measure maternal mortality: A review of recent experience.

Abstract

The Sustainable Development Goals adopted by the United Nations General Assembly in 2015 (United Nations 2015) set national targets for reducing maternal mortality, putting pressure on governments of countries lacking comprehensive statistical systems to find other ways to measure it. One approach tested since the 1990s has been to collect necessary data through national population censuses. This paper reviews maternal mortality data from the 2010 round of censuses for several countries to determine whether the census is useful for monitoring maternal mortality. Data on births, deaths, and pregnancy-related deaths from two censuses for 10 countries was evaluated using standard methods; adjustments were applied to the reported numbers if so indicated. In general, the censuses underreported births moderately and underreported deaths by larger amounts; except in one case, proportions of pregnancy-related deaths appeared plausible. Adjusted estimates of the pregnancy-related mortality ratio (PRMR) were generally higher than estimates from Demographic and Health Survey sibling data or estimates of maternal mortality developed by cross-national studies. Analysis of recent data confirms results of earlier assessments: Census data provides imperfect but still valuable information on maternal mortality. Data requires careful assessment and often adjustment, resulting in estimates with large uncertainty. This paper provides additional evidence as to whether maternal mortality can usefully be measured by population censuses in countries lacking civil registration data.