Improving the Evidence Base for Maternal Health Outcomes Worldwide

Abstract

Maternal health is a key priority for the international community, as evidenced by commitment to Millennium Development Goal 5. However, evidence on maternal mortality is weak. This thesis addresses the shortcomings in the measurement of maternal mortality through three components. First, it presents a systematic assessment and analysis of all data available for the measurement of maternal mortality over the period 1980-2008, resulting in new estimates of maternal mortality for 181 countries with quantified uncertainty. The results show a remarkable decline in maternal deaths, from 526,300 in 1980 to 342,900 in 2008, an average annual decline of 1.5%. This new evidence suggests there is cause for greater optimism than generally perceived, and that substantial declines in the MMR are possible over relatively short periods of time. Second, the thesis presents an innovative, replicable, and generalizable approach to selecting the multivariable model for the trend analysis of maternal mortality. The third component presents the results of a comprehensive search for maternal deaths in Mexico over the period 2006-2010. This study shows encouraging progress towards the more complete and accurate assessment of causes of maternal mortality in a middle-income country, and offers a useful strategy for improving detailed, cause-specific maternal mortality data at the country level. The results also highlight the growing importance of indirect obstetric causes, which requires rethinking the health system response to maternal mortality, with health personnel trained to treat the entire woman, not just her pregnancy. Together, this dissertation fills important gaps in the maternal health field, providing new evidence and approaches to the measurement issues that have hampered progress in the past.