UC Berkeley

Policy makers in the health and development fields have established important goals and benchmarks for development progress, as showcased in the Declaration of Alma Ata (in 1978) and the Millennium Declaration (in 2000). Within these frameworks a clear and verifiable set of benchmarks were developed as a means of advancing transparency and accountability. Yet, often the data that are required to inform such evaluations, even for basic indicators such as child mortality, are lacking in low-income countries. This dissertation reviews the state of vital registration data systems in low-income countries, and then explores interim methods for measuring mortality when comprehensive civil registration systems are lacking. In particular, I evaluate the reliability of indirect methods to measure mortality rates under the age of 5 years by drawing on available data from Demographic and Health Surveys and from UNICEF's Multiple Indicator Cluster Surveys. I also carry out a validation study of indirect adult mortality estimation methods using historical population registers from northeast China, and assess the quality of mortality data collected at a demographic surveillance site in South Africa. The analysis demonstrates that indirect estimates, especially when adjustments are made to address simplifying assumptions, are generally consistent with adjusted direct estimates (in the case of child mortality) and life table estimates (in the case of adult mortality estimates). These results suggest that indirect estimates have a limited use in low-income countries, namely for populations that have experienced either smooth mortality declines or only short periods of excess mortality in their recent past. When examining data quality at the Agincourt demographic surveillance site, I found that such data provide a strong basis for the examination of health and mortality transitions in low-income countries given that they are less vulnerable to many of the data errors found in household surveys and population censuses in low-income countries. These findings underline the limitations of indirect methods, the promise of demographic surveillance systems as a valuable data source, and the critical importance of sustained improvements to civil registration systems.