We estimated separately under-five mortality (ages 0-4) and adult mortality (ages 15-59) rates for 188 countries using all available data sources including surveys, censuses, sample registration systems, disease surveillance, and vital registration systems. Summary birth history data from surveys and censuses have been analyzed using updated formal demographic methods developed for the Global Burden of Disease study. Death reporting from vital registration and censuses were evaluated for completeness using improved death distribution methods. We also applied improved sibling survival method to survey data on sibling survival modules to correct for survivor bias, zero-survivor bias, and recall bias. Estimates of under-five and adult mortality were generated using the state-of-art data synthesis methods: a combination of spatio-temporal and Gaussian Process regressions.
The final estimates of under-five and adult mortality rates, together with crude death rates due to HIV/AIDS, were used as entry parameters to a new relational model life table system. In this new model life table system developed for the GBD 2013, we have a more flexible standard life table selection process to account for similarities in life tables based on levels of under-five and adult mortality rates and distance between two life tables in terms of time and geographic location. Impacts of conflict, war and natural disaster on age pattern of mortality are modeled separately. Age and sex specific deaths for 188 countries for years 1970-2013 are then generated together with population estimates. Uncertainty has been propagated throughout the entire process and 95% uncertainty intervals have been provided for each metric we generate.