We analyzed 243 Demographic and Health Surveys from 76 countries to develop an empirically-based method to indirectly estimate time trends in age-specific child mortality from summary birth histories. We used complete birth history data to train a discrete hazards model that was able to predict hazard functions for children based on individual, mother, and country-year level covariates. Age-specific mortality estimates were evaluated using cross-validation, and using an external database of an additional 243 censuses and surveys. The model was able to explain between 80% and 95% of the variance in the validation data and bias was near zero in every age group. For trends in all under-5s, performance was comparable methods used for the Global Burden of Disease Study, and significantly better than the standard indirect (Brass) method. Further external validation using census and survey data found close agreement with concurrent direct estimates in the neonatal and infant age groups.
Presented in Session 187. Methodological Innovations in Modeling Health and Mortality