This group's work focuses on the development of novel statistical methodologies for estimating mortality in low- and middle-income countries, where estimates of child mortality in particular come from probabilistic surveys as opposed to vital registration systems or censuses. Estimates of under-5 mortality rates (U5MR) are an important indicator of the health of a country. In low- and middle-income countries (LMICs), estimates of U5MR primarily come from probability surveys as opposed to vital registration systems or censuses. The most prevalent of such surveys are the Demographic and Health Surveys (DHS), which are known to have certain quality concerns endemic to probability surveys. One of these quality concerns is age-heaping, where a disproportionate number of child deaths are reported at common ages, such as 12 months. The exact frequency of age-heaping in DHS surveys is not well-studied, and its impacts on downstream statistical analyses - in particular, NMR, IMR, and U5MR estimation - has not been rigorously explored.
This project will investigate the prevalence of age-heaping in DHS surveys, conduct simulations with various “severities” of age-heaping, and assess the impact of age-heaping on commonly used statistical methods for estimating child mortality summary measures in settings akin to those found in LMICs.