CASM-Child: A Bayesian Hierarchical Model for Multivariate Count Data to Estimate Cause- and Age- Specific Mortality for Children in Data-Scarce Countries.

Austin E. Schumacher , University of Washington, Seattle
Tyler McCormick, University of Washington, Seattle
Li Liu, Johns Hopkins University
Jon Wakefield, University of Washington, Seattle

As investment increases in implementing age-targeted, disease-specific childhood interventions in data-scarce countries, effectiveness requires knowledge of the causes responsible for deaths in this important age group. Development of sample vital registration systems in low- and middle-income countries is progressing, motivating new methods to utilize this important data source. Current methods to model cause- and age-specific child mortality (i) use broad age groups which mask important heterogeneity, (ii) estimate all-cause and cause-specific mortality in two separate frameworks, (iii) produce estimates separately and independently in each age group, and/or (iv) do not account for empirically observed correlations between causes. We propose a novel Bayesian hierarchical model using sample registration data that accounts for correlations between cause- and age-specific mortality rates. We provide theoretical justification for this model, explore its properties via simulation, and use it to estimate age- and cause-specific mortality trends in sample registration data from China.

See extended abstract

 Presented in Session 187. Methodological Innovations in Modeling Health and Mortality