Many studies have documented significant divergence in U.S. county-level mortality trends since 2000. Recent analyses have utilized spatially-explicit Bayesian hierarchical models to make robust estimates of county-level mortality over space and time. However, few studies have examined a comprehensive set of time-varying contextual characteristics within such a modelling framework to illustrate how differences in mortality trajectories by county are associated with shifting levels of these predictors. Combing vital statistics data with county-level characteristics related to healthcare, health behaviors, socioeconomic profile and population composition, this paper utilizes a spatially-explicit Bayesian hierarchical modelling framework to analyze how changing levels of mortality across age groups are associated with changes in county-levels exposures. We employ a Shapley decomposition on the time-varying components of our models to illustrate the additive contributions of each changing characteristic to the observed mortality change in each U.S. county since 2000.
Presented in Session 86. Spatial Distribution of Diseases and Deaths