Spatiotemporal Analysis of Overall Health in the United States Between 2010 and 2018

Abstract

Background Although many previous studies have documented spatial heterogeneity in health outcomes across the United States at different geographic scales, spatiotemporal analyses to understand overall health are scant. Methodology We used the County Health Rankings (CHR) data to analyze the three types of health outcomes, viz., overall health, length of life, and quality of life for 2010-2018 in the contiguous United States employing hierarchal Bayesian methods. Composite scores were created to proxy these outcomes utilizing predefined weights of several variables as recommended by CHR. Our methods assumed a convolution of spatially structured and unstructured errors to model the overall spatial error. Spatial effects were modeled using conditional autoregressive distribution. Results The substantial disparity in these health outcomes was evident, with counties having poorer health outcomes mostly concentrated in the southeastern United States. Models that incorporated county-level demographic and socioeconomic characteristics partially explained the observed spatial heterogeneity in health outcomes. Interestingly, there was no time effect in any of the outcomes suggesting a perpetuation of health disparity over the years. Conclusions County-specific health policy interventions that take into account the contextual factors might be beneficial in improving population health and breaking the perpetuation of health disparity.

Keywords: bayesian statistics; county health rankings; health disparity; overall health; spatiotemporal.

Figure 1. Spatially structured errors for overall health, length of life, and quality of life from null and fully adjusted models.