Bayesian Spatial Hierarchical Modeling of Geographic Disparities in COPD Mortality in U.S. (2000--2007)

• Activity Number: 34
• Type: Contributed
• Date/Time: Sunday, August 4, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Xingyou Zhang*+ and James B. Holt and Anne Wheaton and Earl Ford and Janet B. Croft
• Companies: CDC and CDC and CDC and CDC and CDC
• Keywords: Bayesian Spatial Hierarchical Modeling ; Geographic disparities ; COPD mortality ; Spatial clustering ; spatial mixture model

Abstract

Chronic lower respiratory diseases (primarily chronic obstructive pulmonary disease or COPD) were the third leading cause of death in the US in 2011. Demographic disparities in COPD mortality are well recognized yet research is limited about geographic disparities. We explored a Bayesian hierarchical spatial model for the analysis of geographical variations in COPD mortality among US counties during the period 2000-2007. We examined the associations of county-level poverty, smoking, obesity, and rurality with COPD mortality; we accounted for spatial continuity and discontinuity of COPD geographic patterns by including spatial structured and unstructured random effects. We adopted a full Bayesian approach, via a Markov chain Monte Carlo computation implemented in WinBUGS, to permit more consistent inferences about the quantities of interest. Significant variations in COPD death rates were observed at both state and county levels. Poverty, current smoking and obesity were significantly associated with COPD mortality. Our spatial mixture model had a greater flexibility in specifying the spatial clustering effect and the local heterogeneity effect in geographic patterns of mortality.