Bayesian Spatial Cox Regression of Childhood Asthma Hospital Readmissions

• Activity Number: 83
• Type: Contributed
• Date/Time: Sunday, July 29, 2012 : 4:00 PM to 5:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Chen Chen*+ and Bin Huang
• Companies: Cincinnati Children's Hospital Medical Center and Cincinnati Children's Hospital Medical Center
• Keywords: Cox model ; Spatial correlation ; Conditional autoregressive prior ; Bayesian method ; Asthma ; Census-track SES

Abstract

Identification and analysis of spatial pattern for health outcomes have significant public health and health policy implications. Spatial modeling of health outcomes has seen rapid development recent years. Childhood asthma as one of the leading cause of childhood morbidity and impairment has shown geospatial patterns. Motivated by a research study investigating health disparity of childhood asthma morbidity, this study implements a Bayesian Cox spatial survival model which allows for spatial autocorrelation in neighboring or proximal locations. Hospital records of asthma related readmission between 2008 and 2010, and Census track socio-economic variables from a large Midwest county are examined. A repeated measures proportional hazards model is considered. Unmeasured/unknown spatial confounders are incorporated into the model as random effects with a conditional autoregressive (CAR) prior. Our study results suggest that individual insurance type, race and season of index asthma hospital admission are associated with the risk for readmission. And we also identified neighborhood employment as significant risk factors to the risk of childhood asthma hospital readmission.