Online Program

Saturday, February 21
CS23 Population Modeling Sat, Feb 21, 11:00 AM - 12:30 PM
Napoleon C

Bayesian Spatial Joint Modeling of Asthma Admission and Readmission for Identifying High-Risk Neighborhood (302920)

Andew Beck, Cincinnat Children's Hospital Med Ctr 
Chen Chen, Cincinnat Children's Hospital Med Ctr 
*Bin Huang, Cincinnati Children's Hospital Medical Center 
Robert Kahn, Cincinnat Children's Hospital Med Ctr 
Patric Ryan, Cincinnat Children's Hospital Med Ctr 

Keywords: Bayesian Joint Modeling, Spatial Conditional Autoregressive (CAR) model, model fitting criteria, model prediction

Asthma is the most common pediatric chronic physical health condition. Many children being hospitalized for asthma will be readmitted into the hospital within a year. Spatial modeling has been applied to asthma outcomes for identifying neighborhoods or regions at higher risk. Accurate identification of the high-risk neighborhood will effectively inform better policy and medical decisions. The Greater Cincinnati Asthma Readmission Study (GCARS) is a pediatric hospital patient population cohort study. Considering special features of GCARS: 1) index admission could be a readmission itself of the patient, 2) individual-level risk factors and readmission outcomes are only available to a subset of patients who agreed to a detailed interview, 3) risk factors at census track–level are available to all study participants. We developed a Bayesian joint spatial modeling of asthma admission and readmission outcomes. A large numerical study evaluated the performances of the proposed modeling. The study results suggest joint spatial modeling performs better at identifying high-risk neighborhoods, compared to separate modeling.