|
Activity Number:
|
148
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #310402 |
|
Title:
|
Bayesian Hierarchal Modeling of Small-Area Suicide Rates: A Geographical and Ecological Evaluation
|
|
Author(s):
|
Yi Lin*+ and Ying MacNab
|
|
Companies:
|
University of British Columbia and University of British Columbia
|
|
Address:
|
Room E414, Vancouver, BC, V6H3V4, Canada
|
|
Keywords:
|
Bayesian hierarchical model ; shared component model ; conditional autoregressive model ; suicide rates ; geographical variation ; ecological analysis
|
|
Abstract:
|
We present a Bayesian spatial and ecological analysis of hospitalization and mortality rates of suicide among children and youth in British Columbia, Canada. The Bayesian spatial modeling enabled us to produce more reliable risk estimates and to identify regional level risk factors. Spatial and shared component models were used to explore and quantify spatial correlation in relative risk and shared component effects. Empirical Bayes penalized quasi-likelihood method is used to identify risk factors and fully Bayesian MCMC method is used to quantify uncertainties associated with the estimation of all model parameters, particular the regional relative risks. Principal component method was used to explore multicollinearity among covariates and to control for confounding. Regional characteristics explained considerable variations in suicide rates.
|
