Abstract
Disease mapping is linked to two other scientific areas: small area estimation and ecological-spatial regression. This paper reviews similarities and differences among them. Bayesian hierarchical models are typically used in this context, using a combination of covariate data and a set of spatial random effects to represent the risk surface. The random effects are typically modeled by a conditional autoregressive prior distribution, and a number of alternative specifications have been proposed in the literature. The four models assessed here are applied to a study on alcohol abuse in Portugal, using data collected by the World Mental Health Survey Initiative.
Original language | English |
---|---|
Pages (from-to) | 79-101 |
Number of pages | 23 |
Journal | REVSTAT: Statistical Journal |
Volume | 13 |
Issue number | 1 |
Publication status | Published - Mar 2015 |
Keywords
- Alcohol abuse
- Bayesian hierarchical models
- Disease mapping
- Generalized linear models
- Small area estimation