The analysis of the spatial variation of disease risk is crucial in Environmental Epidemiology studies. In this context, the effects of the presence of a source of pollution on the population health can be evaluated by models that consider distance from the source as a possible risk factor. We introduce an hierarchical Bayesian model in order to investigate the association between the risk of multiple pathologies and the presence of the risk source in the context of spatial case-control studies. Our approach extends some previous works based on spatial point patterns, concerning the risk variation of a single pathology and provides the possibility to incorporate spatial effects and other confounding factors within a logistic regression model. Moreover, spatial effects are decomposed into the sum of a disease-specific parametric component accounting for the distance from the point source and a common semi-parametric component that can be interpreted as a residual spatial variation. The proposed model is estimated by MCMC and is applied to data from a spatial case-control study in order to evaluate the association of the incidence of different cancers typologies with the residential location in the neighborhood of a petrochemical plant in the Brindisi area (South-eastern Italy).
(2010). Spatial variation of multiple diseases in relation to an environmental risk source [working paper]. Retrieved from http://hdl.handle.net/10446/950
Spatial variation of multiple diseases in relation to an environmental risk source
2010-06-01
Abstract
The analysis of the spatial variation of disease risk is crucial in Environmental Epidemiology studies. In this context, the effects of the presence of a source of pollution on the population health can be evaluated by models that consider distance from the source as a possible risk factor. We introduce an hierarchical Bayesian model in order to investigate the association between the risk of multiple pathologies and the presence of the risk source in the context of spatial case-control studies. Our approach extends some previous works based on spatial point patterns, concerning the risk variation of a single pathology and provides the possibility to incorporate spatial effects and other confounding factors within a logistic regression model. Moreover, spatial effects are decomposed into the sum of a disease-specific parametric component accounting for the distance from the point source and a common semi-parametric component that can be interpreted as a residual spatial variation. The proposed model is estimated by MCMC and is applied to data from a spatial case-control study in order to evaluate the association of the incidence of different cancers typologies with the residential location in the neighborhood of a petrochemical plant in the Brindisi area (South-eastern Italy).File | Dimensione del file | Formato | |
---|---|---|---|
graspa38_Calculli_Pollice_Bisceglia.pdf
accesso aperto
Dimensione del file
593.58 kB
Formato
Adobe PDF
|
593.58 kB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
Aisberg ©2008 Servizi bibliotecari, Università degli studi di Bergamo | Terms of use/Condizioni di utilizzo