In the last 30 years, a large number of studies have provided substantial statistical evidence of the adverse health effects associated with air pollution. Statistical literature is very rich and includes a plethora of models to manage different types of spatial data. This paper starts with a thorough discussion on the spatial nature of the available data on health and air pollution. Health data are usually provided by Health Authorities as mortality and morbidity counts at a small area level. Thus we mainly focus on reviewing and discussing the spatial and spatio-temporal regression models proposed for disease count data on irregular lattices. In general, measuring the effect of exposure on health outcomes is an extremely hard task, and to obtain reliable estimates of the exposure effect and associated uncertainty one needs to build models that account for the residual variability not captured by the exposure-response relationship. In this context, Bayesian hierarchical models including spatial random effects play a prominent role: we consider both univariate and multivariate models and discuss some extensions to the spatio-temporal setting. Since model estimation can be prohibitive, practitioners are provided with a list of available software for Bayesian inference that avoids the need for complicated coding.

A survey on ecological regression for health hazard associated with air pollution

IPPOLITI, Luigi;VALENTINI, PASQUALE;
2016-01-01

Abstract

In the last 30 years, a large number of studies have provided substantial statistical evidence of the adverse health effects associated with air pollution. Statistical literature is very rich and includes a plethora of models to manage different types of spatial data. This paper starts with a thorough discussion on the spatial nature of the available data on health and air pollution. Health data are usually provided by Health Authorities as mortality and morbidity counts at a small area level. Thus we mainly focus on reviewing and discussing the spatial and spatio-temporal regression models proposed for disease count data on irregular lattices. In general, measuring the effect of exposure on health outcomes is an extremely hard task, and to obtain reliable estimates of the exposure effect and associated uncertainty one needs to build models that account for the residual variability not captured by the exposure-response relationship. In this context, Bayesian hierarchical models including spatial random effects play a prominent role: we consider both univariate and multivariate models and discuss some extensions to the spatio-temporal setting. Since model estimation can be prohibitive, practitioners are provided with a list of available software for Bayesian inference that avoids the need for complicated coding.
File in questo prodotto:
File Dimensione Formato  
SPASTA-Accepted.pdf

accesso aperto

Descrizione: versione accettata del paper publicato al seguente link: https://doi.org/10.1016/j.spasta.2016.05.003
Tipologia: Documento in Post-print
Dimensione 358.22 kB
Formato Adobe PDF
358.22 kB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11564/669373
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 5
  • ???jsp.display-item.citation.isi??? 5
social impact