Tools for Characterizing Multi-Pollutant Patterns of Exposure for Use in Environmental Health Studies

Lead Research Organisation: Imperial College London
Department Name: School of Public Health

Abstract

Air pollution is made up of gases and particles, with highest concentrations often found in areas with high levels of traffic congestion. There is strong evidence from research studies that exposure to various types of air pollution can increase the risk of disease or death, with particular impacts on the heart and lung. In real-life, most people are exposed to complex mixtures of air pollution components. However, because of a lack of available methods including statistical techniques, research studies are usually only able to consider health effects of one or two air pollution components at a time.
This project aims to develop new, advanced methods to help in investigating the health effects of mixtures of air pollution exposures. It will do this by developing air pollution profiles, composed of many pollutants at once, using very detailed air pollution data from London. It will also develop exposure profiles for metal dusts that are released from vehicles. The air pollution profiles will be clustered into risk groups that are associated with health outcomes. Additionally, the relationship of air pollution mixtures with more and less deprived areas will be investigated.
This project will result in rich sets of data that will help address important questions related to real-life exposures to air pollution and the relation between these exposures and health. The project team includes specialists in air pollution exposure, statistical methods and health research at the newly formed MRC-HPA Centre for Environment and Health at Imperial College London and King s College London. The methods they develop will be made freely available and the results will be disseminated to health researchers and local government and also made available on a website.

Technical Summary

There is now extensively reviewed evidence linking chronic air pollution exposures to all-cause, cardiovascular and respiratory mortality. However, such studies typically examine pollutants singly or in two pollutant models since the highly inter-correlated nature of air pollutants makes it difficult to examine their combined effects on health. However, air pollutants occur in complex mixtures consisting of highly correlated combinations of individual components, which makes marginal, single pollutant models inadequate. The objectives of this proposal therefore focus on the development of new, advanced multi-pollutant methods for investigating the health effects of gaseous and particulate air pollution exposures including metal components of particulate matter, and demonstrating their use in examining the combined effect of these pollutants on health outcomes and associations with deprivation. The methods will be developed and applied as part of the new MRC-HPA Centre for Environment and Health at Imperial College London and King s College London.
Our overall modelling approach is to cluster joint patterns of air pollution exposures, denoted as an air pollution profile, utilizing recently developed powerful Bayesian dimension-reduction techniques to characterize the pollutant patterns. For some applications, we will then link clusters of exposure profiles to outcomes of interest via a regression model, taking into account relevant confounders. Inference is then based on the joint pattern of pollution exposures and will be implemented via a unified Bayesian modelling framework.
We will demonstrate the benefit of using multi-pollutant profiling methods to characterize patterns of air pollution exposures and relate these exposure patterns to indicators of health and socio-economic status using data from the Office for National Statistics. Exposure estimates of NO2, NOx, PM2.5, PM10, and Ozone will be made available from the Environmental Research Group, King?s College London. Health and deprivation data for London will come from the Small Area Health Statistics Unit (SAHSU) at Imperial College, which holds postcoded mortality, hospital admission and cancer registration data and Census derived small area level measures of deprivation. Further, we will exploit the methods developed to examine relationships between traffic concentration and patterns of metal exposures obtained from 16 monitors placed throughout London.
The methods developed in this project will be made available to other researchers for use on their datasets, and the substantive results obtained will be of value to health researchers and policy makers. The research programme has thus a strong methodological as well as substantive component, which we believe will help advance this important field.
 
Title DiPBaC 
Description R package to implement profile regression analysis method 
Type Of Material Data analysis technique 
Year Produced 2012 
Provided To Others? Yes  
Impact Package only released 1 month ago, so no notable impacts that we are aware of to date. 
 
Title Profile Regression models developed and corresponding C++ program: Profile Regression C++ and R program version 2.0.0 by David Hastie 
Description We have developed sofisticated statistical modeling techniques to associate multi-pollutant air pollution profiles with health outcomes after adjusting for confounder variables. A sophisticated and computationally efficient C++ computer program has also been developed to implement the method. This program/modeling framework is currently being used to analyze exposure data in Greater London. 
Type Of Material Model of mechanisms or symptoms - human 
Provided To Others? No  
Impact Scientific papers using this modeling framework/program are forthcoming. 
 
Description Collaboration with Imperial College and Kings College - multi-pollutant modeling 
Organisation King's College London
Department School of Biomedical Sciences KCL
Country United Kingdom 
Sector Academic/University 
PI Contribution We are combining their modeled exposure data with our statistical models to examine important associations between air pollution exposures and health in Greater London.
Collaborator Contribution This research group is providing us with key modeling expertise and corresponding modeled exposure data which we are currently using to produce scientific papers which associate multi-pollutant profiles with health outcomes for individuals living in Greater London.
Impact Papers to be published in high-impact journals are forthcoming and progressing rapidly.
Start Year 2010