~ Pergamon 0277-9536(95)00182-4 Soc. Sci. Med. Vol. 42, No. 6, pp. 831-841. 1996 Copyright © 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0277-9536/96 $15.00 + 0.00 ESTABLISHING LINKS BETWEEN AIR QUALITY AND HEALTH: SEARCHING FOR THE IMPOSSIBLE? CHRISTINE DUNN j and SIMON KINGHAM"* ~Department of Geography, University of Durham, South Road, Durham DH1 3LE, England and -~Department of Epidemiology and Public Health, The Medical School, Universityof Newcastle Upon Tyne, Newcastle Upon Tyne NEI 4HH, England Abstract--Traditional approaches in environmental spatial epidemiology have relied on assessing postulated links between environmental pollution and ill health, often as a response to a perceived public health problem; clearlyit may be necessaryto go beyond this stage in order to establish the nature of potential causal mechanisms. Different disciplines approach this issue in different ways. Many toxicologists favour approaches based on air quality monitoring, where raised levelsof candidate pollutants may subsequently generate hypotheses about adverse health effects. Epidemiologists, however, assess the health of a population and then look for an associated cause. This paper suggests that neither approach is completely satisfactory and that a combination of both is needed. If spatially referenced data are available for both health status and air quality, then geographical analysis is needed to examine possible links, by using techniques such as atmospheric dispersion modelling and Geographical Information Systems. We discuss the benefits and constraints of these approaches, using empirical examples of environmental epidemiology studies for northern England. Taking into account the problems involved in such studies, allied to the high costs incorporated, the paper asks the question: Are we searching for the impossible? Key words--health, environment, air quality INTRODUCTION Workers in many disciplines are potentially valuable contributors to the study of environmental epidemiol- ogy. Such studies may involve assessing the influence of a variety of environmental factors on the health of a population, including drinking water and bathing water quality, food systems through contamination of soils, or geological effects on indoor home environ- ments. The conceptual and methodological approach taken is clearly influenced by the nature of the driving discipline, such that in investigating spatial differences in health status, a social anthropologist, for example, adopts a different perspective to an epidemiologist. To a large extent this type of work has been approached in a rather piecemeal way, with epidemiologists, toxicologists or medical geographers, for example, operating independently and within disciplinary boundaries. Although clearly a multidisciplinary research area, limited resources and the need for tight co-ordination may dissuade individuals or groups from embarking on cross-disciplinary ventures in this type of work. That said, we argue here that an integrated approach, through collaborative research teams, is most likely to be able to address research questions about associations between health status and environmental factors, and more importantly about potential causal mechanisms. There remains~ *Present address: The Institute of Environmental and Policy Analysis, University of Huddersfield, St Peter's Building, St Peter's Street, Huddersfield HD1 1RA, England. nevertheless, a plethora of problems to be faced in answering such questions and in this paper, we examine some of the difficulties faced when attempting to link air quality data from urban environments, to public health, notably respiratory disorders. We consider the relevance of different theoretical frameworks undertaken as part of this type of research, and the constraints involved in integrating results from these alternative perspectives. Case studies from northern England are used to illustrate the arguments and to demonstrate some of the difficulties involved. Spatial and temporal changes in air quality pose particular difficulties in developing hypotheses about causal links to specific measures of health. Although in some cases, air quality assessment may be relatively straightforward where, for example, a single source of industrial pollution is easily identified and other sources of contamination are considered insignificant, in most cases it is important to consider the potential impact of interaction effects between pollutants from multiple point and linear sources. In addition, recent urban and industrial developments have shifted the focus somewhat, at least in developed western economies, away from looking at the health effects of 'traditional' atmospheric pollutants such as smoke and SO,,. In the U.K., for example, Clean Air Acts (1956, 1968) resulted in lowering of the levels of these components in the urban atmosphere and the number of associated routine monitoring stations has subsequently been reduced. Other sources such as road 831