The importance of the exposure metric in air pollution epidemiology studies: When does it matter, and why?

Abstract

Exposure error in ambient air pollution epidemiologic studies may introduce bias and/or attenuation of the health risk estimate, reduce statistical significance, and lower statistical power. Alternative exposure metrics are increasingly being used in place of central-site measurements, with the intent of reducing exposure error. Dependent on the study design, health outcome, and pollutant of interest, these metrics may provide a means of reducing error (leading to less bias and uncertainty in health risk estimates) if they capture variability in exposure which is not represented when central-site measurements are used. We examine the current evidence for answering the question of when the choice of exposure metric matters and why, focusing on studies which examined multiple exposure metrics in the same epidemiologic study. We conclude that for timeseries and case-crossover studies, central-site measurements may be sufficient, especially for homogeneous pollutants, and in cohort and panel studies, approaches that increase spatial resolution of the exposure metrics do impact the health effect estimates. We note that while the current literature is widely dispersed across exposure metrics and health outcomes, the largest collective, common body of literature is focused on birth/pregnancy outcomes and traffic-related pollution. Also additional discussion and agreement is needed on how to classify metrics as Bdifferent^and Bbetter.^Primary recommendations are to provide context for the reasons behind selection of exposure metrics and to encourage collaboration between exposure scientists and epidemiologists when designing and implementing a study, as results can have important implications for the development of policies and regulations.

Concern over health effects of air pollution is associated to NO2 in seven European cities

“…Previous studies have shown that the spatial distribution of traffic-related air pollution is generally stable over periods 10-15 years, supporting the use of an annual average of a single year preceding our questionnaire to assess long-term exposure (Eeftens et al 2011;Gulliver et al 2011). Also the air pollution model was raster-based (100mx100m) and this may have led to exposure misclassification and lower correlations (Dionisio et al 2016). The question on subjective perception of air pollution in the questionnaire referred to air pollution in the neighbourhood of home or work, yet we only tested out the relationship at the home address, which again may have blurred the results.…”

Integrating Modes of Transport in a Dynamic Modelling Approach to Evaluate Population Exposure to Ambient NO2 and PM2.5 Pollution in Urban Areas

“…The principle idea of exposure is the combination of pollutant concentration values in the environments where people spend their time, and the amount of time they spend within them [4]. Thus, the introduced approaches to model urban-scale ambient pollutant concentrations and generic TMA models need to be combined to calculate meaningful exposure metrics [84], such as total population exposure or population-weighted population exposure (PWE).…”

Mapping Modeled Exposure of Wildland Fire Smoke for Human Health Studies in California

“…Previous health studies that use atmospheric models have been conducted at coarser geographic scales (e.g., GEOS-Chem) and have considered only fire-derived pollutants rather than fire in conjunction with PM from other sources (e.g., traffic, utilities) when assessing associations with health, which may overestimate the fire-specific exposures [60,73]. Other studies have used atmospheric modeling in conjunction with adjustments from air quality monitoring data, satellite remote sensing data or additional post-processing statistical techniques [68,71,[74][75][76][77][78][79].…”