Variation in Fine Particulate Matter Exposure Models: Implications for Health, Equity, and Uncertainty Evaluation

We seek to understand how uncertainty in spatiotemporal models of PM2.5 concentration influences exposure, equity, and health metrics in Canada. We evaluated the accuracy of seven longitudinal PM2.5 concentration data sets from 2002 to 2014 and conducted equity and health impact analysis using each. Specifically, we estimated population exposure improvements over time, exposure disparities for vulnerable subpopulations, and PM2.5-attributable mortality across several combinations of inputs. Nationally, reductions in population exposure to PM2.5 were consistently observed in all models between 2002 and 2014 in Canada, with larger reductions among subpopulations of visible minority status, suggesting that PM2.5 exposure and absolute exposure disparities improved over this period. Agreement between PM2.5 models is strongest for PM2.5-attributable mortality, whereas exposure time trends and contrasts between subpopulations differ more widely. Nationally, estimated PM2.5-attributable mortality ranged from 7,400 to 26,000 deaths in 2014 across combinations of seven PM2.5 models, four Canadian risk estimates, and counterfactuals of 0 and 2.5 μg/m3. PM2.5-attributable death estimates are sensitive to inputs of mortality risks, counterfactuals, and PM2.5 models. Improving estimates of Canadian PM2.5 exposure remains important for reducing uncertainty in health studies, particularly for accountability, equity, and epidemiological studies that rely on spatial or temporal contrasts.