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

High Resolution Image Download MS PowerPoint Slide We seek to understand how uncertainty in spatiotemporal models of PM 2.5 concentration influences exposure, equity, and health metrics in Canada. We evaluated the accuracy of seven longitudinal PM 2.5 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 PM 2.5 -attributable mortality across several combinations of inputs. Nationally, reductions in population exposure to PM 2.5 were consistently observed in all models between 2002 and 2014 in Canada, with larger reductions among subpopulations of visible minority status, suggesting that PM 2.5 exposure and absolute exposure disparities improved over this period. Agreement between PM 2.5 models is strongest for PM 2.5 -attributable mortality, whereas exposure time trends and contrasts between subpopulations differ more widely. Nationally, estimated PM 2.5 -attributable mortality ranged from 7,400 to 26,000 deaths in 2014 across combinations of seven PM 2.5 models, four Canadian risk estimates, and counterfactuals of 0 and 2.5 μg/m 3 . PM 2.5 -attributable death estimates are sensitive to inputs of mortality risks, counterfactuals, and PM 2.5 models. Improving estimates of Canadian PM 2.5 exposure remains important for reducing uncertainty in health studies, particularly for accountability, equity, and epidemiological studies that rely on spatial or temporal contrasts.