Multiyear wildfire smoke exposure (MultiWiSE) metrics: a data-driven approach to characterizing episodic PM2.5 exposures for epidemiologic research

Abstract

Population exposure to wildfire smoke (WFS) has increased across North America. The health effects of short-term WFS exposure are widely documented, but little is known about longer-term exposures. Most epidemiologic studies use multiyear averages to characterize long-term air pollution exposure, but these do not reflect the episodic nature of WFS which may be associated with distinct health risks. We developed twelve data-driven Multiyear Wildfire Smoke Exposure (MultiWiSE) metrics that characterize the frequency, intensity, and duration of episodic WFS for application in epidemiologic studies examining the effects of longer-term exposures. The MultiWiSE metrics are easily calculable using any long time series of daily fine particulate matter (PM2.5). The approach first establishes a location-specific counterfactual, which is used to separate WFS from non-WFS PM2.5. Estimates of weekly WFS PM2.5 are then used to generate the metrics. We demonstrate this approach in British Columbia (BC), using PM2.5 estimates from the Canadian Optimized Statistical Smoke Exposure Model for 2010-2023. Of the MultiWiSE metrics, two describe cumulative exposure, four describe WFS-impacted weeks, five describe WFS episodes, and one describes recovery. When applied to 652 BC census subdivisions, WFS accounted for 1.7–24.5% of cumulative PM2.5 exposure, with a mean (range) of 45.2 (17–80) WFS-impacted weeks and 8.2 (3–15) WFS episodes. WFS episodes lasted up to 24 weeks, with an average recovery of 78.2 (41.1–178.5) weeks between episodes. Eleven metrics were positively correlated, with correlations ranging from 0.20 to 0.99 and a mean of 0.70, indicating they capture both overlapping and distinct features of multiyear WFS exposure. The MultiWiSE metrics characterize the frequencies, intensities, and durations of episodic WFS exposure and can be calculated using any multiyear time series of PM2.5. They can be used in epidemiologic studies for a more nuanced and actionable understanding of the health risks of longer-term exposures.