Background: Wildfire smoke, particularly fine particulate matter PM2.5, represents a significant public health risk because of climate change-driven increase in fire activity. Although wildfires are becoming more frequent in South Korea, comprehensive...
Background: Wildfire smoke, particularly fine particulate matter PM2.5, represents a significant public health risk because of climate change-driven increase in fire activity. Although wildfires are becoming more frequent in South Korea, comprehensive nationwide studies quantifying the spatiotemporal heterogeneity of respiratory morbidity attributable to wildfire smoke are scarce.
Methods: A nationwide exposure–health assessment was conducted across 254 administrative districts from 2015 to 2024. A machine learning-based counterfactual model, specifically the Histogram-based Gradient Boosting Regressor, was used to distinguish wildfire-attributable PM2.5 from background concentrations. Exposure was quantified using two metrics: (1) daily "smoke days" (wildfire-specific PM2.5 ≥15μg/m^3) and (2) hourly "wildfire-affected" status determined by hourly wildfire events and a geometric plume dispersion model. Associations with respiratory emergency department (ED) visits were estimated with a time-stratified case-crossover design, while spatial heterogeneity was assessed with a Spatial Bayesian Hierarchical Model (BHM).
Results: Wildfire smoke exposure was significantly associated with increased respiratory ED visits. In the daily analysis, the strongest effect was observed at lag 1 (OR = 1.062), corresponding to approximately 28,935 excess visits nationwide. Hourly analysis indicated that health risks were both immediate and sustained, with a consistent elevation in risk during wildfire-affected hours (lag 1 OR = 1.056). The Spatial BHM analysis revealed substantial heterogeneity, with high-risk clusters identified not only in fire-prone inland mountains but also in southwestern coastal regions and major metropolitan areas. These associations were not significantly modified by community-level socioeconomic factors, indicating widespread vulnerability.
Conclusion: Wildfire smoke imposes a substantial and geographically uneven respiratory health burden in South Korea, affecting both ignition zones and downwind populations. These findings underscore the limitations of current daily-average monitoring and highlight the urgent need for geographically targeted mitigation strategies, particularly in vulnerable coastal regions.