Improved wildfire smoke exposure estimates are needed to gain a clearer understanding of the health impacts of wildfire PM2.5. Characterizing PM2.5 concentrations from wildfire smoke is challenging due to the transient nature of smoke. Current methods of determining smoke exposure rely on satellite retrievals of aerosol optical depth (AOD), estimates from chemical transport models (CTMs), or values reported by surface monitoring sites; each of these data sources has some limitations. To improve the accuracy of our exposure estimates, we developed new methods to blend these data. Our results indicate that blending information from the above-mentioned data sources along with counts of wildfire-smoke-related social-media posts results in better characterization of smoke exposure than any individual tool.
We link our daily smoke PM2.5 exposure estimates with hospitalization and urgent-care admission data from Washington, Oregon, and Colorado during several fire seasons as well as prescription filling data from Oregon. We find a robust relationship, where a 10 μg m-3 increase in smoke is significantly associated with a 9.5% (95% CI: 6.2, 12.9) increase in the rate of asthma admissions and a 7.7% increase (95% CI: 6.5. 8.8) in the risk for respiratory rescue medication prescription refills. There was no significant association between smoke exposure and any cardiovascular endpoints.
Our findings support the association of wildfire smoke exposure with adverse respiratory events, including subclinical outcomes, but we did not find significant associations with any cardiovascular outcomes. Public health messaging should target vulnerable populations to avoid smoke exposure during wildfire events.