METHODS: We obtained and analyzed daily data on hospital admissions, PM10 levels and BB days from five districts from 2005 to 2015. We identified BB days by evaluating the BB hotspots and backward wind trajectories. We estimated PM10 attributable to BB from the excess of the moving average of PM10 during days without BB hotspots. We fitted time-series quasi-Poisson regression models for each district and pooled them using meta-analyses. We adjusted for potential confounders and examined the lagged effects up to 3 days, and potential effect modification by age and sex.

RESULTS: We analyzed 210 960 respiratory and 178 952 cardiovascular admissions. Almost 50% of days were identified as BB days, with a mean PM10 level of 53.1 µg/m3 during BB days and 40.1 µg/m3 during normal days. A 10 µg/m3 increment in PM10 from BB was associated with a 0.44% (95% CI: 0.06, 0.82%) increase in respiratory admissions at lag 0-1, with a stronger association in adults aged 15-64 years and females. We did not see any significant associations for cardiovascular admissions.