目的 PM2.5 is a risk factor for respiratory diseases, but health effects associated with its extreme change are not well understood. To determine whether the number of hospital visits are sensitive to the PM2.5 extreme changes, we use an opportunity offered by greatly restricted air pollution emissions and the smog event in Beijing, China to measure pollutants daily and the outcomes during our study period. 方法 We used generalized additive modeling with a log link and allowed a time-series Poisson regression model to address possible serial correlation between PM2.5 concentrations levels and the number of hospital visits to estimate the improvement in outcome levels during the good air quality and the anticipated reversal of outcome levels in the followed smog period, and as sensitivity analyses, we replaced daily mean PM2.5 with daily hourly maximum or minimum PM2.5 to examine whether our results were robust to the use of alternative indicators.