Background: Studies have examined the variation in mortality risk associated with ambient ozone. We conducted an international cooperation study for investigating the non-linear ozone–mortality association accounting for lag effect and for examining the ozone level with significant health effect. Methods: Daily counts of total non-accidental death and ambient air pollutant concentration were obtained in 17 cities from 3 Eastern Asian countries or regions (Taiwan, Korea, and Japan). The total study period was from 1979 to 2010 and differed by city based on data availability. The ozone–mortality association in each city was estimated by running a time-series quasi-Poisson regression model, allowing for overdispersion. The city-specific estimates were then pooled by country by using multivariate random effects meta-analysis. Results: The non-linear ozone exposure–mortality response curves were generated in 17 cities from Taiwan, Korea, and Japan. The association curves in the three countries all showed increased mortality with elevated ozone, and the significant mortality effects of ozone exposure were observed at level higher than 40, 50, and 40 ppb for Taiwan, Korea, and Japan, respectively. These associations are unaffected by co-pollutant of particulate matter in Taiwan and Japan. But the potential confounding effect of co-pollutant could not be ignorable in Korea. Conclusions: Our study provides evidence that exposure to a relative low level of ambient ozone is associated with an increased risk of mortality. Our results emphasize the continual need to examine the existing standard by documenting potential human adverse effects.
