Epidemiological studies have widely demonstrated association between ambient ozone and mortality, though controversy remains, and most of them only use a certain metric to assess ozone levels. However, in China, few studies have investigated the acute effects of ambient ozone, and rare studies have compared health effects of multiple daily metrics of ozone. The present analysis aimed to explore variability of estimated health effects by using multiple temporal ozone metrics. Six metrics of ozone, 1-h maximum, maximum 8-h average, 24-h average, daytime average, nighttime average, and commute average, were used in a time-series study to investigate acute mortality associated with ambient ozone pollution in Guangzhou, China, using 3 years of daily data (2006-2008). We used generalized linear models with Poisson regression incorporating natural spline functions to analyze the mortality, ozone, and covariate data. We also examined the association by season. Daily 1- and 8-h maximum, 24-h average, and daytime average concentrations yielded statistically significant associations with mortality. An interquartile range (IQR) of O3 metric increase of each ozone metric (lag 2) corresponds to 2.92 % (95 % confidence interval (CI) 0.24 to 5.66), 3.60 % (95 % CI, 0.92 to 8.49), 3.03 % (95 % CI, 0.57 to 15.8), and 3.31 % (95 % CI, 0.69 to 10.4) increase in daily non-accidental mortality, respectively. Nighttime and commute metrics were weakly associated with increased mortality rate. The associations between ozone and mortality appeared to be more evident during cool season than in the warm season. Results were robust to adjustment for co-pollutants, weather, and time trend. In conclusion, these results indicated that ozone, as a widespread pollutant, adversely affects mortality in Guangzhou.
