Is Non-Linear Concentration-Response Function between Daily Mean 03 Concentration and Mortality Biased?

摘要

Background: The linearity of concentration-response (C-R) function between ambient O3 concentration and mortality has been controversial. Previous reports of non-linear C-R function has been suspected to be biased from unmeasured confounders. The aim of the present analysis is to examine the C-R function between 03 concentration and mortality with instrumental-variable approach. Methods: We extracted data of hourly meteorology, hourly O3 concentration and daily non-accidental mortality in Seoul from 2001 to 2009. We divided the dataset into two, odd-number (training set) and even-number years (testing set). Using the training set, we constructed a prediction model with support vector regression estimating the daily variations of mean O3 concentration caused by sun irradiance, wind speed and direction, uncorrelated with temperature and temporal trend. Then we predicted variance of daily O3 from the testing set using the prediction model, thus creating an instrumental variable. We analyzed the association between the instrumental variable and daily mortality. We also analyzed the association according to the different ranges of daily mean O3 concentration. Result: The cross-validated R2 was 0.20 and 0.15 for the prediction model with training set and testing set, respectively. The instrumental variable was significantly and negatively associated with daily mortality (3=-0.004$2SE=0.0009, P<0.0001). The negative slope of the association diminished as the quartile increased and the slope became positive in the 4th quartile (O3>29.7 ppb). The interaction between quartiles and instrumental variable was significant (P=0.0485). The slope was negative under 30 ppb of daily O3 concentration (B=-0.0041, SE=0.0011, P=0.0001), but it was positive over 30 ppb ((3=0.0012, SE=0.0022, P=0.6000). Conclusion: We observed unequal C-R function between ambient O3 concentration and mortality according to the different ranges of daily mean O3 concentration with instrumental-variable approach.