Document Type


Publication Date



Ozone (O3)-induced health effects vary in terms of severity, from deterioration of lung function and hospitalization to death. Several studies have reported a linear increase in health risks after O3 exposure. However, current evidence suggests a non-linear U- and J-shaped concentration-response (C-R) function. The potential increasing risks with decreasing O3 concentrations may seem counterintuitive from the traditional standpoint that decreasing exposure should lead to decreasing health risks. Tus, the question of whether the increasing risks with decreasing concentrations are truly O3-induced or might be from other C-R mechanisms. If these potential risks were not accounted for, this may have contributed to the risks observed at the low ozone concentration range. In this study, we examined the short-term effects of photochemical oxidant (Ox, parts per billiion) on outpatient cardiorespiratory visits in 21 Japanese cities after adjusting for other air pollutant-specific C-R functions. Daily cardiorespiratory visits from January 1, 2014 to December 31, 2016 were obtained from the Japanese Medical Data Center Co. Ltd. Similar period of meteorological and air pollution variables were obtained from relevant data sources. We utilized a time-stratified case crossover design coupled with the generalized additive mixed model (TSCC-GAMM) to estimate the association between Ox and cardiorespiratory outpatient visits, after adjusting for several covariates. A total of 2,588,930 visits were recorded across the study period, with a mean of 111.87 and a standard deviation of 138.75. The results revealed that crude Ox-cardiorespiratory visits exhibited a U-shaped pattern. However, adjustment of the oxides of nitrogen, particularly nitrogen monoxide (NO), attenuated the lower risk curve and subsequently altered the shape of the C-R function, with a substantial reduction observed during winter. NO- and nitrogen dioxide (NO2)-adjusted Ox-cardiorespiratory associations increased nearly linearly, without an apparent threshold. Current evidence suggests the importance of adjusting the oxides of nitrogen in estimating the Ox C-R risk functions.