Effect of Confounding, Effect Modification, and Exposure Measures on the Association of Long-Term Ozone Exposure and Cause-Specific Mortality

Abstract

Introduction: Associations between long-term ozone (O3) exposure and cause-specific mortality have been demonstrated, although with varying exposure metrics and confounders.Methods: We assessed the association of warm season (April-September) averages of 1-hour daily maximum O3 and cause-specific mortality for 22.2 million Medicare beneficiaries between 2000-2008. We modeled the national and regional associations between O3 and rate of mortality using age, gender, and race stratified log-linear regression models. We examined confounding by including ecologic and neighborhood-level behavioral covariates and PM2.5 estimated using spatio-temporal models. We also performed sensitivity analyses using exposure assessed as the warm season averages of daily 8-hour maximum and 24-hour average concentrations. We fit a log-linear model with a restricted cubic spline of O3 to estimate the shape of the dose-response relationship between ozone and cause-specific mortality and examined effect modification of these associations by age, gender and race.Results: Our findings show that 1-hour maximum O3 is significantly associated with an increased rate of death for all-cause (1.013 per 10 ppb increase; 95%CI: 1.012, 1.014), respiratory (1.036; 95%CI: 1.032, 1.039), cardiovascular (1.027; 95%CI: 1.025, 1.028) and lung cancer (1.016; 95%CI:1.011, 1.020) mortality, but not for cancer mortality. Associations were generally strongest in the northeast US for all causes as compared to other US regions. Risk ratios were robust to adjustment for ecologic covariates, behavioral covariates and to PM2.5. Results remained significant but were attenuated when 8-hour daily maximum and 24-hour average measures of O3 were used as the exposure measure, although 24-h averaged O3 was associated with decreased all cause and cardiovascular mortality.Conclusions: Long-term exposure to O3 was associated with increased mortality due to all-cause, respiratory, cardiovascular and lung cancer mortality.