Abstract
BACKGROUND: The understanding of the complex interactions among ozone, temperature, various atmospheric conditions, and the intricate mixtures that make up the air exposome remains limited. Whether exposure to ozone concentrations consistently below current U.S. regulatory limits (70 ppb, 8-hour average) is associated with newly-diagnosed chronic obstructive pulmonary disease (COPD) is unclear, especially among demographic subgroups. We examined associations between residential ozone and incident COPD in a large prospective cohort study in the U.S., and assessed heterogeneity by demographic subgroups. METHODS: The All of Us research program followed >848,000 volunteers enrolled in 2017-2023. Among 596, 926 participants whom consented to the release of EHR data, annual average ozone concentrations from satellite measurements from 2000-2016 were linked to residential location for 376,535 participants. Multivariable Cox regression was used to estimate associations between ozone and incident COPD over the study period, adjusting for co-exposures and potential confounders. We assessed effect modification using cross-product terms and stratified analyses with race, sex, age, income, and smoking status. To account for non-probabilistic sampling, we applied cell weighting based on the distribution of key factors from the 2018-2022 American Community Survey. RESULTS: We identified 7,907 incident COPD cases over an average 4-year follow-up. Residential ozone concentrations (Median: 37.75 ppb, Min-Max: 29.79-51.39) were similar across subgroups. Overall, we observed a positive, non-monotonic relationship between ozone and COPD risk, driven by the highest quartile of exposure (HR (Quartile 4vs1) =1.09, 95% CI: 1.01, 1.18). We found evidence of heterogeneity among different subgroups. Importantly, the exposure-response relationship was monotonic and most apparent in male never-smokers (P-trend (continuous) =6.62×10 (-6) ). CONCLUSION: Even consistently below U.S. regulatory limits, the respiratory effects of ozone were apparent independent of smoking. Our findings provide greater precision into which subgroups might be more susceptible.