Cancer and Mortality in Relation to Traffic-related Air Pollution among Coronary Artery Disease Patients: Using an Ensemble of Exposure Estimates to Identify High-risk Individuals
2Dept. of Statistics and Operations Research, School of Mathematical Sciences, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University
3Technion Center of Excellence in Exposure Science and Environmental Health, Technion Israel Institute of Technology
4Epidemiology Unit, Braun School of Public Health and Community Medicine, Hebrew University and Hadassah Medical Organization
5Dept. of Geography, Hebrew University of Jerusalem
6Dept. of Cardiology, Rabin Medical Center (Beilinson and Hasharon Hospitals)
7Dept. of Cardiovascular Medicine, Sackler Faculty of Medicine, Tel Aviv University
Background: Traffic-related air pollution (TRAP) has been associated with cancer and mortality, yet the estimated health effect is small. Considering the limited agreement across exposure modeling approaches, misclassification is probable. We used an ensemble of modeling approaches to reevaluate the strength of the association in a large cohort of coronary patients.
Methods: The cohort comprises all patients undergoing percutaneous coronary interventions in a major medical center in Israel from 2004-2014. Subjects were followed for cancer until 2015 (median, 7 years) through the National Cancer Registry and death until 2017 (median, 9 years) through the Population Registry. Annual residential levels of nitrogen oxides (NOx) –a proxy for TRAP– were estimated by two high-resolution modeling approaches: optimized dispersion model (ODM) and land use regression (LUR). We created mutually exclusive groups of subjects classified as exposed to NOx concentrations of ≥25 ppb by none of the methods, ODM alone, LUR alone or both methods. We then repeated the analysis for more extreme cutoffs. Cox models were constructed to study relationships with cancer and mortality.
Results: Among 12,784 candidate patients, 10,627 (mean age, 69 years; 76% men) had both ODM and LUR NOx estimates (rP=0.50). During follow-up, 741 (8%) incident cancer cases were diagnosed and 3,051 (28%) deaths occurred. In adjusted models, associations of the ensemble estimator with outcomes were considerably stronger than those obtained by using each method separately, particularly for the composite outcome of lung, breast and prostate cancer previously linked to TRAP (Figure). Similar patterns, albeit less striking, were observed for mortality.
Conclusion: Using an ensemble of TRAP exposure estimates yields strong association with various cancers and mortality. This approach, if further confirmed, may alter patient, clinician and policy-maker perspectives on health threat of air pollution exposure.
Powered by Eventact EMS