Association of air pollution and mortality in individuals with high cardiovascular risk

Abstract

Background and Aim: Non-communicable diseases, e.g. cardiovascular diseases, are among the leading causes of mortality worldwide. Well-established risk scores, such as the ESC SCORE 2 (European Society of Cardiology Systematic COronary Risk Evaluation) are used to predict the onset of cardiovascular disease and mortality. However, they consider patient intrinsic risk factors, e.g. smoking but not extrinsic risk factors like air pollution. Yet, there is extensive evidence for an unfavourable associa tion between air pollution exposure and cardiovascular and other mortality. This dissertation aims to explore the time-dependent association between the extrinsic risk factor air pollution and mortality in a population with pre-existing cardiovascular risk factors in Hesse, Germany. Moreover, it explores the utility of considering air pollution and socioeconomic status in mortality risk prediction, compared to existing risk scores. Methods and results: Between 2010 and 2019, patients (N = 4610, 32% female, median age 69 years) scheduled for coronary angiography were enrolled in a prospective registry cohort at a certified German centre for cardiovascular research. Mortality was the outcome variable (1122 registered deaths). Air pollution data were retrieved from the Hessian State Agency for Nature Conservation, Environment and Geology (HLNUG) and assigned to patients according to postcode information. Air pollution markers of interest were: carbon monoxide (CO), nitrogen monoxide (NO), nitrogen dioxide (NO2), ozone (O3), particulate matter (PM10 and PM2.5 with an aerodynamic diameter of <10µm and <2.5µm, respectively) and sulphur dioxide (SO2). Receiver Operating Characteristic (ROC) analysis was conducted to identify the time period with the highest prognostic importance of air pollution exposure for its association with mortality within the range of 3 years prior to study enrolment. Exposure time periods with the highest prognostic ability varied for respective air pollution markers (CO [3 days], O3 [348 days], PM10 [232 days], PM2.5 [402 days], NO [1077 days], NO2 [1014 days], SO2 [1058 days]). Mean air pollution exposure concentrations for time periods of maximum predictive power were then each analysed in multivariate logistic regression models, with single air pollution markers. In these single air pollution marker models, adding NO, NO2, PM10, PM2.5, SO2 but not CO and O3 to a model including the ESC SCORE 2 variables, could significantly improve a models’ mortality prediction power. There was no significant benefit of adding purchasing power as an explanatory variable. Conclusion: The findings of this dissertation indicate that considering air pollution exposure in mortality risk prediction for individuals with cardiovascular burden is likely to enhance mortality risk estimation and should be subject to further research.