External exposome and all-cause mortality in two large administrative cohorts: Rome (Italy) and Greece

Abstract

Background and Aim Many studies reported associations between long-term exposure to several environmental stressors and mortality, however little is known on the complex interplay between the urban exposome and health. We characterized the external exposome through a limited number of exposure domains and investigated their association with mortality in two large cohorts (Rome and Greece). Methods Within the EXPANSE project, we analyzed two population-based cohorts: Rome (Italy) (30+ years; 2011-2019) and Greece (37+; 2014-2019). Multiple exposures were assigned at the residential addresses, and were divided into three a-priori defined domains: 1) land-built environment (NDVI, imperviousness surfaces, distance to water); 2) air pollution (fine particulate matter (PM₂.₅), nitrogen dioxide (NO₂), black carbon (BC) and warm-season ozone (warm-O₃)); 4) temperature (annual, cold-season and warm-season mean and standard deviation (SD)). Each domain was synthesized through Principal Component Analysis (PCA). We applied Cox proportional hazards regression models using the obtained principal components as exposures. Results In each cohort and for each domain, we selected the first principal component. It explained more than 55% of variance, ranging from 57.4% (land-built environment for Rome) to 83.3% (air pollution for Rome). We estimated a protective effect between the land-built environment domain and all-cause mortality, with hazard ratios (HR) of 0.984 (95%CI: 0.977-0.991) in Rome and 0.980 (0.976-0.984) in Greece, per interquartile range increments. Associations with the air pollution component were positive and statistically significant, with HR of 1.011 (1.005-1.018) and 1.060 (1.051-1.068), respectively. The air temperature domain was associated with mortality only in the Greek cohort. Results were similar in multi-component analyses. Conclusions We found associations between different components of the external exposome and mortality in two large administrative cohorts. The same approach will be applied in the other administrative cohorts of EXPANSE: Switzerland, Sweden, Netherlands and Catalunya (Spain). Keywords: exposome, population-based cohort, principal component analysis.