Low-level air pollution and natural cause mortality in Rome: comparison of results based on European wide and local land use regression models within the ELAPSE project

Abstract

PDS 65: Exposure assessment: implications for epidemiology, Exhibition Hall (PDS), Ground floor, August 27, 2019, 1:30 PM - 3:00 PM Background: The multicentre “Effects of Low-Level Air Pollution: A Study in Europe (ELAPSE)” study has been designed to collect data from multiple European cohorts and apply standardized exposure models. Alternative exposure assessment based on local models serves as a comparison. We evaluated associations with mortality of annual average air pollution exposure based both on a Europe-wide hybrid Land Use Regression (EW-LUR) model and on local LUR (L-LUR) models in the Rome Longitudinal Study (RoLS). Methods: We selected >1 million subjects aged 30+ years and followed them up from 2001 to 2015 through record linkage with mortality registries. Annual average concentrations of PM2.5 and NO2 were centrally modelled within ELAPSE at 100 m spatial resolution (2010). Local LUR models were developed for PM2.5 (ESCAPE, 2010) and for NO2 (2007). We applied Cox proportional hazard models considering age (time axis), sex (strata) and several individual and contextual variables. We evaluated the effects of each pollutant through single and two-pollutant models. Results: We observed 235,543 deaths from natural causes out of 15,300,400 person-years. Average exposures were 16.7 μg/m3 and 32.9 μg/m3 for PM2.5 and NO2, respectively for EW-LUR, and 19.6 μg/m3 and 43.4 μg/m3 for L-LUR. Pairwise correlations (PM2.5 and NO2) were 0.76 in EW-LUR and 0.42 in L-LUR. We found significant associations between PM2.5 and NO2 and mortality for both exposure models, with hazard ratios (HRs) =1.07 (95% CI: 1.03, 1.10) and 1.03 (1.02, 1. 04) for PM2.5 and NO2 from EW-LUR, and 1.02 (1.01, 1.03) and 1.02 (1.02, 1.03) for exposures from L-LUR (per increments of 5 and 10 μg/m3), respectively. PM2.5 estimates decreased and lost significance in two-pollutant models while NO2 estimates were robust to PM2.5 adjustment under both exposure conditions. Conclusions: Long-term exposure to air pollutants (PM2.5, and NO2) was associated with natural mortality using two different exposure models.