Contribution of the in-vehicle microenvironment to individual ambient source-derived NO2 exposure concentration

Abstract

In-vehicle exposures to traffic-related air pollution (TRAP) may be an important part of an individual's overall exposure to pollutants. Prior epidemiological and exposure assessment studies used outdoor NO₂ concentrations as a surrogate for individual TRAP exposures and generally did not explicitly account for in-vehicle exposures. In this study, we assessed the contribution of the in-vehicle microenvironment to an individual's ambient source-derived NO₂ exposure concentration in a subset of MESA Air participants residing in Winston-Salem, NC (n=46). We measured time-integrated microenvironmental (residential indoor, residential outdoor, and in-vehicle) and personal NO₂ concentrations as surrogates of TRAP during a winter sampling campaign; time-location data was collected through a time-location diary. For our primary aim, we calculated microenvironmental contributions to each individual's ambient source-derived NO₂ exposure concentration using a time-weighted model. Secondarily, we evaluated the correlations between microenvironmental NO₂ concentrations and the ambient source-derived individual NO₂ exposure concentration using Pearson correlation coefficients. Finally, we attempted to predict in-vehicle NO₂ masses using the time spent in different travel conditions using multiple linear regression in a person-level analysis, whose model was evaluated using leave-one-out cross-validation. Data for analysis was available on 38 participants, who spent an average of one hour per day in-vehicle. The in-vehicle microenvironment was a major contributor to each individual's ambient source-derived NO₂ exposure concentration, accounting for an estimated 48.9±10.4%. Indoor ambient source-derived and outdoor NO2 concentrations showed stronger correlations (r=0.73) with each individual's ambient source-derived NO₂ exposure concentration than in-vehicle NO₂ concentrations (r=0.44). The time spent in different travel conditions explained 44% of the in-sample variation within in-vehicle NO₂ masses with a RMSE of 0.25 ppb, and demonstrated good predictive value when tested out of sample (CV R<super>2</super>=0.42; CV RMSE=0.15 ppb); the number of trips driven during the two-week sampling period was the most important predictor of in-vehicle NO2 masses (partial R<super>2</super>=0.33). These results indicate a high contribution of the in-vehicle microenvironment to individual ambient source-derived NO₂ exposure concentration, suggesting the need to account for in-vehicle exposures in studies of traffic-related air pollution.