Exploring methods of air pollution exposure and intake in active populations

Abstract

Ambient air pollution epidemiological studies are based on limited data on outdoor concentration at residential address neglecting people’s daily activities and how these influence the actual concentrations they are exposed to or how much of a pollutant is inhaled. The research presented here deals with such omission exploring methodologies that account for people’s activity patterns to model personal exposure concentrations and daily inhaled doses of air pollutants.

Three models are presented: First, a model based on call detail records (CDRs) from 8 million callers in Italy is used to identify activity-based nitrogen dioxide (NO2) hotspots. A second model using geolocation, accelerometer data, and activity diaries from 180 people in Barcelona is presented and percentage contributions to NO2 and particulate matter (PM2.5) daily personal exposure and inhaled doses from different microenvironments are computed. Measured black carbon (BC) exposure data from 122 participants living in Antwerp, Barcelona and London is then used to generate the third and last model based on linear regressions.

Population-activity-based exposures from the first model are lower than those based on home location. The method does identify areas of relative importance due to the time spent by people in them. The second model shows that despite people spending 10% of their day in transport, transport microenvironments contribute with 25% of participants daily exposures and with 45% of their daily inhaled doses. Measured BC exposures showed a similar trend with people spending 7.5% of their time in transport, which in turn contributed with 10.5% of daily BC exposures and 18.1% of daily inhaled doses. Linear regression models had an acceptable performance with adjusted R2 values for BC exposures of 0.4 and for BC inhaled doses of 0.35.

The results offer new evidence on the importance of including activity patterns and ventilation rates in future epidemiological studies to avoid biases.