You are here

Accounting for mobility in air pollution exposure estimates in studies on long-term health effects

Principal Investigator: 

Swiss Tropical and Public Health Institute and Basel University, Switzerland

This study aims to improve our understanding of the contribution of individual mobility in air pollution exposure estimates. The investigators will use location tracking on a mobile phone application for 2,000 individuals in the Netherlands and Switzerland.

Funded under

Poster abstract for HEI Annual Conference 2022

MOBI-AIR: Accounting for MOBility in AIR pollution exposure estimates in studies on long-term health effects

Kees de Hoogh1,2, Nicole Probst-Hensch1,2, Danielle Vienneau1,2, Ayoung Jeoung1,2 Benjamin Fluekiger1,2, Gerard Hoek3, Roel Vermeulen3, Kalliopi Kyriakou3, Derek Karssenberg4, Meng Lu5, Oliver Schmitz4 

1Swiss Tropical and Public Health Institute, Basel, Switzerland; 2Basel University, Basel, Switzerland3; Institute of Risk Assessment Sciences, Utrecht University, Utrecht, the Netherlands; 4Department of Physical Geography, Utrecht University, Utrecht, the Netherlands; 5Department of Geography, University of Bayreuth, Bayreuth, Germany

Background. Large scale epidemiological studies investigating long-term health effects of air pollution typically only consider the residential locations of the participants, ignoring the space-time activity patterns that likely influence total exposure. Neglecting mobility patterns in exposure assessment may lead to incorrect exposure distributions and bias in downstream exposure response relations. The main aim of this study is to assess whether more sophisticated estimates of individual exposure, considering population mobility, decreases the bias in health studies.

Methods. We developed a modelling framework combining the space-time mapping of pollution and activity-based mobility simulation of individuals. Our framework contains: a) an activity schedule generator to express the type, location and duration of an individual's activity as a function of their profile defined by e.g. age, gender or occupation, b) a spatial context generator ― providing the location (home, work, leisure) or travel mode (car, bicycle, on foot) of an individual during a particular activity, and c) an exposure estimator ― combining the spatial contexts for each activity with air pollution concentrations at corresponding times. The modelling framework will be applied for exposure assessment of large cohorts in the Netherlands and Switzerland.

Results. The modelling framework was applied to 3000 individuals of the SAPALDIA cohort, Switzerland. The simulation assumed a regular daily activity schedule, and for commuters calculated routes between residential and work locations (known for most individuals in the cohort). The diurnal trend of exposure to NO2 was calculated for each individual in the cohort. For most commuters, exposures were highest during commuting corresponding with high estimated NO2 concentrations during rush hours. Compared to exposure assigned from residential address alone, exposures calculated with our modelling framework show somewhat less contrast between individuals.

Conclusions. The activity-based mobility simulation provides a relatively detailed representation of space-time activity of individuals. Simulations can be executed with acceptable runtime, which is important for future application to very large cohorts. The approach can be modified based on the availability of individual level data; if work locations are not known, these can be randomly drawn from typical work locations in a Monte Carlo setup. In the next phase of the project we will use observed tracks (from GPS devices) to evaluate the performance of our simulations.