You are here

Optimizing exposure assessment for inference about air pollution effects with application to the aging brain

Principal Investigator: 

University of Washington

This study will compare and contrast scientific and logistical benefits of different approaches to air pollution exposure assessment. The investigators will leverage large air pollution datasets obtained from low-cost sensors, mobile monitoring, and passive samplers. They will apply the exposure assessment approaches to determine associations with cognitive decline and dementia incidence in an ongoing cohort study, Adult Changes in Thought Air Pollution (ACT-AP).

Funded under

Poster abstract for HEI Annual Conference 2022

Evaluation of the Impact of Mobile Monitoring Network Design on the Quality of Air Pollution Exposure Prediction Models 

Magali N. Blanco,1 Elena Austin,1 Edmund Seto,1 Timothy Larson,1,2 Julian Marshall,2 Lianne Sheppard,1,3

1Department of Environmental and Occupational Health Sciences, School of Public Health, University of Washington, Seattle, WA, USA; 2Department of Civil & Environmental Engineering, College of Engineering, University of Washington, Seattle, WA, USA; Department of Biostatistics, School of Public Health, University of Washington, Seattle, WA, USA

Introduction. Short-term mobile monitoring campaigns are increasingly being used to assess long-term air pollution exposure in epidemiologic investigations. However, most collect limited data and produce moderate or poorly performing prediction models.

Methods. To address this gap, we leverage an extensive, multi-pollutant mobile monitoring campaign in the greater Seattle area intended to assess annual average air pollution exposure levels in an epidemiologic cohort. The campaign consisted of 309 stationary sites, each with approximately 26 temporally balanced visits per site, and a complete set of measurements of ultrafine particulate (UFP), black carbon (BC), nitrogen dioxide (NO2), fine particulate matter (PM2.5) and carbon dioxide (CO2). We use Monte Carlo simulations to subsample data and investigate the quality of the resulting exposure prediction models when we select a different number of monitoring sites, repeat site visits, and sampling periods. We further explore the spatial extrapolation limits of these models as the geographic distance and covariate differences between the monitoring and prediction locations increase.

Results. While there were some differences across pollutants, model performance generally increased with a greater number of campaign visits and when samples were collected using temporally-balanced approaches including all seasons, days of the week, and hours of the day. The performance of these models was maximized when the monitoring and prediction sites were as close as possible and had similar geographic covariate characteristics.

Conclusions. Short-term mobile monitoring campaigns can improve the performance of annual average exposure prediction models by increasing the total number of visits and by conducting temporally balanced sampling. The spatial compatibility between the monitoring and cohort locations of interest as well as the spatial and temporal characteristics of the pollutants of interest should additionally be considered.