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).
Optimizing exposure assessment for inference about air pollution effects with application to the aging brain
Lianne Sheppard1, Ali Shojaie1, Adam Szpiro1
1University of Washington, Seattle, WA, United States
Background: Although adequate exposure assessment is fundamental to understanding the health impact of environmental exposures, limited work has been done to determine the optimal exposure assessment approaches. Thanks to recent funding from HEI, we plan to improve the inferential strength of air pollution cohort studies by expanding the diversity of high-quality exposure metrics available and assessing their value in the context of existing exposure metrics. We will accomplish this goal by leveraging the exposure assessment activities in the ongoing NIH-funded Adult Changes in Thought Air Pollution (ACT-AP) cohort study in order to harness its two novel measurement technologies (low-cost sensors and mobile monitoring) and improve exposure assessment modeling approaches. Our overall objective is to compare and contrast the scientific and logistic features of different exposure measurements and sampling designs for air pollution exposure assessment in environmental epidemiology.
Methods: Our broad objective is to optimize estimates of long-term average outdoor air pollution exposures to use in cohort study inference, focusing on 1) criteria air pollutants typically measured at multiple locations in a study region (fine particles (PM2.5) and nitrogen dioxide (NO2) and 2) other pollutants often unavailable in cohort studies (black carbon (BC), ultrafine particles (UFP), oxides of nitrogen (NOx), ozone (O3), carbon monoxide (CO), and multi-pollutant mixtures).
Our specific aims are:
Aim 1: Identify key design choices to improve long-term average exposure prediction using 1a) mobile monitoring campaigns and 1b) stationary networks of low-cost sensors and/or Ogawa samplers.
Aim 2: Develop annual average TRAP exposure predictions from a mobile platform using novel statistical methods not previously employed, including 2a) spatial ensemble methods, 2b) predictions that leverage road network information combined with spatial proximity, and 2c) multi-pollutant versions.
Aim 3: Determine the impact on inference of different predictions based on sampling designs and analysis approaches. We will focus on cognitive decline and dementia incidence.
Aim 4: Address the overall value of incorporating novel exposure data and modeling by comparing and contrasting both their impact on inference and the logistical features (cost and time) of using different sampling designs and analysis choices.
Conclusions: This project will advance the state of exposure science for epidemiology by making recommendations to researchers as to how to optimize data collection and prediction in their air pollution exposure assessments for inference about health effects in cohort studies. To our knowledge there has never been funding for this type of detailed investigation into how different designs compare and what features of exposure sampling campaigns are critical to the success of an epidemiologic study. Our focus on brain health and less commonly studied pollutants has direct policy relevance; these important policy implications are magnified given the Puget Sound’s relatively low levels of air pollution.