AS552 - Harmonized and pooled approaches in multiple cohorts to understand low concentration health impacts of air pollutants

Investigator Names and Contact Information

Joel Kaufman (joelk@uw.edu)

Introduction/Intent

Long-term exposure to air pollutants has been consistently associated with adverse health outcomes, especially cardiovascular events, in several cohorts, including the WHI-OS. There is reason to believe that air pollution health effects may exist at low-levels, including below current regulatory standards. However, substantial uncertainty exists about the concentration-response function at low-levels. The study of low-level air pollution health effects in observational cohorts is challenging, requiring not only a study of adequate size, but also exposure heterogeneity within the range of interest and extensive individual confounder information. Moreover, complex exposure assessment methods may be required to estimate individual exposures in such settings. An efficient approach to overcome such issues is to pool data from large well-characterized cohorts and apply a unified state-of-the-art exposure assessment approach. In this proposal, we propose to pool information from a set of well-established cohorts in the US, including the WHI OS and CT cohorts, each with appropriate outcome, home address, and individual confounder level detail, to create an amply powered study. These approaches, combined with state-of-the-art fine-scale hybrid modeling of air pollutant concentrations at geocoded participant locations, will provide information to robustly address the research question. This proposal is responsive to a request for proposals from the Health Effects Institute.

SPECIFIC AIMS

Aim 1: To assess the effects of long-term exposure to ambient air pollutants (PM2.5, NO2, O3) on cardiovascular diseases and mortality in a pooled-analysis of well-characterized epidemiologic cohorts in the United States. We will further develop and apply our advanced spatio-temporal models of air pollutant concentrations from 1990-2014 which utilize regionalized universal kriging incorporating a large suite of geographic covariates via partial least squares regression, with inputs from both satellite remote-sensing data and chemical transport modeling.

• Aim 1a: Using harmonized outcome and covariate approaches in each cohort, and pooling cohort data, we will estimate the effect of low-levels of exposure to PM2.5(e.g, concentrations < 12 μg/m3), NO2, and O3 on incident cardiovascular disease events, including fatal and nonfatal myocardial infarction, stroke, congestive heart failure, sudden cardiac arrest from high-quality outcome data. Analyses are adjusted for individual-level potential confounders.
• Aim 1b: Using harmonized outcome and covariate approaches in each cohort, and pooling cohort data, we will estimate the effect of low-levels of exposure to PM2.5, NO2, and O3 on all-cause mortality, non-accidental mortality, and cause-specific mortality (especially cardiovascular disease mortality). Analyses adjusted for individual-level potential confounders.
• Aim 1c: To characterize the concentration-response function between long-term exposures to air pollutants and incident disease and death, with specific attention to low-levels of exposure.
• Aim 1d: To estimate single and multi-pollutant relationships of long-term exposure to air pollutants and incident non-accidental death, cause-specific death, and clinical events.

Aim 2: To correct single and multi-pollutant models for measurement error incurred by using predicted concentrations in place of true exposures in all health analyses.