AS251 - Acute cardiovascular events and air pollution

[This page is intended to provide a study summary, the sections of which are below. Please complete these sections, as applicable. The headings below are suggested headings. You can remove inapplicable sections, or add new ones relevant to your study]

Investigator Names and Contact Information

Gregory A. Wellenius, ScD [ gwelleni@brown.edu ]

Abstract

The research described in this ancillary study proposal examines the cardiovascular disease (CVD)-air pollution association.  Its focus is on the population burden of acute CVD events attributable to air pollution as well as methodological innovations aimed at improving its estimation and informing policy.  It contrasts sudden death with other CVD outcomes, emphasizes particulate matter (PM) exposures, and addresses the error in causal inference that occurs when aggregate- and individual-level associations are assumed to be identical within the same population.  In studies of PM—including smoke, soot, and condensed or transformed gases directly emitted into the air by motor vehicles, power plants and factories—this ecologic fallacy reflects the acknowledged inability of aggregate PM mass concentration data to capture the heterogeneity of pertinent exposures between individuals within groups.  Recent studies suggest that this heterogeneity may be related to inter-individual differences in small-scale geographic, socioeconomic and / or environmental exposures, e.g. to traffic-related air pollutants and other source-specific chemical components of PM.  From a public health perspective, a better understanding of these sources of and methods for reducing ecologic bias in studies of incident CVD is therefore required, especially if National Ambient Air Quality Standards (NAAQS) are inappropriately high.  To this end, we will leverage the substantial data resources that we have assembled; the geocoding, kriging, and error modeling methods that we have carefully validated; and the corresponding error estimates that we have generated in AS #140,"The Environmental Epidemiology of Arrhythmogenesis in WHI" (Whitsel [PI]).  We will estimate exposures of WHI CT participants to traffic-related air pollutants and other source-specific chemical components of PM, examine the associations of these exposures with incident, validated CVD events, and describe the effects of exposure measurement errors on the magnitude and precision of the former associations.  Errors to be evaluated include those associated with using [1] commercially geocoded (vs. measured) address coordinates, [2] ambient (vs. personal) exposures, and [3] spatially interpolated (vs. monitored) exposures.  Collectively, the analyses will endow WHI with a uniformly standardized and validated set of environmental exposure data.  Moreover, they will improve understanding of the cardiovascular disease-air pollution association, its effect modifiers, and sources of / methods for reducing important biases overlooked in prior studies.  Such understanding will be useful to investigators broadly concerned about the impact of exposure measurement error on exposure-outcome associations in multi-level studies, and more specifically, to epidemiologists,  environmental health scientists, and federal regulators responsible for accurately evaluating air quality standards in terms of their ability to protect the cardiovascular health of women living in the U.S.

Specific Aims

The research described in this ancillary study proposal examines the cardiovascular disease (CVD)-air pollution association.  Its focus is on the population burden of acute CVD events attributable to air pollution as well as methodological innovations aimed at improving its estimation and informing policy.  The study's specific aims are to

1. Estimate the following measures of air pollutant burden from baseline through follow-up in the WHI CT
1. Measures of traffic-related air pollutant exposure
2. Spatially interpolated, residence- and exam site-specific, daily mean concentrations of the major PM chemical components including coal, diesel and gasoline combustion species
2. Examine associations between incident, validated, acute CVD events and the above exposures, contrasting
1. Sudden death with other CVD outcomes
2. Acute, independent and joint effects of the above exposures and those of PM mass concentrations already estimated by AS # 140
3. Putative clinical, geographic, socioeconomic, climatic and environmental effect modifiers, including chronic exposures averaged over 30 & 365 days
3. Assemble full probability distributions of pertinent errors, i.e. those associated with using
1. Commercially geocoded (vs. measured) residence and exam site address coordinates
2. Ambient (vs. personal) PM exposures
3. Spatially interpolated (vs. monitored) PM concentrations
4. Evaluate in a single statistical model how the observed CVD-PM associations are influenced by errors associated with the prior sources of uncertainty, using the most comprehensive and flexible methods available
1. Bayesian hierarchical modeling
2. Clustering of highly correlated exposures
3. Markov chain Monte Carlo (MCMC) sampling
5. Assess the reliability of findings in the more representative, but less well-characterized WHI OS population
6. And to thereby
1. Endow WHI with a uniformly standardized and validated set of environmental exposure data
2. Improve understanding of the CVD-air pollution association, its effect modifiers, and sources of / methods for reducing important biases overlooked in prior studies

Inform epidemiologists, environmental health scientists, and federal regulators responsible for accurately evaluating air quality standards in terms of their ability to protect the cardiovascular health of women living in the U.S.