Comparison of PM2.5 Air Pollution Exposures and Health Effects Associations Using 11 Different Modeling Approaches in the Women’s Health Initiative Memory Study (WHIMS)

Background: Many approaches to quantifying air pollution exposures have been developed. However, the impact of choice of approach on air pollution estimates and health-effects associations remains unclear. Objectives: Our objective is to compare particulate matter with aerodynamic diameter ≤2.5μm (PM2.5) concentrations and resulting health effects associations using multiple estimation approaches previously used in epidemiologic analyses. Methods: We assigned annual PM2.5 exposure estimates from 1999 to 2004 derived from 11 different approaches to Women’s Health Initiative Memory Study (WHIMS) participant addresses within the contiguous US. Approaches included geostatistical interpolation approaches, land-use regression or spatiotemporal models, satellite-derived approaches, air dispersion and chemical transport models, and hybrid models. We used descriptive statistics and plots to assess relative and absolute agreement among exposure estimates and examined the impact of approach on associations between PM2.5 and death due to natural causes, cardiovascular disease (CVD) mortality, and incident CVD events, adjusting for individual-level covariates and climate-based region. Results: With a few exceptions, relative agreement of approach-specific PM2.5 exposure estimates was high for PM2.5 concentrations across the contiguous US. Agreement among approach-specific exposure estimates was stronger near PM2.5 monitors, in certain regions of the country, and in 2004 vs. 1999. Collectively, our results suggest but do not quantify lower agreement at local spatial scales for PM2.5. There was no evidence of large differences in health effects associations with PM2.5 among estimation approaches in analyses adjusted for climate region. Conclusions: Different estimation approaches produced similar spatial patterns of PM2.5 concentrations across the contiguous US and in areas with dense monitoring data, and PM2.5-health effects associations were similar among estimation approaches. PM2.5 estimates and PM2.5-health effects associations may differ more in samples drawn from smaller areas or areas without substantial monitoring data, or in analyses with finer adjustment for participant location. Our results can inform decisions about PM2.5 estimation approach in epidemiologic studies, as investigators balance concerns about bias, efficiency, and resource allocation. Future work is needed to understand whether these conclusions also apply in the context of other air pollutants of interest. https://doi.org/10.1289/EHP12995


Table of Contents
. Summary statistics of estimated annual PM 2.5 exposures (in ug/m 3 ) at WHIMS participant addresses, 1999 and 2004.Table S2.Results of linear mixed models identifying factors that influence absolute agreement of annual PM2.5 exposure concentrations across estimation approaches, 1999 and 2004.
Table S3.Results of linear mixed models identifying factors that influence relative agreement of annual PM2.5 exposure concentrations across estimation approaches, 1999 to 2004.Table S4.Summary statistics of estimated annual PM2.5 exposures (in ug/m 3 ) at WHIMS participant addresses, 1999 and 2004, by climate region grouping.
Table S5.Summary statistics of estimated annual PM2.5 exposures (in ug/m 3 ) at WHIMS participant addresses, 1999 and 2004, by distance to road, distance to air pollution monitor, development, greenspace, and urbanicity.Table S6.Pairwise percent agreement across contextual factors in 1999 and 2004.Table S7.Hazard ratios and 95% CIs for the association between a 10 ug/m3 increase in mean PM2.5 exposure and death due to natural causes, CVD-related mortality, and incident CVD events from 2005-2021 in WHIMS using 11 different estimation approaches.
Table S8.Hazard ratios and 95% CIs for the association between a 10 ug/m3 increase in mean PM2.5 exposure and CVD-related mortality from 2005-2021 in WHIMS participants using 11 different estimation approaches by region.
Table S9.Hazard ratios and 95% CIs for the association between a 10 ug/m3 increase in mean PM2.5 exposure and CVD-related mortality from 2005-2021 in WHIMS using 11 different estimation approaches stratified by contextual factors.

Figure S2 .
Figure S2.Heatmaps of pairwise mean bias and pairwise mean error in 1999 and 2004 for each pairwise combination of PM2.5 approaches at WHIMS participant addresses.

Figure S3 .
Figure S3.Boxplots and density plots of PM2.5 exposure distributions at WHIMS participant addresses using 11 different estimation approaches in 1999 by region.

Figure S4 .
Figure S4.Boxplots and density plots of PM2.5 exposure distributions at WHIMS participant addresses using 11 different estimation approaches in 2004 by region.

Figure S5 .
Figure S5.Heatmaps of pairwise mean bias in 1999 for each pairwise combination of PM2.5 approaches at WHIMS participant addresses by region.

Figure S6 .
Figure S6.Heatmaps of pairwise mean error in 1999 for each pairwise combination of PM2.5 approaches at WHIMS participant addresses by region.

Figure S7 .
Figure S7.Heatmaps of pairwise mean bias in 2004 for each pairwise combination of PM2.5 approaches at WHIMS participant addresses by region.

Figure S8 .
Figure S8.Heatmaps of pairwise mean error in 2004 for each pairwise combination of PM2.5 approaches at WHIMS participant addresses by region.

Figure S9 .
Figure S9.Boxplots and density plots of PM2.5 exposure distributions at WHIMS participant addresses using 11 different estimation approaches in 1999 (Panel A) and 2004 (Panel B) by distance to a PM2.5 air pollution monitor.

Figure S10 .
Figure S10.Heatmaps of pairwise mean bias in 1999 and 2004 for each pairwise combination of PM2.5 approaches at WHIMS participant addresses by distance to a PM2.5 air pollution monitor.

Figure S11 .
Figure S11.Heatmaps of pairwise mean error in 1999 and 2004 for each pairwise combination of PM2.5 approaches at WHIMS participant addresses by distance to a PM2.5 air pollution monitor.

Figure S12 .
Figure S12.Heatmaps of pairwise spatial R 2 in 1999 and 2004 for each pairwise combination of PM2.5 approaches at WHIMS participant addresses by distance to a major road.

Figure S13 .
Figure S13.Hazard ratios and 95% CIs for the association between a 10 ug/m 3 increase in mean PM2.5 exposure and death due to natural causes, CVD-related mortality, and incident CVD events from 2005-2021 in WHIMS using 11 different estimation approaches.

Figure S14 .
Figure S14.Hazard ratios and 95% CIs for the association between a 10 ug/m 3 increase in 1999-2004 mean PM2.5 exposure and CVD-related mortality from 2005-2021 in WHIMS using 11 different estimation approaches, stratified by contextual factors.