Spatial-temporal Analysis of Particulate Exposure in Those Experiencing Homelessness Compared to the Housed Population in Seattle, WA
Date:
Presented at the American Public Health Association (APHA) 2026 Annual Meeting and Expo, Monday, November 2, 2026, 8:30–10:00am. Program: Environment; Session: Environmental Justice and Health Equity.
Authors: Yehong Deng, Allen J. Gassett, Jianzhao Bi, Aja Sutton, Joel D. Kaufman, Abraham D. Flaxman, Amy Hagopian, and Zack W. Almquist.
Abstract: People experiencing homelessness face disproportionate environmental health risks, yet the sustained health burden of air pollution exposure for this population remains critically unquantified. This study provides the first direct quantification of PM2.5 exposure disparities and attributable disease burden across housed, sheltered, and unsheltered populations within a single U.S. city. Crucially, we further stratify the housed population by neighborhood income to isolate the effect of housing status from broader socioeconomic gradients. Using geocoded data for 675 tent encampments, 125 emergency shelter sites, and 716 residential tax lots in Seattle, Washington—each linked to a validated spatio-temporal PM2.5 prediction model—we document that unsheltered residents face systematically higher mean annual PM2.5 concentrations than housed residents (6.39 vs. 5.84 µg/m³, a 9.4% gap) across all six study years (2016–2021). Marked Poisson point process analysis confirms this exposure gap reflects structural spatial segregation, with tent encampments clustering disproportionately in Seattle’s high-pollution central corridor (segregation test p < 0.005).
We translate 32-year cumulative exposure histories (1990–2021) into Disability-Adjusted Life Years (DALYs), a metric capturing total disease burden through years lost to premature death and years lived with disability. To rigorously propagate uncertainty in the concentration-health relationship across six PM2.5-attributable diseases, we employ 1,000-draw Monte Carlo simulations integrating IHME Burden of Proof relative risk curves and GBD baseline rates. To disentangle housing status from socioeconomic position, we stratify housed tax lots into quartiles by census tract median household income. We find that Unsheltered (tent) individuals accumulate approximately 11% more PM2.5-attributable DALYs than the highest-income housed residents (Q4) and approximately 6% more than the lowest-income housed residents (Q1). Sheltered individuals accumulate approximately 12% more than Q4 and 7% more than Q1. These findings establish housing insecurity as a distinct, severe axis of environmental injustice that cannot be explained by neighborhood income alone.