Effects of fine particulate matter from wildfire and non-wildfire sources on emergency-department visits in people who were housed and unhoused in San Diego County (CA, USA) during 2012–20: a time-stratified case–crossover study

Background

Being unhoused can increase vulnerability to adverse health effects due to air pollution. We aimed to quantify changes in emergency-department visits during and after exposure to wildfire-specific and non-wildfire particulate matter 2·5 μm or less in diameter (PM_2·5) in San Diego County (CA, USA) in people who were both unhoused and housed.

Methods

For this time-stratified case–crossover study, we used data on exposure to wildfire-specific PM_2·5 in California and individual-level data for people admitted to the emergency departments of two hospitals (UC San Diego Health emergency departments at La Jolla and Hillcrest, San Diego) in San Diego County between July 1, 2012, and Dec 31, 2020. People with a postcode outside of San Diego County were excluded. Demographic information was age group, race or ethnicity, and transport to the emergency department. Wildfire-specific PM_2·5 concentration at the postcode level was previously estimated using an ensemble model that combined multiple machine-learning algorithms and explanatory variables obtained via data on 24-h mean PM_2·5 concentrations from the US Environmental Protection Agency Air Quality System. Conditional logistic regression models were applied, adjusting for specific humidity, wind velocity, and maximum temperature extracted from the US Gridded Surface Meteorological Dataset. Housing status was established by registration staff or triage nurses on arrival at the emergency department. For people who were unhoused, exposure was defined based on the weighted mean PM_2·5 concentration at the city level proportional to the number of people who were unhoused in each specific city across urban centres in San Diego County. For people who were housed, we used residence postcode to measure exposure. We assessed the association between PM_2·5 from wildfire and non-wildfire sources and emergency-department visits in people who were housed and unhoused.

Findings

There were 587 562 emergency-department visits at the two hospitals, 76 407 (13·0%) of which were by people who were unhoused. People who were housed had a higher exposure to overall PM_2·5 (24-h mean over the study period of 9·904 mg/m³, SD 3·445) and non-wildfire PM_2·5 (9·663, 2·977) than people who were unhoused (9·863, 3·221; 9·557, 2·599). However, people who were unhoused had a higher exposure to wildfire-specific PM_2·5 (0·305, 1·797) than people who were housed (0·240, 1·690). Overall PM_2·5 exposure was associated with increased odds of emergency-department visits for both people who were housed (odds ratio 1·003, 95% CI 1·001–1·004 per 1 μg/m³ PM_2·5 for 0–3 days after exposure) and people who were unhoused (1·004, 1·000–1·008 for 0–3 days after exposure). We found that non-wildfire PM_2·5 was associated with emergency-department visits among people who were housed (1·003, 1·002–1·005 for 0–3 days after exposure) and wildfire-specific PM_2·5 was associated with emergency-department visits in people who were unhoused (1·006, 1·001–1·011 for 0–3 days after exposure).

Interpretation

People who were unhoused in San Diego County were more likely to visit emergency departments after exposure to increased wildfire-specific PM_2·5. As the intensity and frequency of wildfires increase, understanding risk factors for vulnerable populations, such as people who are unhoused, is crucial to develop effective adaptation strategies.

Funding

US National Institutes of Health, National Institute on Aging.