Challenges of Using Publicly-Available Hospital Data to Quantify Health Effects from Wildfire Plumes in the East San Francisco Bay Area Communities of California, USA

Abstract

In the summer and fall of 2018 and 2020, major wildfires in Northern California (USA) impacted the San Francisco Bay Area. The remote 2018 and 2020 wildfires produced the highest PM2.5 concentrations experienced in the Tri-Valley of the East Bay Area during those two years. The Tri-Valley is composed of the San Ramon, Amador, and Livermore Valleys, surrounded by local terrain that creates a small airshed. In 1967, the California Air Resources Board created 15 Air Basins defined by regional geography, topographic and meteorological conditions. Airshed is sometimes synonymous with an urban-scale component of an Air Basin. We use airshed as a Tri-Valley component of the Bay Area Air Basin. This airshed spans across two counties (Contra Costa and Alameda) and encompasses four cities: San Ramon, Dublin, Pleasanton, and Livermore. PurpleAir (PA) sensors provided good geographic coverage of variation in PM2.5 in the Tri-Valley airshed. Several studies have established significant health effects from wildfire plumes by associating daily hospital visits with PM2.5 air quality data at local and regional scales. We hypothesize that during the wildfire smoke periods of 2018 and 2020 in the Tri-Valley area, there was an increase in hospitalizations and ED visits for respiratory (asthma and COPD-related) health effects, as compared to the same time periods during years with less fire activity. The primary goal of this study was to confirm health effects from wildfire plumes on a community scale using 5 years of publicly-available health data. However, with only monthly hospitalization data available, directly linking respiratory hospital and emergency department (ED) visits with PM2.5 concentrations was unsuccessful. Also, because COVID-19 masked all other causes of hospital visits in 2020, that year was ultimately eliminated from this study. However, visits during November 2018 being much higher than any other November in 2016, 2017, and 2019 implied a potential cause and effect. Daily hospitalization and air quality data are required to quantify any relationship by regression analysis. These findings help inform future studies on the health effects of air quality at community scales. KEYWORDS: PM2.5 Air Quality; Air Pollutant Exposure; Air Quality Monitoring; Wildfire Smoke; Respiratory Health; PurpleAir Sensors