The effects of residential air conditioning and social vulnerability on heat-related hospitalizations in California

Lack of access to cool spaces is considered to be a key determinant of heat vulnerability. However, assessments of the moderating effect of access to cooling on heat health outcomes have typically used spatially coarse estimates of residential air conditioning (AC) prevalence. Here we estimate patterns of inter- and intra-urban AC ownership across 1,582 zip code tabulation areas (ZCTAs) in 27 California cities, use the results to assess the covariation between AC and social vulnerability, and examine how AC prevalence and social vulnerability modify the association between extreme heat and morbidity. Our approach combines Census records, the CDC Social Vulnerability Index (SVI), meteorological variables, and hospital admissions across the 2012-2019 warm season from the California Department of Health Care Access and Information. We first construct individual- and ZCTA-level estimates of residential AC prevalence, which we then compare with SVI between and within metro areas, and separately assess their effect modification on all-cause and cause-specific morbidity using a space-time-stratified case-crossover design with conditional Poisson regression models. We found that AC prevalence is weakly positively correlated with social vulnerability (r = 0.15) across the state and weakly negatively correlated within cities (r = -0.25). We found evidence of effect modification from AC prevalence on associations between extreme heat and all-cause (RR 1.04 (1.03, 1.05) for low AC versus RR 1.01 (1.00, 1.02) for high AC) as well as some cause-specific hospitalizations (e.g., heat-related illness: RR 1.42 (1.32, 1.52) for low AC versus 1.21 (1.11, 1.32) for high AC). Importantly, we find that residential AC reduces - but does not eliminate - the risks of adverse heat health outcomes.