Online expressions of heat risk: social media posting and social vulnerability in California

Abstract Heat waves in California are occurring with increasing frequency, duration, and intensity, largely due to climate change. Furthermore, the burden of extreme heat disproportionately affects various communities both geographically and socioeconomically. As a result, understanding how heat is discussed on social media is critical for informing the design of public health interventions. This study involved a spatiotemporal analysis of heat-related tweets (N = 24,598) posted across California between April and October from 2016 to 2022. Poisson regression models were used to examine county-level counts of heat tweets, with population size included as an offset term in the model. Logistic regression models estimated the probability that a heat tweet reflected coping or life impact themes. Separately, regression models estimated the likelihood of heat tweets expressing positive or negative sentiment. Explanatory variables included socio-demographic and housing variables derived from the Healthy Places Index and American Community Survey. Sensitive housing and income predicted posting of heat tweets about coping, whereas income and percentage of active commuters predicted posting of heat tweets about life impacts. Active commuters, tree canopy, and sensitive housing (living in an RV, van, or boat) predicted negative sentiment heat tweets. The Hispanic population in California also appeared vulnerable, showing a modest tendency toward greater life impact and fewer coping-related heat tweets. These findings suggest that public health interventions should prioritize outreach to individuals living in sensitive housing, with an emphasis on indoor heat mitigation strategies. Overall, results demonstrate the value of social media heat discourse as a complementary tool for identifying vulnerable communities and informing targeted extreme heat awareness campaigns.