Community firearm violence during extreme temperatures in the continental United States: A spatiotemporal epidemiological analysis (2014–2024)

Firearm violence and climate change represent intersecting public health crises in the United States. While temporal associations between temperature and violence have been established, critical gaps remain in understanding the spatial dimensions, underlying mechanisms, and differential impacts of both heat and cold extremes on firearm violence patterns. To identify and characterize spatial clusters of firearm violence during extreme temperature events across the continental United States from 2014 to 2024, and to examine how incident characteristics, temporal factors, and contextual circumstances vary across these clusters. We analyzed 427,312 firearm violence incidents from the Gun Violence Archive and temperature data from NOAA's ClimGrid dataset between 2014 and 2024. Using county-specific thresholds for heat and cold (95th and 5th percentiles), we identified extreme temperature events (1-day, 2-day, and 3-day) using a 31-day moving window approach. Purely spatial Bernoulli models in SaTScan detected statistically significant clusters of firearm violence during temperature extremes. We employed logistic regression to identify predictors of cluster membership and used natural language processing to extract contextual patterns from incident narratives. During heatwaves (≥95th percentile), we identified three high-risk clusters: central Florida (24 counties, relative risk [RR] = 1.59, p < 0.001), southern Texas (50 counties, RR = 1.47, p < 0.001), and the New York City area (13 counties, RR = 1.33, p < 0.001). Three low-risk clusters emerged in the Midwest and Southeast (RR = 0.71-0.75, p < 0.001). During cold extremes (≤5th percentile), high-risk clusters appeared in Arkansas/Oklahoma/Missouri (RR=1.65, p<0.001) and Kansas/Missouri/Nebraska (RR=1.65, p<0.001), while low-risk clusters concentrated in Florida (RR=0.54- 0.58, p<0.05). Heat-related violence showed increased outdoor incidents, public space violence, and criminal activity, while cold-related violence demonstrated elevated domestic violence, substance-related incidents, and defensive gun use. Weekend and holiday effects varied by cluster type, with holidays amplifying the risk of cold-related violence. Firearm violence during temperature extremes exhibits distinct spatial clustering patterns, with regional variations suggesting differential vulnerability to climate-violence interactions. The contrasting mechanisms between heat and cold extremes—with heat increasing opportunistic outdoor violence and cold elevating planned domestic violence—necessitate differentiated prevention strategies. These findings support the integration of geographically targeted interventions that combine violence prevention with climate adaptation strategies.