Feed

The 2-1-1 information and referral system connects vulnerable populations to essential services during public health emergencies. However, there is limited evidence examining the specific needs of unhoused populations during these events. This study examines associations between temperature variations, tropical cyclones, and service utilization patterns among unhoused populations using 2-1-1 call data from southern Louisiana (2014-2023). Analyzing over 20,000 calls, predominantly in urban areas (65%), we employed negative binomial regression, distributed lag non-linear modeling (DLNM), and natural language processing to uncover patterns in help-seeking behavior during temperature variations and tropical cyclones. Negative binomial models demonstrate declines in 2-1-1 utilization both during cyclone events (IRR: 0.77, CI: 0.60-0.99) and in the recovery period (IRR: 0.86, CI: 0.76-0.97). Significant associations were observed across the temperature distribution with both the coldest (D1, IRR: = 1.22, 95% CI = 1.08-1.38) and warmest (D10, IRR = 1.21, 95% CI = 1.06-1.38) temperature deciles, indicating increased call volume at the coldest and warmest temperatures compared to moderate temperatures. DLNM results corroborated a U-shaped association at temperature extremes, though effects were predominantly significant only at the coldest temperatures. Computational text analysis of call narratives revealed that mental health mentions increased significantly with temperature (from 19.4% in coldest conditions to 30.5% in warmest conditions), but decreased during cyclone periods (from 29.1% in non-cyclone conditions to 15.5-20.2% during various cyclone phases). While 2-1-1 data captures only those with phone access and service awareness, potentially underrepresenting the most marginalized, these findings provide actionable insights into help-seeking patterns among service-connected (e.g., 2-1-1) unhoused populations during and following extreme weather events.