Case Definition Analysis and Implications for Allocating Heat Emergency Resources in Washington, D.C.

As global temperatures rise, identifying heat-related illness (HRI) and optimizing the positioning of resources to respond are urgent needs in Washington, DC. Cities like Washington, DC, are at higher risk of experiencing heat emergencies due to the urban heat island effect, which intensifies already warming temperatures. HRI is identified based on a set of potential signs and symptoms, but they are not definitive indicators. That uncertainty demands attention to the case definitions used to identify HRI incidence, and the implications of such definitions for heat emergency resource positioning. The project aims to support DC Health’s decision-making regarding (1) how HRI is identified and (2) the operations of cooling centers-one of the main public resources to combat HRI during a heat emergency. DC Health currently operates with three HRI definitions. Using patient care reports (PCRs) from the DC State Emergency Medical Service Repository, we characterize and compare those existing definitions with a new fourth definition that leverages PCR narrative data and a large-language model (LLM), Llama 3.3. These four definitions are compared against a gold standard and with an optimization model of cooling centers. This project demonstrates the applicability of an LLM in supplementing future HRI surveillance.