Hospital selection for suspected stroke: risk-averse approach considering the minimal risk of exceeding the therapeutic time window.

Abstract

Timely treatment within the therapeutic window is critical for patients with stroke. This study adopts a risk-averse optimization approach to maximize the likelihood of receiving treatment within this window. We developed an optimization model using data from a citywide stroke registry (July 1, 2019 to December 31, 2020). The model included scene locations, data from 10 stroke centers, Cincinnati Prehospital Stroke Scale (CPSS) scores, prehospital times, and in-hospital processing times for endovascular thrombectomy (EVT) in patients with large vessel occlusion (LVO) stroke and intravenous thrombolysis (IVT) in patients with non-LVO stroke. Transport times were calculated using Google Maps and hospital delivery was determined through mathematical programming. We compared the Risk-Averse strategy against four other prehospital strategies (Nearest, Mothership, Optimistic, and Pessimistic strategies) using Monte Carlo simulations for 726 patients. The Risk-Averse strategy consistently outperformed others for patients with LVO stroke, with 82.0% (95% CI 79.8% to 84.1%) receiving EVT within 6 hours, representing a 4.8% increase over the Mothership strategy and a 13.7% increase over the Nearest strategy. The average time to EVT was 268.1 min (95% CI 258.3 to 278.5). For patients with non-LVO stroke, 80.6% (95% CI 80.0% to 81.6%) received IVT within 4.5 hours, with no significant differences between strategies. The Risk-Averse hospital selection strategy significantly improves EVT rates for patients with LVO stroke without delaying IVT for those with non-LVO stroke.