Dispatching policies during prolonged mass casualty incidents

Mass casualty incidents that result from prolonged increases in patient arrivals represent a unique modeling challenge with patients arriving and queuing over extended periods of time. In this paper, we consider how to optimally dispatch ambulances to prioritized patients during these incidents. Patients arrive, queue, and renege, and their conditions deteriorate over time, and ambulances are allowed to idle while less emergent patients are queued, thereby lifting several assumptions typically made in the literature. We formulate the ambulance dispatching problem as a Markov decision process model with patients prioritized by the benefit they will receive from ambulance care and with two classes of ambulances. Computational results are presented for a real-world example, and an extensive sensitivity analysis is performed. We observe that under the optimal policies, ambulances often remain idle when less emergent patients are queued to provide quicker service to future, more emergent patients. We propose and evaluate heuristics that represent static idling policies to study how to practically implement the results. The results suggest that delaying service to low priority patients when the system is congested enables ambulances to immediately respond to future high priority patients who may need care and whose conditions may deteriorate.