Patient admission control and bed allocation during a pandemic using dynamic programming and simulation

Purpose This paper aims to examine patient admission control (AC) policies aimed at reducing patient waiting times during a pandemic. Unlike previous studies that focused on AC within a single hospital, this research seeks to minimize waiting times across multiple hospitals. The primary objective is to ensure that patients are admitted to the most appropriate hospital to reduce congestion during a pandemic. Design/methodology/approach This paper proposes two stochastic dynamic programming (DP) models. The first model treats the number of available beds as a fixed parameter, while the second model considers the number of beds as a decision variable. In the first model, the main decision is determining which hospital a pandemic patient should be assigned to. Bed allocation is not addressed in this model. The rationale for presenting two models is based on the dynamic nature of hospital resource allocation. In some situations, hospital administrators must decide how to configure beds between pandemic and nonpandemic wards. In other scenarios, bed allocation is predetermined and remains constant throughout the pandemic. DP algorithms are used to precisely solve small-scale instances of the problem and generate policies for patient assignment. These policies are then evaluated against alternative heuristic policies using larger-scale problem instances and simulation tools. Findings Simulation results reveal that implementing an AC unit and adopting an appropriate patient allocation policy can reduce average patient waiting times by approximately 50%. Moreover, when equity is a consideration (when the objectives are in the form of min-max), the policies derived from the DP approach outperform heuristic policies. However, some heuristic policies are more effective when equity is not a primary concern. Originality/value The findings of this research can assist health-care managers in making informed decisions by highlighting the implications and performance of various strategies.