A proactive transfer policy for critical patient flow management

Area of Science:

• Operations Research
• Healthcare Management
• Computer Science

Background:

• Hospital emergency departments (EDs) face overcrowding, leading to long patient wait times and negatively impacting health outcomes.
• Delays in patient transfers exacerbate ED congestion and can increase mortality rates.

Purpose of the Study:

• To develop and evaluate a Markov decision process (MDP) model for optimizing patient flow between EDs and other hospital units.
• To implement a proactive patient transfer policy to improve ED efficiency and patient care.

Main Methods:

• Formulated an MDP model to estimate bed demand and guide proactive transfer decisions.
• Utilized approximate dynamic programming (ADP) to derive an optimal decision policy for bed allocation.
• Tested the model on different-sized instances and simulated real hospital data.

Main Results:

• The model demonstrated that optimal patient transfers vary with changes in arrival rates, but remain stable under proportional changes across units.
• Simulations using real data from a Chilean hospital showed significant improvements: over 50% reduction in patient wait times, a 21% increase in patient capacity, and a decrease in queue abandonment from 7% to less than 1%.

Conclusions:

• The proposed MDP model effectively improves ED performance indicators, including wait times, capacity, and patient abandonment.
• Proactive patient transfer management using ADP is a viable strategy for enhancing hospital operational efficiency and patient outcomes.