Simulation Models for Just-in-Time Provision of Resources in an Emergency Department

Abstract

We address the issues for modeling problems of obtaining medical resources, such as beds and personnel, on a just-in-time basis for providing patient services in a hospital's emergency department [ED]. Simulation modeling is one approach to solving this type of problem. However, most past ED simulations have been highly detailed models that specifically consider individual doctors, nurses, paramedics, orderlies, X-ray machines and other unique types of lab testing equipment as separate resources. When arriving patients request any of these resources, the model queues them if the requested resources are busy with other patients or administrative duties. We argue that these previous simulation models are appropriate for operations planning. However, for operations control a much simpler simulation can be appropriate. We present a simpler model that forecasts in real time when a generic ED will need additional resources to avoid becoming overcrowded. After obtaining an ED bed (queuing if necessary), each patient's treatment time in the ED is represented simply as an observation of a random variable. Although our model is significantly simpler than previous ED simulation models, it has been used successfully by the Vanderbilt University Adult Emergency Department to activate additional resources, including beds, nurses, and other personnel. This ED is located within a state-of-the-art medical center in a large urban area licensed to treat the highest levels of trauma with more than 50,000 patient visits each year. Our model was used in a pilot evaluation of simulation as a real-time tool for providing additional resources on a JIT basis to prevent significant overcrowding by accurately predicting when overcrowding will likely occur. This has allowed management in real time to inform nurses, assistant managers and educators with joint administrative/patient care duties to work exclusively on patients; to open additional temporary beds in hallways; and to expedite patient discharges. It has also been used to let flexible night-shift staff go home early. Since these resources are mobilized only when needed, their overall efficiency of utilization is increased. We discuss the simplifying assumptions in our simulation model and describe a comparable simulation written in a high-level simulation language, Arena [2008].