Estimating emergency department crowding with stochastic population models.

Area of Science:

• Complex Systems Science
• Operations Research
• Public Health

Background:

• Crowding in high-demand service environments like hospital emergency departments is common and fluctuates unpredictably.
• Previous models for crowding often rely on averages, are overly complex, or use difficult-to-interpret machine learning approaches.

Purpose of the Study:

• To demonstrate that a stochastic population model, successfully applied to natural phenomena, can effectively describe and forecast crowding in hospital emergency departments.
• To identify key factors influencing overcrowding and quantify their impact on event probability.

Main Methods:

• Applied a stochastic population model to five years of minute-by-minute emergency department records.
• Analyzed the sensitivity of overcrowding events to patient arrival rates and length of stay.

Main Results:

• The stochastic population model accurately predicted crowding distributions in emergency departments.
• A 10% increase in patient arrivals tripled the likelihood of overcrowding.
• Reducing patient length-of-stay by 20 minutes (8.5%) halved the probability of severe overcrowding events.

Conclusions:

• A unified stochastic model can explain volatile crowding dynamics observed in hospital emergency departments.
• Predictive modeling of crowding is crucial for preventing service breakdowns and improving patient flow.
• Findings highlight the potential for optimizing operational efficiency in high-demand service systems through dynamic modeling.