Predicting emergency department volume using forecasting methods to create a "surge response" for noncrisis events

Area of Science:

• Health Services Research
• Emergency Medicine
• Operations Research

Background:

• Emergency departments (EDs) face challenges in managing fluctuating patient volumes.
• Accurate prediction of patient surges is crucial for effective resource allocation and staffing.
• Existing methods may not adequately forecast non-crisis-related patient volume increases.

Purpose of the Study:

• To develop and validate mathematical models for predicting future ED patient volume surges.
• To assess the utility of ED variables, specifically physician capacity utilization, in predictive modeling.
• To inform staffing decisions during non-crisis surges by forecasting patient arrival rates.

Main Methods:

• Retrospective analysis of ED data from a large urban teaching hospital (July 2009-June 2010).
• Modeling physician capacity based on historical productivity and treatment capacity.
• Binary logistic regression used to predict ED capacity sufficiency for forecasted patient arrivals across multiple prediction horizons (15 mins to 12 hours).
• Model validation using an independent dataset (July-November 2010) evaluating positive predictive values, Type I/II errors, and real-time accuracy.

Main Results:

• The care utilization ratio (CUR), defined as new patients requiring treatment over total physician capacity, emerged as a robust ED state predictor.
• Prediction intervals of 30 minutes, 8 hours, and 12 hours demonstrated superior performance.
• Validation showed positive predictive values ranging from 0.738 to 0.872, with true positive rates of 74%-94% and true negative rates of 70%-90% for the 30-minute model.

Conclusions:

• The CUR is a novel and effective metric for assessing ED system performance.
• The study successfully modeled the trade-off between prediction accuracy and response time.
• Implementing these predictive models could improve current ED practices by enabling earlier identification of patient volume surges.