Measuring and forecasting emergency department crowding in real time

Area of Science:

• Emergency medicine
• Healthcare management
• Operations research

Background:

• Emergency department (ED) crowding is a significant challenge in healthcare delivery.
• Effective monitoring and prediction of ED crowding are crucial for optimizing patient flow and resource allocation.

Purpose of the Study:

• To quantify the potential of four measures—Emergency Department Work Index (EDWIN), National Emergency Department Overcrowding Scale (NEDOCS), Real-time Emergency Analysis of Demand Indicators (READI) Demand Value, and Work Score—for monitoring current and near-future ED crowding.
• To compare the performance of these measures against ambulance diversion status and occupancy level.

Main Methods:

• Four ED crowding measures were calculated at 10-minute intervals over an 8-week period.
• Ambulance diversion status served as the crowding outcome, with occupancy level as the baseline.
• Discriminatory power for current crowding was assessed using the area under the receiver operating characteristic curve (AUC).
• Forecasting power was evaluated using activity monitoring operating characteristic curves to measure early warning timeliness.

Main Results:

• The Work Score (AUC 0.90) and occupancy level (AUC 0.90) demonstrated the highest discriminatory power for current ED crowding, followed by NEDOCS (AUC 0.88) and EDWIN (AUC 0.81).
• The READI Demand Value showed lower discriminatory power (AUC 0.65).
• Only occupancy level provided a median advance warning of 1 hour and 7 minutes before crowding events, with one false alarm per week.

Conclusions:

• EDWIN, NEDOCS, and Work Score effectively monitor current ED crowding but do not outperform occupancy level.
• None of the evaluated measures provided substantial advance warning of crowding at low false alarm rates.
• Occupancy level appears to be a strong predictor of current ED crowding and offers limited forecasting capability.