Predicting 72-hour emergency department revisits.

Area of Science:

• Health Services Research
• Predictive Analytics in Healthcare
• Emergency Medicine

Background:

• Emergency Department (ED) revisits pose a significant burden on healthcare systems.
• Identifying patients at high risk for revisits is crucial for efficient resource allocation and timely interventions.
• Current methods may not adequately capture the complexity of factors leading to ED readmissions.

Purpose of the Study:

• To develop and validate a predictive model for identifying patients at high risk of ED revisits within 72 hours.
• To enable healthcare systems to implement targeted interventions for high-risk patient groups.
• To improve patient outcomes and optimize ED resource utilization.

Main Methods:

• Multivariate logistic regression was used to build the predictive model.
• Data from 21,141 ED visits across four Veterans medical centers were analyzed.
• Patient demographics, prior healthcare utilization, and comorbidities were used as predictive variables; split-sample validation was employed.

Main Results:

• The final model, incorporating demographics, utilization, and comorbidities, achieved a c-statistic of 0.74 (development) and 0.73 (validation).
• Adding prior year healthcare utilization significantly improved the model's predictive power (c-statistic 0.70).
• Patient demographics alone showed limited predictive ability (c-statistic ~0.55).

Conclusions:

• A robust predictive model was developed to identify patients at high risk for ED revisits.
• This model facilitates targeted interventions for "frequent flyers," potentially reducing healthcare costs and ED wait times.
• The findings support the implementation of data-driven strategies for proactive patient management in emergency care settings.