Effect of a Distributed Split-Flow Model on Length of Stay for Discharged Emergency Department Patients

Emergency department crowding and inpatient boarding are nationwide challenges that impede efficient patient evaluation and lead to increased length of stay (LOS), decreased patient satisfaction, and increased morbidity and mortality. At our large academic ED with approximately 63,000 annual visits, LOS for discharged patients consistently exceeded goals. We sought to improve throughput by implementing a distributed split-flow model within existing geographic care teams.

This quality improvement initiative was conducted in a single large academic ED segmented into six geographic pods, each staffed by a physician–nursing team. Prior to the intervention, vertical flow (low-acuity) patients were cared for in a single "fast track" pod by a single care team. Beginning April 2025, a distributed split-flow model was implemented wherein each team managed both horizontal (bed-based) and vertical flow patients within their pod. Upon arrival in a pod, vertical flow patients were rapidly evaluated in a designated fast track chair, allowing initiation of diagnostic workup before transitioning to an internal waiting area, thereby preserving beds for higher-acuity patients. LOS data for discharged patients was collected 12 months before and 8 months after implementation. Statistical process control analysis using control charts was performed to compare pre- and post-intervention performance.

In the pre-intervention period, median LOS for discharged patients was 339 minutes (moving range 15.9; standard deviation 14.1; upper control limit [UCL] 381; lower control limit [LCL] 296). Following implementation, median LOS decreased to 289 minutes (moving range 5.9; standard deviation 5.2; UCL 305 minutes; LCL 274 minutes). The post-intervention period demonstrated eight consecutive monthly data points below the pre-intervention LCL, indicating special cause variation and sustained improvement.

Implementation of a distributed split-flow model within existing geographic care teams was associated with a meaningful and sustained reduction in LOS for discharged ED patients. This approach scales the split flow model, allowing for multiple parallel vertical flow pathways. This parellelization reduces bottlenecks that occur in a geographically segmented ED, maximizing physical space efficiency and maintaining throughput without additional staffing.