49EMF Designing Emergency Departments to Provide Efficient, Patient-Centered Care: An Analysis of Split Flow and Sub-Waiting Area Models

Abstract

As emergency department (ED) crowding has worsened and its effects catalogued, ED leaders have sought process improvements to improve efficiency while architects have proposed design strategies to achieve the same. Unfortunately, these efforts have largely failed to cross professional boundaries. The present study explored the essential interaction between ED design and flow with a goal to optimize split-flow patient care systems. The study is a 2 factor analysis, examining the interaction of 3 flow models (split by Emergency Severity Index score, split by a physician, and no split) with 3 sub-waiting area types (no sub-waiting, 1 sub-waiting, and 2 sub-waiting). Thus, 9 total models were examined. Outcomes of interest were operational metrics (length of stay (LOS), room utilization rate) and patient-centered metrics (door-to-provider time (D2P), left without being seen (LWBS) rate, and number of movements per patient). We used patient and encounter-level data from 30 randomly selected days at a 100,000 annual visit academic ED to create and specify ARENA discrete event simulation models. We validated 3 models against actual ED data, and then used these to extrapolate performance in the remaining 6 flow-design sub-types. Flow split by ESI with 1 waiting area (the most common model used by EDs) was used as the control. We also sought to determine the best design specification given a fixed flow as well as the best flow specification given a fixed design. Models were compared and analyzed for statistical significance and effect size using one-way analysis of variance (ANOVA) to find the most efficient model, two-way ANOVA to measure the impact of each group of independent variables (flow types and/or design types), and linear and non-linear regression. One-way ANOVA testing demonstrated the superiority of the ED flow split by a physician with 2 sub-waiting areas. This model resulted in the smallest LOS of 189.8 minutes (54 min shorter vs. control), and the highest bed utilization of 5.02 patients/bed/day (41.8% increase vs. control). In addition, physician-directed flow with 2 sub-waiting areas also showed superior performance in several patient-centered metrics, having the best D2P time of 9.6 minutes (vs. 26.3 min, control) and only a 1.17% LWBS rate. Not surprisingly, having 2 sub-waiting areas did result in an increase in the number of different treatment spaces a patient visited, 4.2 vs. 4. For any given flow type, adding 1 additional sub-waiting area resulted in a decreased LOS (range 20.9 - 37.3 min), increases in bed utilization (0.42 - 0.70 patients/bed/day), decreases in D2P (2.6 - 8.3 min), and decreases in LWBS (0.8% - 1.24%). For a given number of sub-waiting areas, flow split by a physician resulted in superior performance, followed by flow split by ESI, followed by no split flow. Modifications to both ED flow and physical design have significant potential to improve both operational and patient-centered metrics. In general, addition of sub-waiting areas and use of a physician to split flow, as opposed to ESI score sorting, significantly improved operational and patient centered metrics. EDs should consider implementation of a physician-based intake model with multiple sub-waiting areas to improve performance.