95 Rotational Patient Assignment versus Physician in Triage: A Comparison of Two Emergency Department Front-End Process Redesigns

Abstract

Rotational patient assignment (RPA) is an emergency department (ED) front-end process in which patients are automatically and algorithmically assigned to physicians. Physician in triage (PIT) is a front-end process in which a physician (often with a nurse) evaluates and treats patients before they are placed into a room in the main ED. Both have been associated with improvements in ED throughput, but their comparative efficacy is unknown. We sought to compare ED operational metrics using RPA versus those using PIT at a single facility in consecutive years. ED operational metrics were length of stay (LOS), arrival to provider time (APT), left without being seen (LWBS), left before treatment complete (LBTC), early (within 72 hour) returns to the ED (72R), early returns to the ED who were admitted (72R/A), and complaint ratio (CR). Design: Retrospective cohort review. Setting: Single site facility with 27,000 visits per year. Type of participants. Patients seen on 23 days during which PIT was used in 2011 to 2012 and a matched cohort of 23 days during which RPA was used in 2012 to 2013. All days in both groups were the busiest days of the week (Monday and Friday) during the busiest season of the year (winter). In the primary analysis, we performed an unadjusted comparison of the two groups. In the secondary analysis, we performed an adjusted comparison of time outcomes (LOS and APT) utilizing multivariate regression to account for identified potential confounders. Identified potential confounders were patient characteristics (age, sex, and emergency severity index [ESI] score), ED daily volume, ED physician staffing, ED nurse staffing, ED holding, and effective hospital occupancy. There were 1,906 visits during RPA and 1,869 visits during PIT. Primary Analysis: In a simple comparison, RPA was associated with a lower median LOS (219 minutes versus 233 minutes; difference of 14 minutes; 95% confidence interval (CI), 5-27 minutes) and lower median APT (25 minutes vs. 44 minutes; difference of 19 minutes; 95% CI, 15-22 minutes) than PIT. There were no significant changes in LWBS, LBTC, 72R, 72R, 72R/A, or CR. Secondary Analysis: Multivariate linear regression incorporating identified potential confounders found no statistically significant difference in the geometric mean of LOS for RPA versus PIT (204 vs 217 minutes; reduction of 6.3%; 95% CI, -3.6% to 15.2%). Secondary analysis did confirm an improvement in geometric mean of APT for RPA versus PIT (14.6 vs 27.4 minutes; reduction of 46.8%; 95% CI, 33.6% to 57.4%). In a single-site study, RPA was associated with a shorter APT than PIT but not a significantly shorter LOS. As RPA and PIT are very different processes, each with distinct benefits and drawbacks, data such as this may be useful as emergency departments choose between front-end process redesigns.