A Virtual Physician-in-Triage Model Decreases Left Without Being Seen Rate

Since the COVID-19 pandemic, rates of Emergency Department (ED) overcrowding have increased concomitant with length of ED visits, time to provider, and rates of Left Not Seen (LNS). Previous studies have found mixed results of a Virtual Physician in Triage (VPIT) model on time to discharge, LNS, Left Before Completion of Service (LBCS) and time to disposition decision . The objectives of our study were to assess the impact of VPIT on our ED’s LNS rates and to assess how long a patient waits prior to leaving after seeing a VPIT provider compared to standard care.

We designed a VPIT workflow to triage patients meeting pre-specified criteria on Monday to Wednesday between 11 am and 5 pm. A single-center, retrospective cohort study was performed to assess the rate of LNS. The analysis cohort was defined as all patients presenting Monday to Wednesday between 11 am and 5 pm. The control cohort was defined as ED arrival between 11a-5p on Thursday and Friday. These two cohorts were also compared to the same time periods one year (Y-1) and two years (Y-2) before the pilot program. Duration of ED waiting room stay was defined as the time of patient check in to the time the patient was roomed in the ED. LNS was defined as patients who left the department without seeing any provider or VPIT clinician.

VPIT eligible patient volume increased during the study with the Y-2 period seeing 8,163 patients, the Y-1 period 8,556 patients, and the pilot period 8,708 patients. The LNS rate in the Y-2 period was 8.6% in the analysis cohort compared to 7.2% in the control cohort. During the Y-1 period, LNS in the analysis cohort was 6.8% compared to 5.2% in the control. During the pilot period, the rates were 6.6% and 7.6% respectively. The rate of LNS decreased after the integration of the VPIT model compared to the Y-2 period (8.6% versus 6.6%, p<0.00001). The rate of LNS in the control group increased (7.2% to 7.6%, p<0.41). The introduction of a VPIT model increased the amount of time patients were willing to wait prior to being seen by a provider although statistically insignificant (274 min vs 213 min, p < 0.32).

Introduction of a VPIT model decreased the LNS rate and showed a trend towards patients being willing to wait longer to be seen, though statistically insignificant. Future directions may include assessing VPIT’s impact on patient satisfaction and time to disposition.