Background: At our multi-hospital system, our utilization management (UM) program found that the length of stay (LOS) for observation patients in one hospital was, on average, 10 hours longer than the rest of the hospital system. There was a challenge of having better observation management, but there was no space or staffing to create a dedicated observation unit.
Purpose: To use our existing staff to create and test a model that had dedicated team members for our observation care patients, without locating all the patients on a geographical unit; in essence, a virtual observation unit.
Description: With strong executive leadership support, a multidisciplinary observation team was assembled, consisting of two hospitalist Advanced Practice Providers (APPs), a dedicated UM nurse, and a physician advisor (PA). Collaborative efforts were also initiated with various hospital departments, including emergency medicine, pharmacy, radiology, lab services and physical therapy, to assist in expediting discharges.Additionally, we enhanced our Utilization Management (UM) staffing by collaborating with another hospital in our system, allowing their UM nurses remote access to our Electronic Medical Records (EMR). This was done to improve admission status accuracy at the Emergency Department entry point. Patients under observation were managed by our observation team. Daily virtual meetings were held for multidisciplinary rounds to review each observation case, identify discharge obstacles, and proactively address them to expedite the process. If patients were not ready for discharge, we reevaluated whether inpatient admission was now necessary.Our baseline data was observation admissions in September 2023 and our virtual observation unit was tested in October 2023. All observation and outpatient in a bed hospitalizations were included. Observation time was defined as hours from arriving on the hospital unit until exiting the hospital. We excluded patients under the age of 18, patients who were later changed to inpatient status, hospice patients, and patients with total admission time lasting less than 8 hours.Figure 1 shows that the mean, median and spread of observation LOS were all less in the pilot compared to baseline. The mean LOS reduced from 38.6 hours to 28.3 hours, with a p-value < 0.01. Mitigating the effect of outliers by comparing the medians showed similar results: 33.7 hours vs 22.6; p-value < 0.01.
Conclusions: With the ever-increasing healthcare costs, it has become imperative for hospitals to optimize care delivery processes (1). We present a novel approach to observation management, using virtual units instead of cohorting these patients in a geographical unit. There was a statistically significant reduction in mean and median LOS even after addressing outliers. To the best of the authors’ knowledge, this is the first description of this type of model for observation management and could be replicated in other hospitals with space and staffing limitations to implement a dedicated observation unit.
