The Effects of Observation Unit Location and Physician Staffing on Patient Outcomes
Observation units have become increasingly common in recent years. Observation units (OUs) can be staffed by emergency physicians (EP) or by internal medicine physicians (IMPs). The location of these units can vary from close proximity to the emergency department (ED) or to remote locations, such as on an inpatient floor. There is a paucity of data regarding ideal staffing models and location of OUs. We sought to evaluate the effect, if any, that OU location and staffing specialty might have on patient outcomes.
This is a single-center, retrospective study completed at a tertiary care academic hospital in an urban setting from 2014 to 2018. The OU at this hospital is a 20-bed closed, complex unit with 24-hour physician assistant coverage. From February 2014 to May 2015, the unit was located in close proximity to the ED and staffed by EPs. In May 2015, the unit was relocated to an inpatient ward, where it was staffed by IMPs. In June 2018, the unit moved back to a care location adjacent to the ED, and it remained under the care of IMPs.
Three-month increments of data were compiled to compare outcomes between the different locations and staffing: 1) near the ED with EPs, 2) inpatient ward with IMPs, and 3) near the ED with IMPs. Outcomes included observation unit mean length of stay (LOS), percentage of patients with less than 8-hour observation LOS (indicating a possibly unnecessary observation unit stay), and rate of hospital admission. Fisher Exact Test or t-stat was used for analysis as appropriate. All statistics were calculated using the R statistical computing software version 3.6.0.
A total of 3,896 patients were seen in the OU during the study period. The mean LOS was longest at 24 hours, when the unit was on an inpatient ward and staffed by IMPs. Mean LOS was shortest, 18 hours, when the unit was in close proximity to the ED and staffed by EPs (p<0.01). LOS decreased from 24 hours to 21 hours after the change in location from upstairs on an inpatient ward to a location in close proximity to the ED, staffed by IMPs (p<0.01). The admission rate was highest at 33% when the unit was on an inpatient ward and staffed by IMPs, whereas the admission rate was lowest at 27% with the unit was adjacent to the ED, staffed by the IMPs (p=1.00). The percentage of patients with an observation unit LOS less than eight hours was greatest, 13%, when the OU was close to the ED, and patients were under the care of EPs. When moved to an inpatient ward floor and staffed by IMPs, the percentage fell to 3% (p<0.01). When the unit was relocated back to close proximity to the ED, but still staffed by IMPs, the percentage of patients with an observation unit LOS less than 8 hours was 6% (p<0.01).
Observation unit LOS was the shortest when the unit was located adjacent to the ED and was staffed by EPs. However, a larger percentage of patients were discharged or admitted within eight hours when the unit was in this location and under this staffing model, thus raising the question of appropriateness of unit use. LOS improved with relocation of the unit from an upstairs inpatient ward to a downstairs location adjacent to the ED, suggesting the importance unit location may play on observation patient LOS. There was no statistically significant difference in admission rates among all groups.
Jonathan Yeo, MD,
Peter Rachlin, MD,
Michael Herscher, MD, and
Nachi Gupta, MD