Document Type


Publication Date



Maine Medical Center, Medical Education, Maine Medical Center Research Institute, Internal Medicine, Emergency Medicine, Center for Outcomes Research & Evaluation

MeSH Headings

Humans, Hospitals, Patients' Rooms


*Purpose*: Patient-centered care is increasingly described as patients receiving “the right care, in the right place, at the right time”. In the inpatient setting, bed placement can contribute to that goal by facilitating high-performance teams while minimizing high-cost patient boarding and the safety implications of off-service placement. Geographic cohorting is a potentially important hospital quality goal for improving efficiency, as it may reduce length of stay and improve patient safety. Geographic cohorting (GCh) aims to place patients in rooms primarily by a team-based physical location. An example of this process would be to assign a hospitalist team to one unit and restrict admissions for that hospitalist team to that physical unit whenever possible. This study investigated the current state of GCh for general medicine patients at Maine Medical Center (MMC), the impact on length of stay (LOS) and other key measures, and potential opportunities and implications for implementing GCh in the future.

*Methods*: An interdepartmental working group determined the operational definition for GCh for general medicine patients at MMC (e.g. patients on the “Medicine R7” team are “cohorted” when placed on the unit R7). These rules were compared to retrospective (FY2017-August 2019) electronic health record (EHR) data for patient admissions. The data elements included admission and discharge date/time/unit, and each treatment team assigned during the patient’s stay (e.g., “Medicine R4”). Using the R programming language and Excel, data were analyzed for cohorting performance (e.g. percent of patients cohorted at admission) by unit, by team, longitudinally, and in different hospital conditions (e.g. high census). Cohorted and non-cohorted patients were compared for length of stay. Using the placement rules, unit capacities, and actual data, a simulated “patient placer” was developed using Visual Basic for Applications (VBA). Using this simulator, altered placement rule sets were compared for impact on cohorting and hospital efficiency.

*Results*: Patient admissions missing crucial data elements were removed from the data set (37.65%), leaving 56,631 patient admissions for analysis; of these, 20,262 (35.78%) were general medicine patients. 52.1% of general medicine patients were cohorted on admission versus more than 60% at discharge. The Medicine P2C and Hospitalist Gibson APP teams were the most commonly cohorted, as were the P2C and R7 units. Cohorting performance did not appear to correlate with hospital census or turnover. Medicine R2/R2APP showed a decrease in median LOS between cohorted and non-cohorted patients of .27 days. The patient placer demonstrated that the prioritization of cohorting during patient placement is likely feasible within the current capacity of MMC.

*Conclusions*: General medicine patients are more often cohorted on discharge than admission, likely due to internal transfers during the patient admission. Cohorting at admission appears to be correlated with a decreased length of stay in certain conditions. Improving GCh appears feasible given the results of the patient placer. This study was limited by missing data and a lack of patient-level demographic and acuity information, especially level of care. Further studies should match patients based on their characteristics in order to evaluate the impact of cohorting on outcomes.


2020 Costas T. Lambrew Research Retreat