Setting and Problem
Over the past several decades, hospitalized patients have become increasingly complex, often with multisystem needs. In response, hospital teams now rely heavily on subspecialty consultants and interprofessional colleagues. While this has improved care delivery, fragmentation of responsibilities has changed the clinical learning environment, and graduate medical education has suffered.
By the mid-2020s, collaborative models of care made it unclear who was responsible for teaching and evaluating residents. Subspecialty consultation—particularly e-consults—were common, but residents and fellows rarely met face-to-face, forfeiting opportunities for workplace learning. Isolation and anonymity overtook any sense of community in the hospital, and rates of burnout soared. Moreover, asynchronous siloed work patterns led to misunderstandings and conflicting recommendations from different teams.
Recognizing the potential for the built environment to impact work patterns and workplace learning, we assessed whether colocating medicine subspecialty fellows in a shared workspace near the medicine resident workroom could increase face-to-face interactions during subspecialty consultation. We hypothesized this would have benefits for communication, teaching, and burnout.
In 2020, our medical center began a 10-year process of building a new hospital. To inform space design and involve end users in the process, we performed surveys and focus groups with residents and fellows to understand the consultation process, assess the feasibility and acceptability of creating a shared workspace for fellows, and measure burnout. After the hospital opened in 2030—with a shared fellow workspace adjacent to the medicine resident workroom—we repeated our surveys and focus groups. We also tracked work patterns by looking at computer logins and performed work sampling in which observers noted who used the new workspace and how (eg, working alone, discussing a patient, teaching at the whiteboard).
Respondents to our baseline survey confirmed that face-to-face communication and teaching during consultation were rare. We also learned that residents and fellows worked on different floors in different buildings (figure, panel A), and few knew each other, leaving many feeling isolated. Over 40% of residents and 50% of fellows met criteria for burnout. In focus groups, fellows reported a willingness to try a shared workspace, especially if it had ample workstations, snacks, and places for socialization and respite.
Based on our review of computer logins during the 3 months after the new shared workspace (figure, panel B) opened, 78% of fellows representing 10 of 11 medical subspecialties (all but gastroenterology) used the workspace. Half used it daily, and a core group of fellows (23%) performed most of their work there. Work sampling data revealed a median of 4 fellows (range 0–9) and 2 residents (range 0–17) in the workspace at any time. Entire medicine teams often came to the room for interdisciplinary meetings, improving teams' agreement on care plans, and fellows were taught using the whiteboard multiple times daily.
In surveys and focus groups, both residents and fellows reported that working in close proximity facilitated getting to know each other and improved community. They noted that face-to-face interactions became much more common, leading to more in-depth conversations about patients, reduced pushback on consults, and increased resident empowerment to ask questions as opposed to “just following the recs.” All trainees reported feeling less isolated, and rates of burnout fell to 20% and 23% among residents and fellows, respectively.
Our novel intervention of colocating medicine subspecialty fellows in a shared workspace near medicine residents increased face-to-face communication, improved teaching, and reduced burnout. This design-based approach could be readily adopted in other departments and institutions—particularly at academic medical centers—that have compelling reasons to improve the trainee experience. Remodeling existing spaces would likely achieve similar goals at substantially lower costs.
The authors would like to thank Anne Marie Amisola for assistance creating the figure.