Trainee Autonomy and Supervision in the Modern Clinical Learning Environment: A Mixed-Methods Study of Faculty and Trainee Perspectives

Background: Balancing autonomy and supervision during medical residency is important for trainee development while ensuring patient safety. In the modern clinical learning environment, tension exists when this balance is skewed. This study aimed to understand the current and ideal states of autonomy and supervision, then describe the factors that contribute to imbalance from both trainee and attending perspectives. Methods: A mixed-methods design included surveys and focus groups of trainees and attendings at three institutionally affiliated hospitals between May 2019-June 2020. Survey responses were compared using chi-square tests or Fisher’s exact tests. Open-ended survey and focus group questions were analyzed using thematic analysis. Results: Surveys were sent to 182 trainees and 208 attendings; 76 trainees (42%) and 101 attendings (49%) completed the survey. Fourteen trainees (8%) and 32 attendings (32%) participated in focus groups. Trainees perceived the current culture to be significantly more autonomous than attendings; both groups described an “ideal” culture as more autonomous than the current state. Focus group analysis revealed five core contributors to the balance of autonomy and supervision: attending-, trainee-, patient-, interpersonal-, and institutional-related factors. These factors were found to be dynamic and interactive with each other. Additionally, we identified a cultural shift in how the modern inpatient environment is impacted by increased hospitalist attending supervision and emphasis on patient safety and health system improvement initiatives. Conclusions: Trainees and attendings agree that the clinical learning environment should favor resident autonomy and that the current environment does not achieve the ideal balance. There are several factors contributing to autonomy and supervision, including attending-, resident-, patient-, interpersonal-, and institutional-related. These factors are complex, multifaceted, and dynamic. Cultural shifts towards supervision by primarily hospitalist attendings and increased attending accountability for patient safety and systems improvement outcomes further impacts trainee autonomy.

Trainees perceived the current culture to be signi cantly more autonomous than attendings; both groups described an "ideal" culture as more autonomous than the current state. Focus group analysis revealed ve core contributors to the balance of autonomy and supervision: attending-, trainee-, patient-, interpersonal-, and institutional-related factors. These factors were found to be dynamic and interactive with each other. Additionally, we identi ed a cultural shift in how the modern inpatient environment is impacted by increased hospitalist attending supervision and emphasis on patient safety and health system improvement initiatives.
Conclusions: Trainees and attendings agree that the clinical learning environment should favor resident autonomy and that the current environment does not achieve the ideal balance. There are several factors contributing to autonomy and supervision, including attending-, resident-, patient-, interpersonal-, and institutional-related. These factors are complex, multifaceted, and dynamic. Cultural shifts towards supervision by primarily hospitalist attendings and increased attending accountability for patient safety and systems improvement outcomes further impacts trainee autonomy.

Background
Balancing autonomy and supervision during medical residency is important for promoting trainee development while also ensuring patient safety. Several regulatory changes and the evolution of patient safety and quality improvement movements in healthcare over the past twenty years have emphasized increasing the quality of trainee supervision by attending supervisors. [1][2][3] Growing tension between trainees and attendings around autonomy and supervision has been reported, particularly when the balance is skewed in one direction or another. Studies have noted that trainees desire more autonomy than attendings are willing to provide, and that attending support for trainee autonomy is perceived to be limited. 4,5 Both extremes are perceived negatively -too much autonomy can cause feelings of abandonment and hinder trainee's clinical decision-making, while too much oversight can foster feelings of distrust and prevent trainees from expressing concerns on rounds. Attendings grapple with this balance as their role has expanded on the team in response to pressure to achieve hospital quality metrics or to compensate for team discontinuity related to duty hour restrictions or other residency curricular activities. 6 For trainees, lack of autonomy has been reported to negatively impact well-being. 7,8 Appropriate supervision promotes safe patient care, resident development and education, and opportunities for evaluation. 1 However, there is debate over whether these goals are being achieved. While one meta-analysis noted a slight decrease in patient mortality after the implementation of supervision and duty hour standards, this has not been reproduced in other studies. 9,10 Similarly, the impact of increased supervision on medical error rates has had con icting results. 11,12 Effect on trainee development is also unclear; some studies suggest increased supervision improved the clinical-learning environment, while others found that structured independence positively affected their development, sense of "ownership" over patient care, and overall satisfaction. 13,14 Entrustment and the variables that affect a supervisor's con dence in a trainee has been an area of robust research. The basis of entrustment decision-making is a supervisor's assessment of a learner's ability, integrity, reliability, and humility; it is speci c to a given patient care activity and clinical situation. 15,16 In an attempt to characterize a trainee's progress through stages of responsibility toward independent practice, Entrustable Professional Activities (EPAs) have been created to capture behaviors and activities that represent a trainee's development through residency. 17 However, these assessments are performed retrospectively by supervising faculty after a period of direct observation, are speci c to a single skilled behavior, and are not easily accessible to future supervisors. 18 In this study, we de ned "autonomy" as the ability to provide patient care in an independent and competent manner and "supervision" as the provision of both direct and indirect oversight by a more experienced individual in a clinical learning environment to achieve the aforementioned goals. This study aimed to understand the current and ideal states of autonomy and supervision with the hypothesis that a discrepancy exists, and further describe the factors that contribute to tension from the perspectives of both trainees and attendings.

Study Design
We conducted a mixed-methods study using surveys and focus groups to explore attending and trainee perspectives on autonomy and supervision during inpatient medicine teaching rotations.

Setting, Participants and Oversight
This study took place at the three teaching hospitals of the University of California-San Francisco (UCSF): quaternary care hospital, public county hospital, and Veterans Affairs (VA) hospital. Eligible participants included internal medicine trainees (interns and residents) and attending physicians from each site.

Survey development and data collection
We developed two surveys to explore perspectives and practices related to autonomy and supervision during inpatient medicine rotations from trainees and attendings (Appendix 1 and 2). These surveys utilized both xed-response questions (i.e. 5-point Likert scale) and open-ended free text responses to evaluate perceptions of "current" and "ideal" state of autonomy and supervision, experience with autonomy and supervision during training, perceived value of attending contributions to patient care and medical education, and perceived impact of autonomy and supervision on burnout and wellness. Both surveys presented six clinical scenarios and prompted participants to select their preferred response of autonomy or supervision along a spectrum of behaviors for each scenario. Between May 2019 and June 2020, surveys were distributed to participants using online survey software (Qualtrics, Provo, UT).

Focus group development and data collection
We developed a study-speci c focus group guide informed by survey responses (Appendix 3) . Questions were open-ended and followed up with prompts to elicit greater detail about participant's de nitions of autonomy and supervision, contributing factors, impact on work satisfaction, and perception of readiness for independent practice. Participants were invited by email to take part in focus groups between October 2020 and April 2021. Trainee and attending focus groups were held separately; all focus groups were digitally recorded.

Survey and focus group analysis
Survey responses were summarized using descriptive statistics. Trainee and attending responses were compared using chi-square tests or Fisher's exact tests. Analysis was completed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA). Survey questions that were open-ended were analyzed using thematic analysis. 19 At least two members of the research team independently performed open coding of transcripts using a data driven (inductive) approach. To ensure methodological rigor, the research team met at regular intervals to develop a code book and to resolve any coding discrepancies using negotiated consensus. 20 Codes were then grouped into higher order themes. Focus group data were professionally transcribed, veri ed for accuracy, and de-identi ed to ensure con dentiality and limit analytic bias. 21 We used thematic analysis to summarize focus group data.

Survey Results
Across all clinical sites, trainees perceived the current culture to be signi cantly more autonomous than attendings (Table 2). However, both attendings and trainees described an "ideal" culture as more autonomous than the current state, though trainees favored more autonomy than attendings. The largest difference between current and ideal state occurred at the quaternary referral center, which was perceived to be highly supervisory by both groups.

Focus Group Results
Focus group analysis of the ve core contributors to the balance of autonomy and supervision revealed a complicated dynamic. We identi ed subthemes within each core contributor to better understand the multiple variables that in uence the balance of autonomy and supervision in the inpatient clinical learning environment (Fig. 2). Representative quotes for each subtheme are summarized in Table 3. "When we spend very little time together all as a team, the attending ends up being the person with the most continuity in the team. And so, they sometimes have…the most information, and are in the most informed place to help make decisions. But I think that that gives us less autonomy, because we know less about the patients, potentially. And part of that is the discontinuity of the team." (resident)

Attending Contributions
The attending contribution to the autonomy and supervision dynamic was multifaceted, incorporating their practice patterns, re ections from their residency training, and experience navigating the various, and often changing, roles that they are expected to ful ll within the team and institution. There was a near universal intention of restraint in order to create space for trainee clinical decision-making, though execution varied signi cantly depending on attending risk tolerance for trainee mistakes and exibility in creating patient care plans. Trainees often were unable to distinguish discomfort with patient-related factors from discomfort with their own clinical skills, which seemed to exacerbate a negative reaction to perceived hypervigilance or "inappropriately" supervisory behaviors and amplify bene ts of attendings' intentional exibility in clinical decision-making.
Attendings' perspectives on the value of autonomy and supervision often related to experiences from their own residency training. Many attendings voiced support for resident autonomy due to perceived bene ts to their own clinical growth from autonomous practice in residency or were able to identify supportive supervisory behaviors that helped ease workload or navigate speci c challenges during their training. In general, attendings were motivated to emulate positive experiences and avoid distressing ones (most often related to inadequate supervision) as a guide to how they approached autonomy and supervision in their own practice.
Attendings described supervising inpatient resident teams as increasingly challenging, with a multifaceted and dynamic role that required adjustments even over the course of one rotation with a given team. They described feeling pressure to ensure high-quality patient care while also satisfying roles as an educator, supervisor, role model, and mentor. Many attendings and trainees identi ed that expectations for engagement with trainees varied by gender, level of clinical experience, and primary specialty (hospitalist versus subspecialist or outpatient physician). Female attendings expressed more di culty earning respect and trust from trainees than male counterparts, particularly when their resident was male. Universally, attendings and trainees reported that junior attendings tended to behave or be perceived as more "hands-on" and engaged than senior attendings. This behavior was hypothesized to be related to clinical uncertainty or anxiety, though many attendings cited perceived expectations from trainees as the reason. Trainees were less likely to identify level of clinical experience as a determinant of engagement, and more likely to identify primary specialty. Trainees reported experiencing greater autonomy with subspecialty and outpatient attendings, though this was not necessarily perceived as positive or intentional for their educational or patient care experiences.

Trainee Contributions
Generally, trainees expected signi cant autonomy in clinical decision-making, but they appreciated attendings who were responsive to their needs for support or guidance. Few trainees described negative experiences with autonomy; most examples of discomfort occurred from perceived overwhelming or inappropriate supervision.
Both groups highlighted trainees' expressed understanding of their own limitations (humility, communicating uncertainty, and openness to feedback) as a behavior that promoted autonomy. Participants agreed that self-assessment of one's own comfort level with patient acuity and complexity, utilizing other team members for their expertise and guidance, and being receptive to other perspectives were critical to developing trust and fostered further autonomy.
Additionally, both groups valued a trainee's ability to articulate their clinical reasoning and prioritize strong communication amongst the patient and provider team. This was particularly re ected in their ability to describe their thought process on rounds, give insightful and accurate feedback to interns and medical students, and proactively and collaboratively communicate with interprofessional team members.
Lastly, team leadership was identi ed as an important determinant of autonomy for both attendings and trainees. Residents described that their attentiveness to detail often translated to attendings offering more autonomy, and that interns' attentiveness to task completion often informed their own decisions about autonomy. Attendings mirrored this statement, noting that a resident's ability to effectively manage the team, pay attention to the details of inpatient care, and capture important details from the EHR were vital to establishing trust.

Patient Contributions
There was almost uniform agreement between trainees and attendings that attending supervision was appropriate in the context of high acuity and complex patients, particularly with co-existing clinical uncertainty. Trainees expressed appreciation and feelings of safety when they knew that attendings would support them and engage in clinical decision-making; attendings acknowledged the opportunity they had to empower resident learning and development, while also ensuring patient safety, in these contexts.
Both trainees and attendings described a generational shift regarding the care of at-risk and underresourced patient populations. Trainees noted that historically, patients who were affected by social drivers of health or without robust support systems were often those with whom they experienced the most autonomy; attendings reported similar experiences during their own training. However, trainees noted that junior attendings and hospitalists now take on a larger role in the care of these patients.
Attendings expressed strong feelings of responsibility and desire to help trainees navigate the complex healthcare system to promote improved outcomes for vulnerable patient populations.
Interestingly, patients with privilege and physician-patients were described as cared for holistically and collaboratively by the medical team, with both trainees and attendings citing patient and family expectations, professional courtesy, and streamlined communication as reasons for adjusting their behavior towards these patients.

Interpersonal Contributions
In addition to any individual team member characteristics or behaviors, the way in which team members interacted and communicated with one another, and how that interaction changed over the course of a shared rotation, was highlighted as a distinct contributor to the autonomy and supervision dynamic. Just as an attending's practice patterns often informed their approach to the attending role on the inpatient team, trainees had expectations for their own responsibilities. When there was overlap in these roles, strong communication ensured that team members felt valued and supported, rather than hindered and intruded upon. The most frequently described example of this overlap was "task sharing:" when an attending attempted to absorb tasks from the trainee. Attendings described task sharing with trainees when they felt they could be particularly helpful -either because of a speci c relationship with a patient and family that had formed, or because of perceived busyness of the trainee -or when they felt that team discontinuity put the attending in a position to ll multiple roles for several days in a row. Trainees had discordant opinions about task sharing; some expressed appreciation for attending willingness to step in, while others found it intrusive on trainee autonomy and growth. However, even with that discordance, both attendings and trainees shared that the impact on team dynamics was improved when expectation setting occurred at the beginning of a rotation and when there was proactive communication about task sharing and roles throughout the rotation.
Both groups identi ed the bene ts and importance of "dynamic supervision" based on the composition of the team that day (accounting for trainee days off). Many attendings described the perceived need to "scale up" supportive efforts during trainee days off, and that these experiences sometimes led them to feel more ownership over patient care than trainees. Trainees, while they acknowledged the attending's dynamic role, strongly preferred that attendings "scale down" their supervision when the team was whole to allow the resident to assume a team leadership role. Attendings expressed di culty adjusting levels of supervision when they perceived their role to facilitate continuity for patient care, though residents struggled with their role on the team when attendings did not achieve appropriate dynamism.
Lastly, there were discordant opinions between and amongst both groups over appropriate balance of autonomy and supervision in the context of patients with signi cant logistical complexity (e.g. multiple consultants contributing to patient care, complicated transitions of care, or previous di cult interactions between a patient and members of the care team). Some trainees and attendings highlighted the bene ts of attending involvement in these areas, including their ability to streamline communication between services, navigate a complex healthcare system, and model effective de-escalation and trust-building techniques. Conversely, many communicated concern about attendings and trainees over-prioritizing the ease of having an attending assume responsibility for these fairly common challenges in the inpatient environment at the expense of trainee development.

Institutional Contributions
These complex interpersonal interactions take place within an institutional context that informed the autonomy and supervision dynamic on the inpatient medicine service, speci cally the "culture" of the institution and residency expectations for trainees and attendings. Culture was di cult for participants to de ne but was described as tightly wound with the "hidden curriculum" of an institution and visible within the team dynamic. Both attendings and trainees noted that they mostly learned their roles by shadowing or experiencing them through their own training (if attendings trained at the same institution), thereby perpetuating whatever culture they had experienced previously. Conversely, new attendings described altering their practice in order to assimilate to the perceived values and expectations of their current peers and trainees. This seemed to act as a balancing measure for the impact of an attending's own training experience.
The residency program's organizational structure of the inpatient medicine rotation was noted to impact the balance of autonomy and supervision most prominently by contributing to team discontinuity due to call cycles and trainee days off. Both groups acknowledged that many of these changes were necessary to satisfy ACGME requirements and important to promoting trainee well-being. However, they also negatively impacted the functioning of the inpatient team. In fact, many of the issues mentioned aboveneed for dynamic supervision, changing expectations of roles within teams, and prioritization of continuity for patient care -were attributed to or worsened by team discontinuity. Both attendings and trainees expressed similar dissatisfaction with a team structure that was often fractured.

Discussion
In this study, we established that the current climate of autonomy and supervision was perceived as more supervisory than desired by both attendings and trainees. There was generally good agreement on the level of autonomy and supervision that was appropriate across both educational and patient care domains, except when admitting a new patient to the hospital. Subsequent focus groups were able to describe speci c features of attendings, trainees, and patients that contributed to the complex dynamic of autonomy and supervision, and how these groups are affected by interpersonal dynamics and institutional factors.
While there has been active research in trainee autonomy and supervision over the past decade, this study is novel in exploring the interactions of factors amongst groups, highlighting previously undescribed factors, and noting a generational shift in how the modern inpatient environment impacts these factors.
Entrustment has long been described as a contributing factor to resident autonomy, though is classically unidirectional -that is, assessed and provided from the direction of the supervisor toward the trainee. 22,23 However, in this study we found evidence of bidirectional entrustment, with trust being applied from trainees towards their attendings for supervisory behaviors, which is a novel understanding of this term. For example, trainees' assessment of an attendings' supervision or provision of autonomy in multiple domains in uenced how they developed trust in their attending: whether their behaviors were intentional or incidental based on their level of clinical experience, reactive to mistrust of the trainee, re ective of clinical uncertainty, or accommodative to the dynamic supervision necessary over the course of the rotation. This phenomenon of "bidirectional entrustment" has important implications in the modern clinical learning environment where attendings and trainees work in increased partnership rather than in a traditional hierarchy and rely on each other more to achieve patient care and educational goals.
This study described how team discontinuity as a result of residency's organizational structure has contributed to the erosion of the resident leadership role, and the challenge it creates for the attending to participate in "dynamic supervision." While team leadership is an important skill in the domain of "Interpersonal and Communication Skills" of the ACGME Core Competencies, 24 it has become more di cult for residents to meaningfully occupy this role when they may miss one or more days per week due to scheduled days off, post-call days (from overnight call), and other competing residency requirements. Some of this can be mitigated on the residency level by minimizing scheduling con icts with inpatient rotations, but residency programs should consider the impact of the organizational structure on trainees and their ability to contribute to the inpatient team when balancing this with other priorities. Attendings can and should continue to participate in dynamic supervision, but the numerous challenges identi ed in this study suggest that this is a stopgap measure to ensure that the resident has an opportunity to occupy a leadership role on the inpatient team, rather than a solution to the problem of team discontinuity.
Prior to this study, the importance of the multifaceted attending identity on the balance of autonomy and supervision in the inpatient environment was not well described. Lastly, this study demonstrated a generational and cultural shift that represents the evolution of the modern clinical learning environment since the regulatory changes of the early 2000's. Current attendings hold a unique perspective and exposure to both the traditional hierarchy and pedagogy of medicine, as well as the modern patient safety, health systems improvement, and medical education movements of the past twenty years. This was re ected in attendings' desires to emulate positive experiences from their training, as well as avoidance of negative experiences of feeling inappropriately unsupervised and unintentionally harming patients. This was also evident in how both groups described care for at-risk and under-resourced patient populations, who were historically perceived to be "opportunities" for more autonomous practice by trainees. 27 The younger generation of attendings -particularly hospitalistsdescribed an intentional effort to add value to trainee education and patient care for at-risk and underresourced patient populations through supervision and education around systems-based practice that would result in improved advocacy and outcomes.
In general, there has been a shift towards hospitalists as primary educators on inpatient medicine teaching services that has aligned with the trend towards hospitalists as primary providers for inpatients at many institutions, including ours. 28 Participants noted several bene ts of this shift, including more predictability in day-to-day activities like rounds, better adherence to health system improvement efforts, and improved education and role modeling in patients with complex inpatient and transition of care needs. Additionally, trainees expressed more trust in the autonomy that was granted by hospitalists compared to subspecialists or outpatient physicians who rarely attended on inpatient teaching services.
Hospitalists were also more likely to err towards inappropriate direct supervision, consultant communication, or task sharing behaviors than their subspecialist or outpatient counterparts, and could readily intrude on resident autonomy if not mindful and restrained in these practices.
This study has a few limitations. First, while we surveyed and conducted focus groups at three distinct inpatient sites serving different patient populations, all sites are a liated with a single institution. The attending groups are distinct from one another, but the trainees rotate through all three sites. Second, this study was conducted during the height of the COVID-19 pandemic, which we believe negatively impacted recruitment for participation and limited our ability to do in-person outreach at all three sites. We believe that despite these limitations, the results of this study are adequately descriptive and generalizable.

Conclusions
The balance of autonomy and supervision on the inpatient medicine service is complex, multifaceted, and dynamic. Trainees and attendings agree that the clinical learning environment should generally favor resident autonomy, and that the current environment does not achieve the ideal balance. When considering the factors contributing to this complex dynamic, one must consider the attending, resident, patient, interpersonal, and institutional variables at play. It is also important to recognize that these factors interact with one another: entrustment is bidirectional between attendings and trainees, and team discontinuity is an important disruptor to the preferred model of resident team leadership. Lastly, it is important for team members -including attendings themselves -to acknowledge how their identities contribute to the autonomy and supervision balance, and that the cultural shift towards more hospitalist attendings and increased attending accountability for patient safety and health system outcomes can further impact trainee autonomy on the inpatient medicine service.

Declarations
Ethics approval and consent to participate: Ethical approval for the study was waived by the University of California-San Francisco IRB (# 19-27849). All methods were carried out in accordance with relevant guidelines and regulations. Informed consent was obtained from all participants.
Consent for publication: Not applicable.
Availability of data and materials: The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Comparison of attending (n=109, hashed lines) and trainee (n=81, solid lines) responses to attending contributions to various aspects of rounds by 5-point Likert scale. Attendings and trainees generally agreed on the attending's role in patient safety, education, acute decompensations, review of EHR data, and consultant communication. Attendings perceived a greater need for supervision in newly admitted patients than residents did (p=0.04).