Autonomy Is Desired, Entrustment Is What Matters

In this month’s journal, Allen et al describe resident autonomy as a core subconstruct of the self- determination theory, which reports the factors that motivate adult learning.”^1–4  Their impressive, multicentered collaboration involved diverse stakeholders working to generate adaptable change strategies aiming to improve senior residents’ perceived autonomy, as measured by a novel “resident autonomy score” (RAS). The “autonomy toolbox” promotes 3 main strategies:

1. autonomy stakeholder development sessions;

2. normalizing shared expectation-setting discussions between senior residents and their attending physicians; and

3. offering recommendations for senior resident-led independent rounding.

The authors created adaptable resources to support these strategies, including videos, preservice agreement forms, just-in-time e-mail templates, and badge buddies. Their multicentered and multidisciplinary approach to cultivating resident autonomy provides generalizability to facilitate spread to additional programs.^5,6  As a result, we anticipate readers to consider local initiatives toward similar aims.

As readers ready themselves for local efforts, we write to encourage reflection on the RAS, but more broadly, the goal of resident autonomy as the primary objective. To this end, we offer 2 interrelated considerations on the outcome of perceived autonomy versus entrustment, a validated metric that balances context-specific learner autonomy with appropriate supervision.^7–24  First, we provide a reflection on the validity evidence behind each metric. Next, we share a case-based example, emphasizing the potential benefit of entrustment.

First, we want to highlight some validity considerations specific to the RAS. Following Messick’s framework,^13,14  the RAS lacks discussion of the score’s consequence validity or the “difference” the score creates for the score recipient.¹³  Congruent with self-determination theory, improving perceptions of resident autonomy may make for happier, more fulfilled and motivated residents, but it is unclear how the RAS relates to a resident’s actual readiness for independent practice. In other words, does a high RAS mean a resident can or should independently provide care to a hospitalized child? As presented, the RAS does not complete the critical, interpretation-use validity argument, which requires a predetermined consequence in response to an achieved score.¹⁴  Further, the monthly average RAS among a small sample of senior residents lacks the additional psychometric calculations in support of internal structure validity (ie, reliability).¹³  Finally, although the authors included important balancing measures of the family and nursing perspectives, further exploration of overall patient care and safety are vital before application of the RAS in the clinical setting.

We commend the authors on their presentation of validity evidence related to the RAS’s content, response process, and relationship to other variables.^13,14  Content validity (ie, how the score represents the construct)¹³  was convincing because the RAS was based on a previously reported autonomy inventory and the composite score factored in the divergent perspectives of both senior residents and their attending physicians.⁷  Response process validity (ie, the relationship between the intended construct and the thought processes of the subjects)¹³  was evident from the multicenter collaboration-generating interventions that improved 5 of the 6 aspects of the RAS, thus an observable connection was made between the interventions and many RAS components. Relationship to other variables (ie, how scores measuring 1 construct compare with those measuring the same or similar constructs)¹³  was appreciated when the authors revealed an anticipated negative inflection of RAS at the start of the academic year when new senior residents lead teams.^8,17  The other variables generated by project stakeholders that were perceived to impact the RAS, such as “attending physician qualities,” “resident skills and work ethic,” and “patient acuity and complexity” provide excellent relationship to other variables’ validity evidence, because they overlap with many of the variables reported to impact the related construct of entrustment.^13,18–20 

This connection between autonomy and entrustment transitions us to our larger point of reflection: Is the goal in graduate medical training perceived resident autonomy or should it be entrustment?

We strongly support entrustment, or borrowing directly from the autonomy toolbox dedicated discussions around entrustment alignment between resident and attending as a preferred outcome measure in future similar initiatives. This would allow for deeper consideration of the specifics of the clinical contexts, skill sets of the resident, and experiences and style of the attending. After all, entrustment balances the learner-centric importance of autonomy^4,7  with supervision, which is ultimately how academic health care systems can ensure safe, patient-centered care.^21,22,25 

In his sentinel work, Olle ten Cate merges the construct of entrustment into the medical field through the entrustable professional activities (EPAs), or “doctoring tasks.”²²  For a given EPA, such as leading an interprofessional health care team,²⁵  a cluster of workplace-based observable competencies are required of the learner to demonstrate to advance to autonomous, independent practice.^{8,9,12,16,17,22,26} 

Measuring EPAs through entrustment scales have demonstrated strong validity evidence in all elements of Messick’s framework.^8–11,17,24  Put simply, if one reflects on an entrustment scale for a given EPA, a clear interpretation-use argument emerges, because a high rating indicates that the preceptor, based on real workplace-based observations of relevant competencies, believes the resident is “ready for independent practice.”^25,27 

To illustrate the role of entrustment, we turn our discussion to a more practical, case-based example to provide our rationale for the promotion of entrustment over perception of autonomy as the ultimate goal in graduate medical education:

You are the hospitalist supervising a resident that is admitting a 4-year-old with a first episode of wheezing. The resident perceives a high RAS in managing this patient, autonomously constructing a care plan on the basis of the hospital’s bronchiolitis clinical care pathway without calling you for oversight (they present a 1-line summary of the case over the phone after the admission). Later in the night, the patient has a sudden decompensation and requires emergent transfer to the PICU for advance airway management and placement of a chest tube to evacuate a pneumothorax. Upon review of the case, the resident ordered a weight-based high-flow nasal cannula flow rate she used previously with infants with bronchiolitis. This clinical decision was deemed to be the root cause for the outcome experienced by the patient.

If the entrustment alignment framework was used by this resident, they might self-assign “trusted to execute without supervision to improve member and team performance and supervise others.”²⁵  However, given that the current pediatric respiratory surge is like nothing the attending has experienced in her career and noting a general lack of senior resident experience in managing respiratory distress because of coronavirus disease 2019 census changes, the attending may select “trusted to execute with supervisor occasionally present to provide advice.”²⁵  During the discussion between resident and attending over entrustment alignment, the attending would engage the resident in determining the needed supervision before they went home for the night. This would set an expectation that a middle of the night “1-liner” was not sufficient to ensure safe and effective care. That overnight discussion would likely highlight concerns about the direct application of a bronchiolitis pathway to an older child and perhaps have led to an adjusted management plan factoring in the potential devastating consequences of prolonged air trapping in children with asthma.

We applaud the authors’ effort moving education forward and believe many of the strategies developed to promote autonomy will also improve resident and attending entrustment alignment. With entrustment alignment as the ultimate goal, the above resident may have called earlier in the patient’s course or provided more details of their initial care plan. This would allow the attending to anticipate unforeseen negative outcomes without compromising the resident’s growth toward independent practice. After all, this resident did not know what they did not know and was overconfident in their perceptions of their abilities to autonomously lead the health care team.

In sum, the construct of entrustment which measures a given EPA provides a framework to guide the appropriate level of autonomy on the basis of learner development. It also ensures our greatest responsibility, the health and safety of the patient. Entrustment ensures that autonomy is in balance with additional considerations such as resident factors, supervisor factors, and the clinical context.^18,22  As such, the autonomy toolbox may motivate adult learning, but an “entrustment toolbox” is likely best for patient outcomes, their safety, and of utmost importance, still prioritize and motivate the residents’ adult learning journey to independent practice.