Learner handover: Perspectives and recommendations from the front-line

The principal investigator (VD) contacted the directors of five large PGME programs (Family Medicine, Emergency Medicine, Pediatrics, Surgery, and Internal Medicine) at McGill University by email and invited them to circulate the recruitment email to their clinical supervisors. The program directors reported circulating our email to an estimated 700 supervisors (although this number does not account for potential duplicates, i.e. supervisors affiliated with multiple departments). Recruitment also occurred through presentations held at departmental meetings. The estimated time commitment was 30–40 min in total to complete the questionnaire and experiment.

Those interested in participating were provided with a link to the study (hosted through LimeSurvey [18]). Following consent, participants completed the experimental component of the study (reported in [17]) and the questionnaire (reported here). All participants who provided their email addresses were given a 100 Can$ (approximately 75 US$) Amazon gift card. All response data were kept in a separate file from participants’ identifying information.

The questionnaire included open-ended items (generating qualitative data) and closed-ended items (generating quantitative data) developed by the research team. Items were reviewed for relevance and clarity by the full research team, which includes several members from the target population of clinical supervisors. The questionnaire was not pilot tested. To limit participant cueing, we used five open-ended questions designed to gather narrative responses regarding participants’ experiences with, and perspectives on learner handover. The closed-ended items included one Likert-style item assessing a participant’s position on learner handover (“Overall, what is your position regarding the sharing of information regarding residents’ previous performance across rotations?”), with the following five response options: Strongly Oppose, Somewhat Oppose, Neutral, Somewhat Favour, Strongly Favour. The remaining closed-ended items comprised demographic questions targeting age, gender, clinical speciality, and the number of years spent supervising and assessing learners. The complete questionnaire is available as Appendix 1 of the Electronic Supplementary Material.

In line with our post-positivist stance, we involved multiple team members in the coding process, recognizing that coders bring prior knowledge and beliefs to the table. In order to minimize the influence of individual bias, two independent coders (STG, NEP) familiarized themselves with the survey responses. For each open-ended survey question, the coders examined responses in batches of 10–20 respondents. Each coder independently broke down a participant’s answer into either one or multiple distinct units of meaning. Then, they assigned specific codes to every unit of meaning, and reviewed all the codes until they agreed on the best code(s) to use to represent each different unit of meaning. They applied this iterative coding process to all responses across the five survey questions. In cases where they could not reach consensus, they consulted a third coder (VD). Once they had established an initial coding scheme, the coders grouped independent codes into overarching categories and subcategories identified directly from the data and by drawing on existing literature [6‐10, 12‐14]. Once a complete round of coding and categorization was finalized, the third coder (VD) independently reviewed all of the assigned codes to ensure consistency in terminology and coding practice. Our research team discussed all codes and subcategories until we identified a final group of core categories. Such consensus-seeking aimed to leverage the different perspectives of team members and reduce the risk of misinterpretation.

We chose to examine the frequency of each code in order to explore how often particular opinions or concerns were mentioned and to facilitate comparison of code frequency with self-reported support of learner handover [24]. The quantification of qualitative data is possible when using a qualitative descriptive stance [20], when all participants are asked the same questions, and when the sample size is sufficiently large [25]. To do this, we computed the percentage of participants who mentioned each code, to gauge how common each mention of a code was within our relatively large (compared with most qualitative studies) sample [24]. To examine patterns of codes among participants with the same and differing positions on learner handover, we created a matrix with codes as rows, position on learner handover as columns, and the number of participants mentioning each code as cells.

Seventy-two supervisors participated (participant characteristics are described in Tab. 1). One participant only completed the closed-ended questions. Response rates for open-ended items ranged from 93–99%, yielding 350 narrative responses out of a potential total of 360 (72 participants multiplied by five open-ended questions).

Positions on learner handover varied across participants on the Likert-style item ranging from Strongly Oppose to Strongly Favour (Tab. 1), with a majority in favour but one third of participants opposing learner handover. Position on learner handover was significantly associated with only two variables: supervisors with more experience in supervising and assessing learners were significantly more favourable towards learner handover (r_supervising = 0.25, p < 0.05; r_assessing = 0.29, p < 0.05).

Appendix 2 of the Electronic Supplementary Material includes a breakdown of all the codes, subcategories, and categories. In examining our matrix of code frequencies by level of support for learner handover, we found no major differences in the codes used across participants based on their level of support of learner handover; therefore, our main findings are presented collapsed across stated levels of support. The percentages reported below refer to the number of participants who mentioned a specific code at least once. The term ‘resident’ refers to postgraduate learners, ‘rotation’ to a clinical placement.

Conceptions of learner handover appeared to vary among participants, with some referring to a ‘learning plan’, suggesting a specific format and process of reporting, and others referring to the sharing of current end-of-clinical placement assessment reports without additional features. Most participants had no formal experience with learner handover but some described experiences of informal forward-feeding (e.g., conversations between supervisors), which they felt had led to positive and/or negative outcomes.

The benefit of learner handover mentioned by most (n = 50, 69%) of the participants in our study was the potential for more individualized learning: “with earlier recognition of potential growth areas of the resident, the learning environment can be shifted to accommodate his/her learning needs” (P2).

Other reported benefits included enhancing the assessment process. Many participants (n = 34, 47%) mentioned how learner handover could improve efficiency: “the faculty [could] spend less preliminary time in assessing the resident [so they] can get on with more specific training in needy areas” (P2). Some participants (n = 21, 29%) also felt that it would improve assessment quality through the monitoring of learner progress. Others felt that it would generate more well-rounded assessments (n = 8, 11%), and reduce the risk of ‘failure to fail’ (n = 6, 8%) (although two participants, 3%, believed that it could increase this risk “in a context where ‘forward feeding’ is permitted, why ‘fail’ someone or give someone a ‘borderline pass’ if we know that the problem can be passed on to someone else down the line to deal with it” (P27)).

A few participants (n = 5, 7%) felt that it would be useful to ‘alert staff’ to weak learners requiring more oversight. A few others (n = 7, 10%) identified how the close monitoring of weak or problematic learners could improve patient safety.

Concerns about how learner handover could lead to labelling and stigma were raised by many participants (n = 33, 46%): “a resident [might be] pigeonholed as being a negative trait (lazy, unprofessional). It biases the new supervisor. The resident does not feel like he gets a fresh start” (P16). Most participants (n = 62, 86%) were concerned that sharing inaccurate information about a learner could bias subsequent assessments, impeding their validity: “a bad evaluation in a previous rotation will likely affect the next clinician’s impression. The earlier evaluation might not be an accurate assessment (i.e. due to evaluator’s biases, due to one rotation of poor performance)” (P58).

Some participants (n = 13, 18%) expressed how learner handover may negatively impact learner training and available educational opportunities. Specifically, concerns about halo effects, self-fulfilling prophecies, and confirmation bias were raised: “strong residents tend to be let off easy whereas residents with difficulty will more likely be scrutinized” (P34).

A few participants (n = 6, 8%) worried that biased assessments would “[lead] to an environment where learners may fear that they have to be perfect because the information will be fed forward” (P22).

Of note, potential legal challenges were raised by one participant.

Participants disagreed on whether the forwarded information should include content related to learners’ weaknesses, strengths, or both. Many participants (n = 26, 36%) felt that handovers should contain information related to a resident’s weaknesses, for example: “the details, specifics, or justifications for weak marks (or fails) are not pertinent to the next rotation but […] areas where the trainee and the supervisor need to focus are best shared” (P46). Regarding a resident’s strengths, one participant described how “it is also useful to feed forward information about residents who are strong, as this helps to know when residents can be given extra trust/responsibility and to challenge these residents to take on more advanced roles” (P47). However, concerns about the sharing of a learner’s strengths leading to halo effects or supervisor and/or resident complacency were also raised.

The use of free-text survey items in medical education has been criticized for lacking depth, richness, and rigor as they often do not contain ‘context, personal meaning, emotional and social nuances and layers of detail’ [33, p. 347]. Further, our open-ended items were developed iteratively by our research team, but were not pilot tested. We adopted a methodology of qualitative description, which is purposeful and often used to obtain straightforward answers to questions that can then be used to inform practice and policy [19]. Given that research on learner handover is limited, we aimed to simply gather a range of opinions towards learner handover and provide context for the responses to the closed-ended survey responses. Future studies, using interviews or focus groups, could examine supervisors’ beliefs about learner handover in more depth. We also recognize some of the opinions reported were mentioned by a small number of participants in this study. However, our goal was to not only capture the most popular opinions regarding learner handover but to identify the full range of beliefs about learner handover and ideas for successful implementation. The response rate at our institution was low (an estimated 10%) which may mean that our findings are representative of a specific type of supervisor thus limiting our ability to generalize our findings to a broader body of clinical supervisors. Participants agreed to commit approximately 30 min to the study as a whole, which may have led us to recruit supervisors who were less busy or who had an interest in medical education. Compared with a shorter study at the same institution, but in different clinical disciplines [34], we found that our participants were more likely to have growth mindsets (99% versus 88% for clinical reasoning, 71% versus 35% for empathy—data reported elsewhere [17]) than participants in the shorter study, suggesting a potential selection bias. Our sample was nonetheless diverse in terms of its demographic characteristics and its positions on learner handover. Further, our sample included clinical supervisors from one Canadian institution, and our findings may be influenced by the institutional policies and culture present at this institution and may not generalize well to other learning contexts. Specifically, at the time of this study, our institution had a strict policy prohibiting learner handover amongst clinical supervisors. As such, only a few participants in our study reported experiences with learner handover and, among those who did, their experiences were informal. Given the novelty of learner handover, especially within Canadian medical education programs, we did not anticipate being able to collect detailed accounts of perspectives on personal experiences with learner handover. Thus, these findings provide baseline data on the range of perceptions that must be considered in preparing clinical supervisors to take on learner handover in a meaningful manner. Our institution has implemented Focused Learning Experiences (FLEX) periods tailored to learners’ educational needs involving coaching and role mentoring, thus working towards normalizing the occasional need for remediation. This may have also influenced our participants’ perspectives on the developmental nature of competence and the appropriateness of learner handover.

While understanding supervisor perspectives is critical to successful implementation, discussions regarding learner handover would benefit from robust evidence regarding the observed benefits and harms of learner handover as well as factors influencing their likelihood and extent. The majority of available experimental studies can be found within the educational, social, and organizational psychology literature [5, 35‐38], with a small, but growing, literature in health professions education with mixed results [17, 39]. Further studies are needed to understand the potential positive and negative impacts of learner handover in medical education settings. The evidence should then be used to inform and engage stakeholders, including clinical supervisors, in collaboratively designing how learner handover should be implemented and carefully monitored for maximal benefits and minimal harms.