Challenges for conducting and teaching handovers as collaborative conversations: an interview study at teaching ICUs

Study sites were the ICUs of our university medical centre (UMC; hospital 1, 46 beds in four locations) and two district teaching hospitals (hospital 2 and 3, with 16 and 13 beds respectively), all in the Netherlands. At all sites, six to ten patients are handed over three times a day at designated times with protected time: 30 minutes at hospital 1 and 3, and 60 minutes at hospital 2. Handovers are given predominantly by supervised residents. All scheduled staff of the incoming and outgoing teams are present, varying from between 3 and 11 physicians in total, of which approximately half are trainees. At the two general district hospitals, handovers take place in a designated quiet room. At the UMC, locations are designated rooms or patients’ bedsides.

Semi-structured interviews with individual participants (12 intensivists, 3 fellows and 6 residents) were conducted between October 2013 and April 2014. The interview protocol was targeted at identifying the timing and manner with which macro-cognitive functions are to be served within ICU shift handovers, and the factors that affect or impede their engagement in these interactions. Participants were first asked to describe what they perceive to be the functions of the handover, in order to identify individuals’ underlying values for the handover process. Then they were asked to describe the circumstances in which the importance of each function might be triggered. Next, they reflected on whether to adapt their presentations to accommodate certain circumstances. In order to explore the perceived requirements for two-way interactions, we then asked them to describe the behaviours of both receivers and senders they find of use to the functions mentioned earlier in the interview. In this section, we also asked for factors that challenge teamwork interactions and for their conceptions of the boundaries to what must and can be achieved in the handover. During the course of the interviews, probes were added based on preliminary analysis of initial interviews. Interviews lasted 40–60 minutes and were transcribed verbatim.

A qualitative ‘conventional’ content analysis was performed on the data following the procedural steps as summarized by Hsieh and Shannon [30]. We felt that the divergence in the literature on handover education and the limited understanding of the complexities of ICU shift handovers justified an inductive approach.

Interview transcripts were de-identified prior to the analysis. Transcripts were first read and segments of interest were then coded using keywords derived from the segments, using ATLAS.ti (Scientific Software Development, Germany). These codes would function as short summaries of the segments of interest. Codes were generated and applied as appropriate across interviews and continued to be generated as necessary across the coding session. These steps were repeated after every three to four interviews to allow insights to develop and have probes added for subsequent interviews. This procedural element from the grounded theory approach [31] was added, given the method’s iterative approach to gathering insights and model building. In order to make sure that the coding process was performed reliably, inter-rater reliability was determined by calculating the percentage of consistently coded segments of the first rater with a second rater for the first three interviews (80.4% of 347 segments). After agreement was reached on how to complete coding, the remaining interviews were coded by the first author. The codes from all interviews were then sorted into categories (e. g., receiver behaviours; team factors) and grouped into meaningful clusters (i. e., by handover function). The codes and the original interview segments were reviewed to identify key messages, nuances and contradictory perspectives. Summaries of the analysis were reviewed by a panel of the interviewed respondents for internal validation.

At the UMC, participants volunteered on the basis of email invitation. At the two district hospitals, the heads of department introduced the study to their staff and provided us with an interview schedule with volunteers. At least two intensivists and two residents were sampled from each study site to maximize the diversity of perceptions. Participants were added until interviews failed to reveal new insights. Participants gave written informed consent. Ethics approval was waived by the institutional review boards of the UMC Groningen and the district hospitals.

Whereas the function of transferring understanding of the case and the plan from one team to the next and the coordination of tasks across shifts is the same for every handover, situational differences affect physicians’ perception of what is required to transfer understanding successfully. One such factor is case complexity and case uncertainty. Complex and uncertain cases require a more comprehensive presentation including arguments and explanations so that colleagues can reflect and fully understand the problem. Also, more elaboration or details may be given if they are functional to illustrating the gravity of the situation.

However, respondents did note that finding the appropriate balance with regards to the amount of elaboration can be challenging (see Tab. 2 for examples of interview segments). Too little detail may leave open questions about the diagnostic theories and how they led to the treatment plan, whereas too much may obscure the main message and leave insufficient time available for other patients.

Whether different expectations between individuals result in difficulties for establishing effective interactions varied. In some hospitals (particularly at the smaller hospital 3), respondents said that they did try to anticipate the receivers’ preferred level of detail, adjusting their presentation to whether the receiver tends to be more concerned with the details of plan execution or to be more reflective. Others (particularly residents in hospital 1) reported being confronted with frustration and impatience if they were perceived as being too detailed. These differing preferences among colleagues were reported to affect residents’ learning: residents receive varying and contradicting feedback regarding preferred handover behaviours and content across supervisors.

Another important factor that affects how physicians communicate information is the experience of receivers: the level of abstraction in the explanation and the specification of actions should be tailored to the receivers’ knowledge base and experience. A challenge for larger teams (as at our hospitals 1 and 2) is that it is not always possible to fully disclose all the details residents may need or to fully include residents in discussions, because of time constraints. For instance, if, in complex cases, discussions are needed, the educational aims of the handover are set aside. Residents are aware of these constraints: they would look things up or save questions for after the handover in case of time pressures or if they feel the atmosphere is not conducive to questions. Furthermore, residents consider whether their questions will benefit the rest of the audience.

Respondents felt that joint reflections and discussions could benefit the patient, particularly in rare, uncertain or complex cases. Associated receiver behaviours included cross-checking, asking for senders’ reasoning, and introducing alternative perspectives. Associated sender behaviours included communicating alternate considerations, doubts or hunches and reflecting on information that is indicative and counter-indicative of diagnoses.

Although such reflection-initiating behaviours can potentially help all members to understand the situation, help catch errors and decide how to proceed, respondents also noted the drawbacks of initiating discussions (Tab. 3). Drawbacks included that discussions might drift away from the main problems and that less time is left available for other patients. The relevance of the discussion to patient outcome was considered an important indicator of the appropriateness of the behaviour. It was also argued that discussions in the handover do not necessarily add new perspectives to those that were generated during consults in the previous shift. Respondents were aware that colleagues held differing views on the functionality of reflection in the handover, and noted that getting the other to join in collaborative reflections would sometimes require some determination. With regards to decision-making, it was often suggested that, given that problems are often too complex to solve in the handover, it is sufficient to inventory colleagues’ perspectives and save actual decision-making for the multidisciplinary round.

Similarly, respondents were not unanimous with regards to the utility of voicing suspicions. Whereas some argued that omitting suspicions could result in dangerous situations when complications are not anticipated, others reflected that communicating suspicions may bias an otherwise fresh perspective. On the other hand, some respondents reported the—limited—occurrence of ‘overly polished’ reports, in which senders provide a self-consistent, tidy report of what was done and why it was justified. Such reports would benefit senders’ personal agendas of appearing more competent, avoiding discussions or of ‘selling a point of view’, but would also cut short discussion that might otherwise contribute to a more positive patient outcome.

Another challenge to engage in joint reflections was a perceived difficulty to ask critical questions while maintaining a positive atmosphere. Questions can very easily be misinterpreted for their intent, as was illustrated in examples where senders felt they were being accused of poor judgment, rather than being invited to enhance shared understanding.

Few respondents reported that teaching (e. g., asking questions to stimulate residents’ knowledge synthesis) was a formal function of the handover (Tab. 4). Instances were also reported in which handovers were used to learn as a unit or to strengthen shared values, for instance, when something had almost gone wrong. Both forms of learning were reported to occur only if the number of patients, the available time and the number of attendants for whom the teaching would be relevant allowed it. As a consequence, residents’ learning was described as a by-product of the handover occurring, for instance, by listening to staff’s discussions. Moreover, the educational aims of the handover are set aside if, for instance, important discussions are needed. Teaching was more likely to occur during handovers in the evening, where fewer staff members were present and the workloads for the upcoming shift were likely to be lower. In hospital 3, it was also said that the small group size (3–4 physicians) provided a low threshold for teaching to occur.

We found that physicians adapt their presentations in order to accommodate complexity, ambiguity or uncertainty in cases: physicians discuss in more detail their assessment of the patient’s problems and how problems are interrelated; their interpretation of diagnostics; and of how treatment success might relate to the diagnoses. Such presentations include more explanations, reflections and details than those of simpler cases. This aligns with previous studies in which assessments of complex and ambiguous cases consistently differed from those of simpler cases [13, 18]. Our findings resonate with the suggestion that handing over patients is more than presenting ‘facts and figures’ [13] and rather requires flexible ways of telling the story [10, 17‐19, 32], skilful verbalization of one’s clinical reasoning and careful expression of uncertainties and options [13, 18] and the provision of details ‘on demand’ [25, 29].

Our finding that variations in handover presentations are a necessity could be seen to raise questions about the current gravitation towards standardized communication structures. A matter of concern is that too rigid communication schemes may fail to appreciate the need for such variation [17, 24]. This notion was confirmed in a recent evaluation study of the I‑PASS mnemonic educational program: residents experienced a challenge between the program’s focus on mnemonic adherence and their ability to use their own judgment in differing circumstances [33]. On the other hand, junior residents can fall back on standardized communication schemes if they lack a more holistic understanding of a patient’s case [8, 17, 29]. Our findings could therefore be seen to imply that teaching a basic structure is a starting point rather than the endpoint. Supplemental training should address flexible narrative approaches that help to synthesize clinical information, express clinical reasoning, and carefully give details at points in the story where their relevance becomes clear.

In addition to transferring understanding, our respondents valued shift handovers as opportunities for jointly reflecting on the patient’s situation and for jointly orienting on future decision-making and planning. These findings are consistent with previous notions that shift handovers in complex settings are an important means to enhance the resilience of patient care [5, 13, 25, 27, 34‐36]. For instance, a recent study showed that cross-covering ICU fellows were more vigilant and made more decisions when compared with continuity-of-care fellows, and thereby reduced mortality [37]. The authors suggest that “the balance between medical errors from handovers and the benefit of a ‘second look’ seems to favour the latter” [37].

Handover training in the ICU should thus advance important non-technical skills. Currently, handover programs are often limited to receivers’ behaviours that help enhance reliability, such as read-backs [8, 12, 38, 39]. Fewer programs target skills for enhancing handovers as a ‘resiliency activity’. Our findings add to the growing case for handover training also addressing skills of exploring the previous team’s sense-making, introducing concerns, suggesting alternative options and discussing anticipated problems.

We found that certain situational factors, such as high case complexity, serve as helpful cues for physicians to determine communication needs, as has been suggested in previous work [18]. However, we found that the simultaneous influence of other factors creates dilemmas for physicians with regards to what must and can be achieved in the handover. Often, dilemmas involved trade-offs between efficiency and thoroughness due to time pressure or a high number of patients to hand over. This trade-off was often reflected in physicians’ tendency to prioritize the function of information transfer over joint reflections or discussions and, particularly, education. However, the pressure to be concise and clear-cut may even hamper the goal of allowing all members to fully achieve understanding of the case.

That colleagues hold misaligned expectations might be explained by the theory that physicians make their own trade-offs depending on how they perceive the factors within the context and how they value the benefits and drawbacks of behaviours within the situation [40]. One way of better coordinating handover dialogues may then be for teams to communicate more explicitly about the desired outcome of the conversation. Team training could be targeted at such interpersonal coordination skills. Also, team training could facilitate consensus finding regarding the boundaries for handover dialogues. A number of issues seem to be of particular interest: systems for identifying which patients might benefit from more extensive discussions; systems for allocating time across patients; and how joint reflections and discussions during handovers add to previous consultations and rounds.

The flexible and variable conceptions of intensivists may leave residents uncertain with regards to best practices. Residents observe a wide range of handover styles and receive feedback that may be contradictory. Without reference to consideration of, for example, the characteristics of the situation or the benefits and drawbacks of certain behaviours, residents will not fully appreciate the values that underlie a behaviour or how the selection of information may or may not apply to other handovers [41].

Our finding that learning was more often seen as a by-product is in stark contrast to views that handovers are ideal platforms for case-based learning [42, 43]. On the other hand, our respondents’ hesitation about engaging in teaching activities included careful consideration of patients’ needs and practical constraints of time. This hesitation closely resembles findings from a survey among intensivists at academic teaching ICUs that increased clinical workloads are associated with insufficient time for patient care and teaching [44]. More efficient ways of incorporating education into daily care activities are needed [45]. Shift handovers may offer one such opportunity because they involve cycles of action, reflection, planning and adaptation within care trajectories that closely resemble learning cycles such as Kolb’s [42].