Introduction

Physicians in acute settings are faced with unpredictable circumstances, diagnostic uncertainty and unknown aetiology of symptoms (Croskerry, 2014; Kovacs & Croskerry, 1999). Therefore residents face the challenge of developing a set of thinking and acting habits that will help them cope with uncertainty and form sound clinical decisions under unpredictable conditions (Elia et al., 2012).

Given the fast-paced and uncertain nature of the ED, novice decision-makers are more prone to cognitive errors than more experienced physicians (Croskerry, 2014). To counter these errors, researchers and practitioners have developed decision aids, such as algorithms, guidelines, and probability programmes (Costantino et al., 2014). However, these aids have their limits (Schriger et al., 2017). Hence there is a need not only to focus on the decision aids, but how residents can be scaffolded so that they develop competencies in clinical decision-making. Therefore, the objective of this study was to analyse residents’ adaptive practices in the emergency setting. Specifically, we focused on exploring how individual and contextual factors might impact residents’ engagement in adaptive practices when diagnosing patients in the ED.

Medical expertise

Medical expertise can be described as proficient pattern recognition, which involves the use of hypothesis testing strategies. According to Elstein and co-workers’ classical view, expertise builds on the premise that good decision-making is a function of accumulation of knowledge and experience, and that difficulties in problem solving are due to insufficient knowledge and experience (Elstein & Schwarz, 2002; Elstein et al., 1990). Medical expertise is contextual and content specific (Elstein, 2009), and clinical judgement is vulnerable to biases and depends on heuristics (Norman, 2009; Norman & Eva, 2010). However, this classical understanding of expertise neither accounts for adaptability amongst novice physicians, nor concerns itself with the application of knowledge. In a rapidly developing healthcare system, adaptability and creative thinking have become imperative (Gube & Lajoie, 2020; Mylopoulos et al., 2016). The resident needs to prepare for an uncertain future, and as pattern-recognition is content and context-specific, it only prepares the learner for what is known at any given moment. It does not, however, prepare them for how to learn new skills or knowledge, which is important in ever-changing settings (Barnett, 2012). It is therefore important to understand how knowledge is organised and used (Eva, 2005), and how to promote adaptability amongst novices (Gube & Lajoie, 2020; Schwartz et al., 2005; Zimmerman, 2013).

Almost forty years ago, the concept of adaptive expertise was proposed by Hatano and colleagues (Hatano, 1982; Hatano & Inagaki, 1986). Based on Piaget’s epistemology of cognition as adaptation, assimilation and accommodation (Furth, 1981), Hatano and colleagues described a difference between a person’s procedural knowledge, which is knowledge of how to perform specific procedures, and conceptual knowledge, which is the act of giving meaning to procedural knowledge (Hatano & Inagaki, 1986, p. 2). This ‘giving meaning to’ can be used by the person to understand when and how to apply procedural skills by forming ‘mental models’ “…with which people can run mental simulation, and thus make predictions/explanations about an unfamiliar object/situation…” (Hatano & Inagaki, 1986, p. 2). According to Hatano (1982), as opposed to the routine expert, the adaptive expert can also adapt and invent new procedures to accommodate an unexpected situation. Importantly, adaptive expertise is the ability to both appropriately apply past procedural knowledge and develop new understandings based on conceptual knowledge when needed (Hatano, 1982; Hatano & Inagaki, 1986). There is a complementary balance between routine and adaptive practices (Bransford & Schwartz, 1999) where adaptive experts are assumed to be exceedingly proficient in both (Schwartz et al., 2005). Routine expertise relies solely on acquired procedural knowledge with no accommodation when the context demands a new solution. Currently, educational programmes are often structured in ways that promote routine practices of expertise. As a result, in order to fully understand how we can support the development of adaptive expertise during resident education, we were interested in gaining a better understanding of adaptive practices. Accordingly, in this study, we dichotomise routine and adaptive practices, and routine practices are therefore seen as disengagement of adaptive practices.

Hatano (1982) also suggests that curiosity and flexibility act as mediators of adaptive expertise. Conceptual knowledge can be gained through trial-and-error by simplifying the problem at hand. For example, when new residents associate an inverted position of a leg with the diagnosis of a fractured pelvis, they know the concept of a fractured hip and how this presents without an x-ray. They have, as such, distilled the diagnosis of a fractured hip to a specific symptom when paired with the knowledge that the patient had fallen. However, in complex cases, a flexible conceptual understanding is needed in order to construct ‘mental worlds’ and foresee complex interactions of patient conditions and related symptoms (Hatano, 1982). Here, conceptual knowledge is not enough. Adaptive expertise, according to Hatano, requires the continuous adaptation of new information to existing conceptual models. As such, while necessary, it is not solely the amount of conceptual knowledge that predicts adaptive expert behaviour but the degree of curiosity and flexibility the physician demonstrates as she critically appraises the applicability of her knowledge to a specific situation. Successfully diagnosing a fractured hip can be done without an expert understanding of orthopaedics. The success relies on the ability to effectively adapt to the unforeseen circumstances of the presentation of symptoms. As when the leg does not bend inward, as it typically would, but the resident, contrary to the guideline, orders an x-ray because they learn that the patient has fallen and knows that this could result in an unsuspected fracture.

In more recent studies (Mylopoulos et al., 2017; Van Der Schaaf et al., 2021), adaptive expertise has been described as a cognitive skill set that encompasses three approaches to knowing. First, epistemic distance is the person’s ability to distance his or her beliefs and knowledge from the problem at hand. Thus, adaptive experts are able to posit a separation between what constitutes their past conceptualisation of a problem and the emerging representation of this problem (Mylopoulos & Woods, 2009, p. 410). Second, self-regulation refers to the ability to continually observe one’s level of understanding of the problem space in order to regulate behaviour towards actions to improve understanding. Third, the ability to orient oneself to novel content and leverage opportunities for learning is a way for adaptive experts to improve their level of competence by improving existing skills and acquire new ones. As such, the key behaviour of adaptive experts is their ability to be aware of their knowledge and to accommodate to unknown situations by actively seeking out appropriate learning opportunities (Mylopoulos & Woods, 2009).

Learning to perform as a physician

The construction of knowledge is dynamic and evolving rather than static and accumulative (Mylopoulos & Regehr, 2007, p. 1163), and it is an integral part of clinical decision-making. Therefore, it is important to understand how the social context mediates and moderate residents’ clinical judgement (Redelmeier et al., 2001). Newly graduated physicians face many challenges when starting residency as they begin taking primary responsibility for their patients. The experience of uncertainty is therefore a natural part of residency, but it is also a fundamental part of clinical practice throughout a medical doctor’s career. Uncertainty can arise when one’s knowledge of the observed, observable and unobservable world is insufficient (Djulbegovic et al., 2011) and has been linked to adaptive expertise (Cupido et al., 2022). In medical literature, uncertainty can refer to several things. Uncertainty can arise due to ambiguity in decision-making processes (Alam et al., 2017; Cristancho et al., 2013). It can also be related to self-doubt, arising due to insufficient or conflicting knowledge, or due to a lack of confidence in one’s ability to solve the problem at hand (Spafford et al., 2007). Djulbegovic et al., (2011) describes it as epistemic uncertainty referring to Fox’ definition, meaning that uncertainty arises from “…incomplete or imperfect mastery of the existing knowledge… a consequence of limitation of the current knowledge… [or] a combination of the first two.” (Djulbegovic et al., 2011, p. 321). In contrast, referencing Lonergan’s work, Engebretsen et al. (2016) argues that decision-making requires assessing information to gain new objective knowledge, and that this arises through novel events, which are inherently surprising and uncertain. They argue that uncertainty is thus important in decision-making, so long as it is accompanied by self-awareness:

Objectivity is self-appropriation. It is the result of a gaze turned inwards not outwards. Thus it does not exclude uncertainty but encompasses it. (Engebretsen et al., 2016, p. 598)

Turning one’s gaze inwards, requires one to doubt oneself. Thus, uncertainty as described as ambiguity cannot be isolated from self-doubt. This is further supported by studies, reporting correlations between low uncertainty (ambiguity) in diagnostic tasks and low anxiety, indicating an important psychological aspect of uncertainty, reaching beyond epistemic uncertainty. This is also underlined by Fox’ work on how students evolve in their ability to manage uncertainty, describing an initial anxious focus on individual limitations (Fox (1957) in Lingard et al., 2003).

Uncertainty has been linked to the development of professional identity. In their study on healthcare professionals’ communication during patient treatment and management, Lingard et al., (2003) depict how presenting and discussing uncertainties was an important socialising feature during case presentations for the development of professional identity. Other studies have further shown how such verbalisations help manage uncertainty, impacting residents’ chances of developing adaptive expertise (Apramian et al., 2015; Cristancho et al., 2013).

Despite the importance of managing clinical uncertainty, the continued ambition of standardising clinical work has led clinicians to negate the inevitability of uncertainty (Engebretsen et al., 2016). Uncertainty has come to be interpreted as an indication of incompetence (Ott et al., 2018), which has harmful implications for clinical training (Lingard et al., 2003). This is so, even if uncertainty is described as an integral part of clinical work, mediating the development of expertise (Cristancho et al., 2013). Thus, aligning professional identity and tolerance of uncertainty is important for adaptive expert development (Mylopoulos & Regehr, 2007). Mylopoulos and colleagues argue that, “[a]daptive experts view themselves as ‘accomplished novices’ rather than ‘answer-filled experts’…” (Mylopoulos & Woods, 2009, p. 410). This definition emphasises both the expert’s tolerance of uncertainty as well as a willingness to accept and expect knowledge gaps. The adaptive expert sees this as an opportunity to seek out learning opportunities to close these gaps. Thus, the development of adaptive expertise requires what has been termed ‘clinical courage’ (Fowler et al., 2021). However, being courageous is dependent on social structures allowing space for uncertainty and, therefore, the development of adaptive expertise is embedded in social structures. Although some scholarly attention has been devoted to the study of the development of adaptive expertise in educational settings, much less attention has been paid to how adaptive practices are influenced by clinical environments. (Kua et al., 2021). Thus, a need exists for investigating what mediates and moderate residents’ opportunity to perform adaptive practices.

Methodology

Design

This was a cognitive ethnographic study (Ball & Ormerod, 2000) of two EDs: one at a university hospital and one in a regional hospital. We followed Post Graduate Year 1 (PGY- 1) ED residents when they encountered geriatric patients, from the time of notification of a patient until a diagnosis was reached, and a treatment plan had been drawn up.

Cognitive ethnography

The field of distributed cognition argues that cognition is inevitably imbedded in a social, cultural and physical world (Hutchins, 1995). With its origin in cultural-historical activity theory (Lave, 1988; Leont’ev, 1978), distributed cognition argues for the interchangeable, dynamic interaction between the individual’s cognition and the context (Hutchins, 2010), which is an uniquely human competency (Tomasello et al., 2005). By taking its point of departure in an ecological psychological approach, distributed cognition is a way of investigating how cognition is embedded in social practices and how affordances of the physical surroundings, such as social and technological means, shape cognition (Gibson, 1986; Hutchins, 1995). The method of cognitive ethnography aims to analyse distributed cognition and is a subfield of ethnography. The aim of this methodology is to use traditional ethnographic methods to explore participants’ interaction with their physical surroundings during cognitive tasks. Here, the aim is to both describe the meaning of social, cultural, and physical resources and to analyse processes of how these resources are employed during cognitive activities. As such, cognitive ethnographers organise observations based on an operationalisation of cognitive phenomena into specific episodes of activities. From these observations of episodes, the analysis aims to identify and explore which material and conceptual resources are present and how they are used in the construction of this specific cognitive activity (Williams, 2006).

Cognitive ethnography can be used to explore adaptive practices within broader discourses embedded in the clinical setting (Ball & Ormerod, 2000). The cognitive ethnographic approach is often utilised in clinical settings due to its more focused scope and shorter duration. This requires the researcher to specify observable behaviours that demonstrate the cognitive phenomenon studied.

As described above, the cognitive phenomena of interest in this study were adaptive practices. We therefore operationalised behaviours in the following subcategories: epistemic distance, self-regulation and orientation towards novel content. These operationalisations are described in Table 1. We analysed occurrences of activities, providing a narrative for how development of adaptive expertise was hindered or facilitated in the culture around residents in Danish EDs. To do this, we integrated observations with information from descriptions of cognitive artifacts in terms of objects within the physical milieu, such as patient journals, equipment, or technological aids. This allowed us to analyse how the culture of the ED facilitated clinical decision-making. Additionally, we continually performed clarifying inquiries, typically lasting five to twenty minutes, with the participants. This served to clarify thought processes and counteract possible misinterpretations of their actions. These inquiries were either performed at the end of the observation or during the observation at time points where it would not disturb the participants’ diagnostic process and interaction with patients, relatives, and colleagues.

Table 1 Operationalisation of the adaptive expert framework into adaptive practices
Full size table

Participants

The primary investigator (MLG) observed 27 PGY-1 resident participants who treated, in all, 32 geriatric patients. Of these residents, nineteen were female, and they had on average 1,5 months of experience. Residents were recruited through a chief physician at the ED based on their availability on the day of observation. Transitioning into clinical practice and taking on the role of authority is challenging. Therefore, we chose to observe PGY-1 residents as we wanted to explore the starting point of adaptive practices to better understand how to structure postgraduate training which caters to the development of adaptive expertise. As described earlier, many medical education programmes promote the development of routine practices, and less is known about adaptive practices. Therefore, it was relevant to explore these practices amongst novices.

On the day of observation, the chief physician or the supervising physician would direct a geriatric patient to the recruited resident. Inclusion criteria for geriatric patients were that they be above the age of 60. During the observations, participants were observed to interact with eight fellow PGY-1 residents, 30 supervising physicians in introductory or specialist training, 33 nurses, 13 other healthcare professionals (emergency medical technicians, social workers, etc.) and 17 relatives or accompanying healthcare professionals. Data were collected from August 2019 until December 2020 with observation periods between 7 AM and 10 PM. The project was approved by directors at both departments. At one department, a chief specialist physician in charge of the PGY-1 residency programme also acted as a key informant.

The Danish context: geriatric patients in emergency medicine

PGY-1 residents are important members of the day-to-day healthcare teams employed in Danish EDs. Alongside residents, the team of EM professionals consists of specialist physicians, medical students, nurses, nursing students, community care workers, medical laboratory technicians, physiotherapists, pharmacists, and secretaries. The team also includes residents and physicians from assisting departments, such as orthopaedic, lung, geriatric, and anaesthesia departments.

Geriatric patients are a growing population who make up a large percentage of patients in Danish EDs, and they are often characterised by the complexity of their health conditions (Franklin et al., 2011). These elderly patients are at higher risk of adverse outcomes (Galvin et al., 2017) than younger adult patients (Boltz et al., 2013; De Decker et al., 2014; Hwang & Morrison, 2007; Schumacher, 2005), often due to age-related frailty (Alexa, 2017; Laging et al., 2014; Melady, 2018). In addition, implicit ageism biases may impact decisions regarding geriatric patients (Cook et al., 2017). For this reason, investigating residents’ clinical decision-making in relation to geriatric patients in the ED was thought to be an ideal means of exploring the performance of adaptive practices. It is well researched that physicians make cognitive errors in EM in general (Croskerry, 2014) and with geriatric patients in particular (Galvin et al., 2017; Schumacher, 2005), making geriatric emergency care an appropriate context for the present study. Furthermore, recent studies has highlighted that intra- and inter-personal aspects of geriatric care is challenging to adaptive experts (Kua et al., 2022). It is thus relevant to investigate contextual circumstances for novices when engaging with this patient group. We operationalised geriatric patients in terms of age and accessibility. This choice was informed by the literature, which often defines geriatric patients as above the age of 65 years. However, in our study, we also included one patient who was 60 years old, as he presented as a geriatric case in terms of multimorbidity and frailty.

Data

The primary investigator (MLG), who is a psychologist with two years of clinical training, had prior experience with various qualitative methods. She was supervised by PM, who had thorough experience with observational studies, both as a researcher and a supervisor.

MLG conducted 80 h of observations, taking the role of passive observer participant (Spradley, 1980), following the residents but not taking part in the diagnostic processes or patient encounters. She wrote field notes consisting of actions related to the concept of adaptive expertise. The field notes were guided by pre-defined observation cues (see Table 1: Operationalisation of the adaptive expert framework), notes from the clarifying inquiries and memos of analytical thoughts. Memos included reflections on the theoretical framework, common features observed among participants, changes to the operationalisation of the adaptive expert framework or similar notes relevant to the analysis. These notes were written out, just after data had been collected, as narrative descriptions of the entire patient encounter. These narrative descriptions consisted of: arrival notification, preparation, history uptake with the patient or relatives, consultation with specialist physician, ordering of tests, analysis, diagnosis, and treatment plan in consultation with a specialist physician, and handover to a nurse. Some days of observations included more than one PGY-1 resident, and these field notes were divided into separate, rich narrative descriptions (henceforth referred to as ‘observations’) of the patient encounter for each participant.

Analysis: operationalisation of the adaptive expert framework

The first analytic step was to operationalise the construct of adaptive expertise in order to identify relevant observable behaviours (Ball & Ormerod, 2000). Based on the definitions of adaptive expert cognition provided by Mylopoulos and Woods (2009, p. 410), the primary investigator operationalised specific behavioural markers at both a contextual and an individual level. This operationalisation was then reviewed by two of the co-authors, PM and MM, where consensus was reached. MM is an experienced researcher within operationalisation of cognitive processes as adaptive expertise. The operationalisations were continually refined together with one co-author (PM) during data collection. The final operationalisations are described in Table 1.

The second step of the analysis was performed continuously during data collection, by MLG, PM and MM, and summative results were discussed by all authors in the author group during data analysis. We analysed each observation for adaptive practices based on the operationalisations (Table 1). Here, we coded each observation within the three cognitive constructs (epistemic distance, self-regulation and orientation to novel content) by sectioning the text and condensing interpretations of meaning within taxonomic tables for each construct (see Appendix 1). These taxonomic tables described three levels of analysis: firstly, who or what the resident interacted with (“actor”). Secondly, what were the material, cognitive or social moderators of their behaviour (“moderator”). Thirdly, what were the outcomes of this moderation (“consequence”).

Results

The taxonomies for each cognitive construct provided insights on how different contextual elements, like how patients and electronic devices, could impact adaptive practices. The tables were structured to illustrate how an actor (e.g., the supervisor) moderated behaviours, and what impact that had on adaptive practices. We termed this impact ‘consequences’. We took the point of departure in the actors, as they were the main moderator of adaptive practices. Examples of these organizations of behaviours are illustrated in Table 2 below. For the full tables of each cognitive construct, please consult Appendix 1.

Table 2 Examples of taxonomy of adaptive expertise
Full size table

The taxonomic tables clearly showed that residents’ adaptive practices were both positively and negatively impacted by patients, relatives, artefacts (electronic patient journals, online medical handbooks, etc.), supervising physicians, peers, nurses, and physiotherapists (Fig. 1).

Fig. 1
figure 1

Adaptive practices distributed in the social and physical space. The figure shows how residents’ adaptive practices were impacted by patients, relatives, artefacts (electronic patient journals, online medical handbooks, etc.), supervising physicians, peers, nurses and physiotherapists

Full size image

From the analysis that led to the taxonomic tables, two key moderating aspects of the social and physical space were identified; clinical uncertainty and decision-making disruptions. These aspects were pervasive across actors and often had a negative impact on adaptive practices.

Clinical uncertainty

Uncertainty was often displayed by residents in the form of ordering more tests and a display of rigidity. Uncertainty was observed during all adaptive practices but was especially prevalent in self-regulation processes, as illustrated by the following excerpt:

Sophie (resident) is making decisions on how to get the best x-ray picture of the suspected fracture: “I’m not sure if I should order both shoulder and upper arm, because I don’t know how specific they are… But I had one done on Saturday…”. She looks up the x-ray from the other patient, and based on that, decides that an x-ray of only the shoulder would be adequate.

The excerpt illustrate how Sophie uses the electronic patient journal to self-regulate, when she encounters uncertainty.

However, uncertainty could also result in negative consequences, such as checking behaviours, as illustrated by the excerpt from Sarah’s case:

Sarah (resident) is treating her second patient of the day, whom has a possible fracture. She expresses needing more information for her to be able to decide and looks for a more experienced physician. When she cannot find one, she orders more X-rays to confirm her hypothesis before deciding on a treatment.

This excerpt illustrates that when a supervising physician is not available, it requires the resident to think on their own, which can infer uncertainty.

In terms of epistemic practices, several actors were in play. Supervising physicians could promote epistemic distancing by helping the resident gain a deeper understanding of the problem space by guiding them how to identify knowledge gaps. Here, residents would gain confidence in their exploration, which improved their sense of agency and adaptive practices. A few factors led to obstruction to the performance of epistemic practices. In some cases, the supervising physician seemed preoccupied with work that did not involve the resident. This led the resident to experience uncertainty because they missed the supervising physician’s support. In cases where the supervising physician continuously questioned the residents’ decisions, residents reacted by reinforcing checking and dependent behaviour. Here, checking behaviour is referring to using the supervisor, medical handbooks, nurses or peers to ‘check’ their thinking, decisions or opinions. In this study, most checking behaviours were asking their supervisor, if their decision or interpretation of symptoms was correct. The use of the electronic patient journal proved to be a key tool in prompting epistemic distance, both when the resident actively explored the problem space in focused read-throughs but also in regular preparatory discourses, helping residents detect knowledge gaps. Lastly, patients and relatives also impacted the residents’ ability to employ epistemic distance, both facilitating and hindering residents’ ability to maintain epistemic distance. Patients and relatives could both prompt identification of knowledge gaps (i.e., asking questions), but could also disrupt the resident’s thought processes.

With respect to the supervising physician, self-regulation behaviours were observed when the supervising physician supported resident agency, challenging them to reflect critically and model adaptive practices. Asking clarifying questions could decrease uncertainty and cultivate the residents’ behaviour of monitoring their own level of knowledge. The residents’ sense of agency seemed to diminish when the supervising physician adopted a more authoritative role, possibly because this hindered self-regulation and fostered dependent and checking behaviours.

Many helpful artefacts in the residents’ environment prompted and guided their adaptive practices; for example, electronic patient journals, online medical handbooks, laminated flowcharts and guidelines, and pocket booklets helped residents monitor their level of knowledge and decide which steps were relevant to take, in order to close knowledge gaps. The use of these tools was mainly prompted by patient complexity, uncertainty or when residents met unknown symptoms. Furthermore, referrals could cue self-regulating behaviour based on the expectation of knowledge gaps.

In general, behaviours demonstrating orientation towards novel content were sparsely observed amongst residents and were often related to uncertainty. These behaviours were primarily observed as a result of conferring with the supervising physician or other healthcare staff. The electronic patient journal acted as a tool to ensure future orientation towards novel content as residents would save patients and follow their outcomes; or, because of inconclusive medical tests, would acquire help from specialists and observe their decision-making in order to learn from their specialised knowledge. Another prompt could be when residents perceived that insufficient help was available from either online sources or supervising physicians, urging residents to actively seek out and identify learning opportunities.

During analysis of participant behaviours, a pattern of how residents oriented themselves towards a professional role was revealed as it impacted their adaptive practices. A lack of display of emotions and self-perception could be an indication of the high degree of perceived professionalism that was enacted in the culture of the ED. When questioned about this issue, residents referred to the importance of acting as a professional. However, supressing emotions had implications for how they tolerated uncertainty, as residents associated displaying emotions with unprofessional behaviour. However, registering and acting on one’s emotional response is an important function for tolerating uncertainty. This is both supported by literature, describing that adaptive expertise requires one to lean into uncertainty (Wineburg, 1998), which requires being comfortable with uncertainty and using negative emotions to guide and regulate one’s behaviours (Eva & Regehr, 2007).

A resident’s identity of being the senior to a medical student or in general being put in a superior, supervising or decision-maker role made residents gain confidence. This role, in turn, came with high expectations for residents to act as an expert, and was a fragile one that could easily be threatened. For instance, confidence and ability to act as an adaptive expert waned when a medical student questioned Sarah’s (a resident) hypothesis. Sarah failed to diagnose a fracture of the hip, and only succeeded in detecting it by prompt of the medical student. Her surprise of being wrong made her uncertain. This, in turn, prompted her to subsequently employ checking behaviour. Thus, her inability to meet the medical student’s expectations that she would perform just as well as her more experienced colleagues and not show uncertainty led to disengagement of adaptive practices. Moreover, Sarah abandoned her role as a supervisor for the medical student, being entirely preoccupied with the process of finding the right diagnosis.

Decision-making disruptions

Disruptions could be described as interactions that made residents stop their decision-making process. They were characterised by displays of overlooking contextual cues, checking their decisions with available colleagues, or enforcing a rigid structure. Disruptions impacted both epistemic and self-regulated practices. They could for example, lead residents to possibly overlook important cues, as demonstrated in the excerpt from Catherine’s case:

Catherine (resident) is gathering anamneses on a patient whom she has difficulty establishing a contact, as the patient seems confused and distant. She starts to question the daughter about her mothers’ medication and chronic obstructive pulmonary disease. The daughter replies and mentions the patients’ memory problems. The patient interrupts: “I feel unkempt… I have for some time.”

The resident tries to calm her, saying that she doesn’t see her as unkempt at all. The patient looks lost, and it seems to me that the patient is trying to communicate a general discomfort or underlying problem, which the resident doesn’t pick up on.

However, disruptions could also have positive impact on adaptive practices. The following excerpt illustrates how disruptions from a lack of alignment between the patient’s symptoms and her working hypothesis, helped Christina to self-regulate:

Cued by contradicting findings in the physical examination of the patient’s pain, Christina (resident) re-examines the patients’ swelling. She wants to clarify if he is experiencing any pain in his face, other than locally from where he has broken his jaw and looks under his tongue for any haemorrhages. Christina comments that “I do not think you have any new fractures; I think it is the swelling”. However, she is not satisfied with the adequacy of her own knowledge and calls a specialist to confer her findings.

As with uncertainty, residents could experience disruptions from several actors and this affected all adaptive practices in both positive and negative ways. In terms of epistemic distance, relatives’ or patients’ behaviours disrupted residents, which could prompt them to further explore the problem space or make them overlook relevant symptomatic cues.

In terms of self-regulation, disruptions prompted residents to monitor their knowledge levels and actively take steps to address knowledge gaps by ordering more tests, doing additional physical examinations or seek out help from other medical staff. Nurses played an important role in residents’ ability to monitor knowledge levels and closing knowledge gaps by engaging in knowledge sharing and consulting situations. However, other healthcare staff could also hinder the residents’ ability to self-regulate. This would typically happen when they disturbed the patient or interrupted the resident. When this happened, the resident enforced a more rigid structure of history taking to keep track of their progression.

The interplay between disruptions and uncertainty

By doing a narrative analysis of interactions between the residents and actors, we frequently observed an interplay between disruptions and uncertainty. The following case vignette demonstrates one example of how disruptions could affect how residents handled information and how disruptions impacted residents’ ability to orient themselves towards novel content and learning opportunities.

Case vignette: Example of relation between disruptions and uncertainty

Participant: PGY-1 resident “Julie” (less than one month of training)

Julie was treating a 91-year-old female patient who had been admitted because of dyspnoea. While Julie was trying to obtain a history from the patient, she was continually disturbed by other healthcare professionals who were there to draw blood samples, trying (and failing) to place a catheter. This caused increasing pain and frustration in the patient. The level of chaos in the history uptake then carried over to the conferral with the supervising physician which became sporadic and unfocused as Julie had a hard time sorting out the relevant information and generating hypotheses. As such, her approach to the conferring and the lack of help in organising and prioritising the information from the supervising physician did not initiate a shared reflective practice but rather became more similar to a handover, resulting in Julie the supervising physician just telling Julie what to do. From then on, Julie articulated and displayed uncertainty by checking all her decisions with the supervising physician.

As exemplified by the above example, disruption of contextual learning could often trigger uncertainty. This was mainly observed when nurses or supervising physicians interrupted residents’ line of thought, or if test results or available information did not fit their working hypothesis. The interaction between observed disruptions from different actors and residents’ behaviours indicating uncertainty is illustrated in Fig. 2.

Fig. 2
figure 2

Interplay between disruptions and uncertainty. The figure illustrates the interaction between observed disruptions from different actors and residents’ behaviours indicating uncertainty

Full size image

As demonstrated in Fig. 2, checking behaviour was often indicative of uncertainty and the ‘abandonment’ of adaptive practices. For example, continuous checking behaviour could be instigated or enforced by supervising physicians when they too strongly curated the patient meeting or completely took over the patient case. This often reduced resident autonomy, making them more dependent on the supervising physician and less inclined to make independent decisions.

Discussion

The observed behavioural markers allowed us to observe the complexity of how residents performed adaptive practices throughout the geriatric patient encounter in EDs. It allowed us to understand the interaction between residents and their environment when residents made clinical decisions. The findings can be summarised in three main issues characterising workplace learning. Firstly, all residents engaged in adaptive practices, but they faltered when they faced uncertainty caused by interruptions. Secondly, how the resident perceived their role as an authority and accepted their own uncertainty, impacted their ability to handle uncertainty and perform adaptive practices. Thirdly, how the environment tolerated and supported the residents through this uncertainty was crucial to counteract inappropriate performance of routine practices.

Adaptive practices amongst residents

In this study, all residents engaged in adaptive practices. In conformity with existing evidence, epistemic distance, self-regulation and orientation towards novel content are not separate, static processes (Mylopoulos & Regehr, 2011). Our findings convey an understanding of adaptive expertise as a dynamic and fluid capability. This means that the medical practitioner is not an adaptive expert in every clinical circumstance. Rather, the practitioner engages in adaptive practices by shifting focus when facing uncertainty. This observation further supports the argument that uncertainty does not indicate incompetence (Ott et al., 2018) and echoes findings from a recent meta review, that adaptive practices are engaged when physicians are faced with a change in their environment (Pelgrim et al., 2022). In this study, we added nuance to this picture, in that such changes could result in unproductive uncertainty. It is important to note that adaptive expertise relies on the accumulation of experience and knowledge, but that it is possible to engage in adaptive practices when being a novice. Physicians actively engaging in adaptive practices have the potential to develop adaptive expertise, over time. At the same time, routine practices serve a vital function in daily clinical work and are therefore a complementary dimension of adaptive expertise (Schwartz et al., 2005). A recent study explored routine practices amongst surgical residents, finding that standardization of practice was not feasible, but suggested that stabilization was a more feasible educational focus (Ott et al., 2021). This study builds this argument even further, suggesting that stabilization requires both routine and adaptive practices, as residents will then also be able to manage the inevitable novel problems they will encounter in their work, leading to overall stable performance.

Professional role and uncertainty

The high demand for professionalism requires residents to assume the role of authority from their first day of residency. In this study, this perceived expectation was shown to lower their tolerance of uncertainty as uncertainty often did not align with how authoritative behaviour was modelled by more experienced physicians, or how they thought an authority was expected to act. If residents were able to counter uncertainty with curiosity, they were often able to act adaptively. This was also a good strategy when the context disturbed and disrupted their decision-making process It speaks to the importance of reconciling the authoritative expert role (Lingard et al., 2003) with the inevitable uncertainty of clinical practice (Engebretsen et al., 2016) when learning to become an adaptive expert (Apramian et al., 2015).

How contextual support can mediate residents’ tolerance of uncertainty

In this study, residents who were more confident in their skills, while tolerant of uncertainty, did not rely on interactions with other healthcare professionals to guide and cue adaptive practices. On the other hand, residents who were less tolerant of uncertainty and less confident in their role as an ‘expert’ seemed to need external validation to find certainty, which then either came from checking behaviour (seeking validation amongst experienced staff), security measures (mindlessly asking for tests and information) or anchoring to their own existing beliefs. It is therefore imperative for residency training that the context supports the alignment between uncertainty and professional identity.

This study demonstrated the importance of supervisors performing uncertainty in the clinical settings, supporting earlier studies that also stress the importance of fostering a culture of acceptance of uncertainty to improve patient safety (Cristancho et al., 2013; Lingard et al., 2003; Ott et al., 2018). Residents were more adaptive when they were allowed to be uncertain, and when this uncertainty did not impact how the context viewed their ability to be an authority. As such, this study shines light on how supervising practices and resident’s understanding of what professionals ‘do’ had an impact on their ability to develop adaptive expertise (Rowland, 2017; Rowland et al., 2020). The study describes the negative consequences for learning when this expectation is present. The consequences were that residents were overpowered by uncertainty or forced to be overly sure of their diagnosis, which bore the risk of them making errors because they were not supported in acting as adaptive experts. This problem is especially unfortunate when uncertainty tolerance is linked with several desired healthcare-related outcomes amongst trainees (Strout et al., 2018), and uncertainty-related anxiety has a negative impact on the clinician’s own satisfaction with their decision-making competency (Libert et al., 2016). Here, studies argue that introducing a vocabulary for uncertainty in professional settings can mediate learning at all levels of experience (Lingard et al., 2003; Rowland, 2017). Uncertainty plays a role in developing adaptive practices and could be mediated through targeted verbalisation. LaDonna et al. (2017) described how learners are pushed to stage their performance in accordance with ‘textbook’ approaches when being watched. This hinders their ability to learn to act as adaptive experts when they are assessed in real clinical settings. Through an analysis of uncertainty rhetoric, Spafford and colleagues showed differences in novices’ and experts’ uncertainty, and how expert uncertainty made room for adaptations and innovations in their clinical work (Spafford et al., 2006, 2007). The present study underlines the importance of identifying and verbalising uncertainty in order for clinicians to act as adaptive experts.

The question of uncertainty is especially relevant when discussing how we train residents and, more importantly, how we assess them. A recent article by ten Cate et al. (2021) discusses the educational benefits of using entrustable professional activities as a tool of assessment, arguing that through this practice of workplace-based assessment, education and training will naturally act as a framework for ‘entrustment decision making’ (ten Cate et al., 2021). This speaks to the problem of acceptance of the inevitable risk involved in workplace-based training; namely that learning to adopt adaptive practices presupposes that the supervisor extends agency and trust towards the resident. However, this framework of entrustment decision making does not consider the importance of incorporating a curriculum for uncertainty.

Limitations

While cognitive ethnography was a deliberate choice to critically and thoroughly investigate how cultures of residency training impacted learning adaptive expertise, it is important to highlight the specificity of our data. While the Danish clinical context bears some comparability to other international clinical contexts, some elements were also specific. For example, in these two particular EDs, all residents were required to confer all patients with a supervisor before moving on with the treatment plan. As such, supervision played a large role in residency training at these departments, and supervising physicians took pride in having this workflow implemented as a way of ‘taking care’ of their residents. However, such an arrangement is particular to these departments and is not implemented in all EDs in Denmark.

The choice of PGY-1 residents and the case of geriatric patients also carries some limitations. Several potential challenges come with treating elderly patients, such as communication and the complexity of their condition; and were chosen for these very reasons. In some instances, the patient was not able to cooperate. This made observations difficult like in the case of a patient with dementia who had expressive aphasia. This choice also excluded some good opportunities for observation of challenging situations of patients younger than 60 years old. However, complexity was argued to be important, and geriatric patients provided such complexity. Furthermore, PGY-1 residents never bore the full responsibility, and this could affect the observations and might explain why some residents were quick to seek help.

Lastly, some risks and limitations in regard to the cognitive ethnographic method were identified. Observer effects in ethnographic studies have been widely discussed, and it has recently been argued that they are not as prominent an issue as previously argued (Bressers et al., 2019; Monahan & Fisher, 2010; Varpio et al., 2017). However, given the specificity of the observations, interfering in the ecological setting of the participants’ distributed cognition may have an effect on the results. While MLG did not disclose to the participant the specific observed cognitive behaviours, she was obliged to informed participants of the aim of the observations. By disclosing this, she may have affected the ecology of their cognitive behaviour and have increased their reflexivity, which would be coded as adaptive practices. During the analysis, however, increased reflexivity or adaptive practices were observed neither in the beginning of the observations nor during events where the participant was especially engaged with MLG, which may indicate that this suggested effect is at a minimum. Furthermore, reliability was enhanced by including more than one investigator in the analytic process, which ensured that such effects were identified.

Conclusion

One of the questions that research on adaptive expertise has grappled with is the role of uncertainty in the development of adaptive expertise. This study found that residents learned to engage in both routine and adaptive practices when faced with uncertainty. Adaptive practices were observed amongst all residents to various degrees and were seen as fluid rather than static; i.e., they occurred under some conditions and were absent in other situations. Uncertainty was induced by factors in the environment, especially patient characteristics and medical case complexity, unforeseen incidents and disruptions. Residents’ confidence in their own ability and how well they were supported in aligning their professional identity with the naturally occurring uncertainty of clinical work impacted how well they retained adaptive practices.