Article Text

Download PDFPDF

Use of head camera-cued recall and debrief to externalise expertise: a systematic review of literature from multiple fields of practice
  1. Vivienne Isabella Blackhall1,2,
  2. Kenneth Grant Walker1,2,
  3. Iya Whiteley3,
  4. Philip Wilson4
    1. 1 Centre for Health Education Research and Innovation, College of Life Sciences and Medicine, Aberdeen University, Aberdeen, UK
    2. 2 NHS Highland Department of Medical Education, Centre for Health Science, Inverness, UK
    3. 3 Centre for Space Medicine, Mullard Space Science Laboratory, University College London, Holmbury St Mary, Surrey, UK
    4. 4 University of Aberdeen Centre for Rural Health, Centre for Health Science, Inverness, UK
    1. Correspondence to Miss. Vivienne Isabella Blackhall, Centre for Health Education Research and Innovation, College of Life Sciences and Medicine, Aberdeen University, Aberdeen AB24 3FX, UK; vblackhall{at}googlemail.com

    Footnotes

    • Contributors PW is the guarantor. VIB drafted the manuscript. All authors contributed to the genesis of the selection criteria. VIB and PW developed the search strategy. VIB, PW and KGW screened the papers for inclusion. All authors read, gave feedback and approved the final version of the manuscript.

    • Funding No funding was received for this work.

    • Competing interests IW authored two of the papers under review, under the names I Solodilova and I Solodilova- Whiteley. The remaining authors have no competing interests.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Collaborators Professor Jennifer Cleland (Centre for Health Education Research and Innovation, College of Life Sciences and Medicine, University of Aberdeen).

    View Full Text

    Statistics from Altmetric.com

    Introduction

    Naturalistic decision making is that which occurs out with the laboratory setting in complex (time-critical, high-stakes, dynamic) situations.1 The study of such decision-making processes poses several problems for the researcher.2 While concurrent reporting can interrupt the flow of a procedure and alter the cognitive processes being studied, retrospective reporting, if not cued by real events, can be associated with bias and omission of key information. In response to these limitations, a technique known as video-stimulated cued-recall-debrief was developed by Omodei et al.2

    The technique is designed to explicate the thoughts and actions of an individual undertaking a complex task. The individual wears a head-mounted camera which records the task from their own point of view. Footage captured is reviewed after the event along with a facilitated debrief to help externalise cognitive processes. The theory posits that footage from an ‘own point-of-view perspective’ contains motion, auditory and visual cues, which together stimulates memory processes, allowing the individual to undergo a high level of experiential immersion. This facilitates recall of specific cognitive processes as well as ‘pre-verbal’ (or non-verbal) processes such as affect and motivation. Experiential immersion is important because recollection retrieval is superior in situations which more closely resemble the original event’s context.2

    The method has been described in several domains, including firefighting, orienteering, clinical medicine and the aviation industry. The aim of this systematic review is to examine the current evidence for video-stimulated cued-recall-debrief as a means of explicating expert thoughts and feelings in complex tasks. The scope was intentionally not limited to a medical environment as the technique was originally described in other domains. Understanding its application in these contexts will allow consideration of the potential range of uses of the technique.

    The primary outcomes of interest are whether video-stimulated cued-recall-debrief is:

    • Feasible (in terms of general practicability and extent of interference by the camera).

    • Associated with experiential immersion (and if so to what degree).

    • Able to generate unique insight into the performer’s decision-making processes.

    In addition, an evaluation of the current applications of the technique is offered.

    Methods

    Data collection

    Our systematic review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) on 6 March 2017 (trial registration number CRD42017057484).

    A systematic literature search was conducted in May 2017 with the assistance of one librarian at our institution. Key terms were searched for in MEDLINE, EMBASE, Education Resources Information Center, SPORTDiscus, PsycINFO and Google Scholar from first records to May 2017. The literature search was limited to the English language and human subjects. A later repeat search without language limits revealed no additional, non-English papers. Following database interrogation and identification of the relevant studies, the reviewers checked for additional relevant cited and citing articles by scanning paper titles. The search terms were ‘cued-recall (debrief)’, ‘decision making’, ‘skills’, ‘video recording’ or variations thereof. A detailed search strategy is provided in table 1.

    Table 1

    MEDLINE search strategy

    Literature search results were uploaded to the reference manager Mendeley. After removing duplicate studies, the team developed and tested screening questions based on the inclusion criteria. VIB independently screened the titles and abstracts of manuscripts against the inclusion criteria. When there was dubiety as to the relevance of the article, the full text was also reviewed. The full text for all titles that appeared to meet the inclusion criteria was obtained (n=20). A flow chart displaying the selection process is provided in figure 1. The full texts were formally assessed for eligibility in relation to the inclusion criteria. This was undertaken by three researchers. VIB reviewed all texts. The manuscripts were arbitrarily divided into two and reviewed by PW and KGW, respectively. One hundred per cent agreement was established between the reviewers. The researchers were not blinded to the journal titles, study authors nor the institutions.

    Figure 1

    PRISMA diagram (Preferred Reporting Items for Systematic Reviews and Meta-Analyses).

    All types of study design were included. This decision was made following an initial scoping search, where it became clear that the literature base was small. To our knowledge (and following interrogation of trial registries), there are no randomised controlled trials examining the utility of the ‘cued-recall-debrief’ technique.

    Papers were included if the participants were considered to be experts (by the authors’ own definition), undertaking any complex task (time-critical, high-stakes, dynamic conditions).1 No limit was placed on the type of task and could, for example, include playing a tennis match. The search included all studies describing video-stimulated cued-recall-debrief, or indeed any other similar feedback process of another name which is designed to elicit detailed recall in order to explicate the expert’s thoughts and feelings after the task. Comparators were all or none.

    Studies were included if they examined one or more of the following features of the technique:

    • Feasibility.

    • Extent of experiential immersion.

    • Ability to generate unique insight into the performer’s decision-making processes.

    • Current applications.

    Quality assessment

    All studies had a qualitative component. To assess the quality of the papers, Critical Appraisal Skills Programme (CASP) checklists were used.3 There are no other published reporting guidelines which include a tool for assessing qualitative research in medical education. Manuscripts were not graded in terms of their quality, as the CASP checklist guidance notes state the checklists are designed to be used as educational tools, not scoring systems. The results were tabulated and areas of concern highlighted using a colour coding system (table 2). For mixed-methods studies, quality assessment of quantitative components was included in a separate column.

    Table 2

    Quality analysis

    Results

    Nine studies were included in the synthesis, the features of which are outlined in table 3. There was significant heterogeneity in terms of the participants, which included clinicians (occupational therapists4–6 and emergency physicians7) and others (orienteers,8 computer war games players,9 fire station officers10 11 and military pilots12).

    Table 3

    Features and methodology of included studies

    There were major differences in how an ‘expert’ was defined. One study stipulated satisfaction of a 6-point checklist in order to qualify, which included having been nominated by a hierarchical superior and holding the highest clinical hospital grade.7 The checklist did not include years of experience, which is known to be a poor surrogate marker for expert performance.13 Despite this, many papers used this to define their expert population.4–6 12 One study included no definition of the expert.9

    There were small variations in the way in which the intervention was described. True to the original description,2 all participants wore a head-mounted camera while undertaking a complex task and subsequently undertook a facilitated video-stimulated ‘cued-recall-debrief’ process to generate a commentary of the episode. The majority of studies used the authentic ‘non-directive’ approach to the debrief; however, one used a novel, semistructured format.7 In five of the nine studies, the individual facilitating the debrief was poorly defined, referred to as ‘an experimenter or member of the team’, and there was no description in any of the studies relating to their level of experience or training.

    All studies were observational and had a qualitative component. Four were purely qualitative,4 5 7 12 and the remainder used a mixed-methods approach.6 8–11 In the majority of the studies, the qualitative component involved transcription and coding of the cued-recall-debrief commentary.5–7 9–12 Two coding approaches were described: a priori or emergent. Generally, where a relevant cognitive model existed in the literature, coding proceeded in an a priori manner.5 6 9–11 Sometimes, this generated new emergent codes which did not fit with the existing framework.5 Other studies adopted a wholly emergent code approach to coding.7 12 Irrespective of how they were generated, emergent codes were used to perform a thematic (or similar) analysis and gain new understanding of cognitive processes or generate novel theoretical models.5 7 12

    Three methods of quantitative analysis were described in the literature:

    • Counting the frequency of codes encountered.6 9–11

    • Using a computer program to generate quantity of codes as a function of time.9

    • A questionnaire measuring participant perceptions of the utility of the intervention in terms of immersion and ability to trigger recollections and insights.8

    Quantitative methods were employed when some form of comparison was taking place:

    • Comparing cognitive activity of different participant groups6 (eg, novice and expert).

    • Comparing level of immersion and recollections generated from cued-recall-debrief with other techniques8 9 (eg, free recall).

    • Comparing the content of recollections from cued-recall-debrief with other established cognitive models.5 10

    • Stratifying the level of performance through the use of rating scales.11

    The studies could be categorised into three major groups depending on their aims:

    • Basic description of the technique with outcome data.4

    • Comparing the effectiveness of cued-recall-debrief with other techniques in studying naturalistic decision making.8 9

    • Applications of the information generated from cued-recall-debrief:

      • Generation of novel insight or theory into cognitive processes.5 7 10 12

      • Comparing the cognitive processes of novices and experts or good and poor performers.6 11

    Each of these categories will be considered in turn.

    Basic descriptive study

    A case report by Unsworth demonstrated that occupational therapists and their clients felt that the use of head-mounted camera during their interaction was acceptable.4 The equipment was portable, did not affect therapist mobility nor interfere with the therapist–client interaction: participants ‘forgot’ about the camera’s presence and were able to ‘continue as usual’.

    Unsworth was able to generate accurate ‘own point-of-view’ footage, which displayed the rich visual cues of the client’s facial expressions (known to be important in generating immersion). No comment was made as to the extent of experiential immersion experienced by the therapist nor the extent of the recollections generated.4

    The main criticism of Unsworth’s paper is that the outcomes were poorly described and interlaced with anecdotal findings from other papers, making identification of novel data challenging.

    Comparison with other techniques

    Two studies compared the effectiveness of cued-recall-debrief with other methods of studying naturalistic decision making (free recall and a ‘walkthrough’ technique).8 9 Effectiveness was measured in terms of the extent of experiential immersion and verbalised recollections.

    Omodei and McLennan compared cued-recall-debrief with free recall, in a group of orienteers.8

    Cued-recall-debrief was associated with increased experiential immersion (4.4 vs 6.33, p<0.01) and allowed the individual to recount between two and four times the amount of detail compared with free recall. Cued-recall-debrief generated more recollections related to thoughts (5.3 vs 6.87, p<0.01) and feelings (4.42 vs 6.68, p<0.01). All participants stated that cued-recall-debrief was the superior method in terms of recalling mental processes.

    Elliott et al compared the effectiveness of cued-recall-debrief with a cognitive walkthrough model in participants undertaking a simulated air defence task.9 The walkthrough model (adversarial crew cognitive walkthrough or ACCW) involved pausing game play when a significant episode occurred in order for researchers to pose questions about the expert’s thoughts and feelings.

    The paper concluded that ACCW was more time-efficient. ACCW took approximately half the time of cued-recall-debrief in generating the same quantity of recollections (1 hour 42 min vs 3 hours 28 min).

    Compared with ACCW, cued-recall-debrief generated between two and three times the number of recollections related to the expert’s own thoughts. The techniques were comparable in their ability to generate recollections relating to their adversary. These results were demonstrated graphically with no specific numerical values for each type of recollection. Furthermore, no formal statistical analysis was undertaken to determine whether the difference between the groups was statistically significant. Experts stated that both techniques were useful and provided rich insights into their cognitive processes. Interestingly, the authors concluded that ACCW was the more effective technique, presumably because it was more time-efficient.9

    Applications of cued-recall-debrief

    These studies did not evaluate the feasibility or effectiveness of cued-recall-debrief as a means of studying naturalistic decision making; these qualities were assumed. These studies had three distinct aims:

    • Generation of additional insight or novel theory into the cognitive processes of the expert.5 7 10 12

    • Comparing the cognitive processes of novices and experts.6

    • Comparing cognitive processes in good versus poor performance.11

    Generation of new theory

    Unsworth used two established clinical reasoning models14 15 as a priori frameworks in order to map the clinical reasoning of occupational therapists.5 She was not only able to demonstrate the most commonly used types of reasoning (procedural, interactive and conditional) and the way in which they were used (in rapid succession and simultaneously), but she also identified an emergent subcategory of reasoning (generalisation reasoning).

    Solodilova-Whiteley and Johnson uncovered the cognitive processes of military pilots.12 Through an inductive, evolutionary data analysis technique, they generated an eight-stage model of how pilots acquire and use information which could be used to help inform the design of more efficient cockpit interfaces.

    The authors found that pilots used previous experience to generate a model outlining the expected course of the flight which was articulated in a preflight brief. Key stages in the flight were bookmarked with expected values, which were compared with actual values. Pilots gathered information from many sources, at two minute intervals, and values were always interpreted relative to one another. Information gathering was supported by common information structures such as checklists or standard operating procedures. On completion of information gathering, pilots anticipated potential problems. A process of problem solving and information organisation followed before pilots reached a ‘point of clarity’ and executed their decision. This entire process was converted to knowledge and experience, banked for future flights.

    Solodilova-Whiteley and Johnson’s paper gives us some general insights into expert decision-making processes. First, experts rely on previous experience to pattern-match situations. Second, they demonstrate good situational awareness: systematically gathering and assimilating data, while anticipating future events. Lastly, they use strategies to reduce the burden on their working memory. Doing things in the same way every time, for example, using checklists, helps to reduce the burden on working memory and ensures data gathering is systematic.12

    Generation of additional insight

    Omodei et al studied the decision-making processes of firefighters using cued-recall-debrief.9 The information gathered by her team was compared with an established decision-making model (recognition-primed decision making16). Some similarities were demonstrated between the data gathered by cued-recall-debrief and the recognition-primed decision model (decision making is primarily intuitive); however, the former revealed novel information: 20% of cognitive activity was devoted to affect and motivation. This led the researchers to postulate that individuals who are more effective in terms of self-monitoring and emotional regulation may be more skilled performers. This hypothesis was confirmed in their follow-on paper,10 which will be considered later.

    Pelaccia et al studied emergency department physicians during a patient encounter.6 The following insights into the decision-making processes of participants were revealed:

    • All generated at least four hypotheses.

    • Most hypotheses were generated fast and without conscious effort.

    • Physicians remained vigilant to the emergence of new hypotheses (ie, avoided premature closure).

    • Hypotheses were rank-ordered and the order was not consistent with standard probability theory.

    Comparison of novice and expert performance

    Unsworth demonstrated important differences in the clinical reasoning of expert and novice occupational therapists.5 While experts demonstrated a greater capacity to reason interactively and conditionally with their clients, novices tended to demonstrate a more procedural, textbook approach. In addition, the study showed that experts were more likely to:

    • Use pattern recognition techniques.

    • ‘Think on their feet’ (improvisation).

    • Use more than one type of reasoning at once (multitasking).

    Comparing good versus poor performance

    Omodei and McLennan’s follow-on study involving firefighters investigated the differences in cognitive activity associated with ‘good’ versus ‘less effective’ decision making.11 All participants were experts and their performance was stratified into poor or good by two senior fire officers using a 10-point scale. The performance standard was peer-referenced: the six best scores were deemed to represent good performance and the six lowest to equate with poor performance.

    There was a significant difference in the way in which cognitive activity was used. Poorer performers tended to report a greater frequency of thoughts related to self-monitoring and emotional regulation (X2 (5; n=485)=27.4, p<0.001), the content of which was uniformly negative. Good performers thought less frequently about self-monitoring and regulation, but when they did they exhibited no such negative evaluations. The authors surmised then that effective decision making is associated with fewer negative self-evaluations, indicating more effective emotional regulation.11

    The qualitative analysis revealed that the poorer decision makers had a tendency to become overwhelmed by the scenario (cognitive overload). Consequently, the capacity of their working memory was reduced. They demonstrated poorer situational awareness and a tendency to commit decision-making pitfalls (eg, anchoring). They tended to focus on only one salient feature of the situation while ignoring emerging threats.11

    More skilled participants demonstrated better situation awareness as they:

    • Kept track of events.

    • Anticipated developments.

    Good performers also reacted to new threats without undue irritation or concern. Interestingly, there was no correlation between level of performance and number of years of experience.11

    Discussion

    In summary, the literature suggests that video-stimulated cued-recall-debrief is a useful and feasible means of gaining insight into expert performance. However, these results should be interpreted with caution due to the small body of literature describing the intervention and the methodological flaws identified in many of the studies. The head-mounted camera is portable and does not appear to interfere with the episode being studied (although this latter observation is purely anecdotal).4 Cued-recall-debrief is associated with a high level of experiential immersion and allows an individual to recount between two and four times the number of recollections when compared with free recall.8 The method is particularly useful in terms of gaining insight into one’s own actions and feelings.8 9 It is however more time-intensive in terms of time than other information elicitation techniques such as cognitive walkthrough.9 A cognitive walkthrough is not always possible in naturalistic decision-making settings. For example, it would be dangerous to ask a surgeon to pause and discuss their thoughts and feelings when performing an emergency tracheostomy.

    The majority of the papers included in this review considered the technique’s applications:

    • Generating additional insight or novel theory relating to expert cognition.

    • Comparing the cognitive processes of two groups of individuals undertaking the same task.

    These studies demonstrate commonalities in the cognitive and affective processes of experts across several disciplines. Generally, experts tended to use intuitive, as opposed to analytical, reasoning. These types of reasoning can be conceptualised using the dual process theory, which posits that two systems of decision making exist.17 System 1 is intuitive, automatic and fast and largely based on pattern matching. It relies on having built up a bank of previous experience. System 2 is slower and analytical and involves a degree of problem solving.

    Other features demonstrated by experts were as follows:

    • An ability to multitask.

    • An ability to improvise.

    • Awareness of and ability to use strategies to avoid common pitfalls: checklists, cognitive strategies (avoiding premature closure).

    • Good situational awareness.

    • Effective self-monitoring and emotional regulation.

    The final point relating to emotional regulation is in keeping with what we know from some of the existing decision-making literature. Too much concern about how well one is doing in a task sometimes disrupts performance by loading short-term memory with pointless thoughts. As controlling thoughts and behaviours is one of the tasks performed by system 2, perhaps experts have a more efficient or better developed sense of analytical reasoning.18 Understanding the way in which expert medical practitioners think may help to support other medical practitioners model their cognitive processes in order to develop their own expertise.

    A common limitation of the studies was a lack of transparency as to the relationship of the lead researcher to the activity being studied. As a researcher, being both an ‘insider’ and ‘outsider’ can have an impact on the way in which data are collected and analysed, and therefore it is important to reflect and comment on this influence.19

    One paper applied no formal statistical analysis when comparing numerical values between two groups.9 This makes it difficult for us to surmise whether the differences between the groups were statistically significant.

    For the majority of the studies, the debrief facilitator and their level of experience was poorly defined. Facilitating a debrief is a difficult skill that requires practice. As a beginner, there is a natural tendency to ask ‘why’, leading the expert to evaluate their performance, rather than to facilitate the recall of their cognitive and affective processes. To maximise the integrity and reproducibility of the technique, we should ensure that reporting is explicit in terms of the debriefer’s training experience and introduce a recommendation for core debriefing competencies to be met ahead of engaging with the technique.

    A limitation of this review is the small number of studies which were included, a reflection of the small body of literature which exists in this domain. Many papers were written by the same authors, and sometimes within the same year. This may introduce an element of confirmation bias whereby researchers unintentionally focus on data which support assumptions about the technique, rather than challenging their theoretical stance. The typically small sample sizes encountered do not necessarily raise concerns, qualitative focuses on indepth and purposeful sampling, even single cases.20

    Conclusions

    To the authors’ knowledge, this is the first systematic review considering the current evidence base for video-stimulated cued-recall-debrief. The authors tentatively conclude that the technique appears to be a feasible and valid method of gaining insight into an expert’s thoughts and feelings. Additionally, it can be a useful tool in building models and patterns of cognitive activity and characterising the way in which more and less skilled individuals tend to think and feel.

    Future research might include expanding the use of the technique in medicine, with attention to technical and non-technical skills. Specifically, this could examine whether there are particular environments (eg, acute medical care environments such as emergency departments, or elective treatment areas), procedures (eg, coronary artery angiography, epidural insertion or colonoscopy) or individuals for which this technique works best. The technique could also be used to determine whether the insights gained into the cognitive processes of experts are transferable to a group of learners and helps to support the development of their own expertise. Indeed, shortening the expert learning curve for medical procedures may help to improve patient safety. Lastly, the impact of the intervention on the activity being studied could be evaluated. Although inanimate technologies and materials tend to be accepted as part of the backdrop in medicine, a head-mounted camera may influence what people think, say and do.

    Acknowledgments

    Professor Jennifer Cleland (Centre for Health Education Research and Innovation, College of Life Sciences and Medicine, University of Aberdeen) advised on the methodology. Database expertise was provided by Rob Polson, senior librarian at the Centre for Health Science, Inverness.

    References

    View Abstract

    Footnotes

    • Contributors PW is the guarantor. VIB drafted the manuscript. All authors contributed to the genesis of the selection criteria. VIB and PW developed the search strategy. VIB, PW and KGW screened the papers for inclusion. All authors read, gave feedback and approved the final version of the manuscript.

    • Funding No funding was received for this work.

    • Competing interests IW authored two of the papers under review, under the names I Solodilova and I Solodilova- Whiteley. The remaining authors have no competing interests.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Collaborators Professor Jennifer Cleland (Centre for Health Education Research and Innovation, College of Life Sciences and Medicine, University of Aberdeen).

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.