Context Institutional interactions are often asymmetrical in that the professional has more control over the conversation. It is difficult to say who the professional is in simulated consultations between simulated patients (SPs) and medical students because these feature a real (educational) institutional context and a simulated (medical) institutional context. This study describes this asymmetry and makes educational recommendations based on the description.
Methods One hundred assessed conversations between SPs and Year 3 students were transcribed and analysed using discourse analysis (DA). We aimed to find linguistic patterns in predefined parts of the conversations (questions, topic initiations, openings, closings) that might suggest conversational dominance.
Results The SP is conversationally more dominant, despite performing the role of the patient, in that he or she asks more direct questions, is more likely to initiate topics, is more likely not to follow topic changes by students, and closes the consultation. The student is likely to follow topics initiated by the SP and to seek permission to pre-close the consultation.
Conclusions The apparently greater dominance of the SP indicates that the simulated consultation differs from the doctor–patient consultation in certain key aspects. It is in that sense unrealistic. We argue, however, that ‘realism’ ought not to be a goal of simulated consultation and that what matters is that such consultations are sufficiently realistic for their educational purpose. We discuss the educational implications that follow from this.
When we use language, we often use it to achieve social purposes, such as to congratulate a newly promoted colleague, to tell our children we love them, to complain about the unfairness of life, to reassure a patient, and so on. Language studies traditionally concentrated on the structural building blocks of language (e.g. its grammar). Discourse analysis (DA) arose essentially as a counterbalance to this, specifically to concentrate on language in the social context. There is a great deal of background reading available on DA.1,2 The University of Ghent’s website offers an easily available overview.3 Roberts and Sarangi4 discuss the issues in the context of their own work. As any background study will attest, a whole range of approaches has arisen and the term ‘discourse analysis’ has been used – not entirely consistently – as an umbrella designation for all of them. It was felt originally that language in use, over texts longer than a few words, was too disorganised for coherent study, but this view has changed. It is now commonplace to accept that languages have ‘rules’ which apply to complete spoken or written texts, just as, if we can adopt Wittgenstein’s metaphor (after Bellack et al.5) for the purposes of the present piece, games have rules. As these authors5 pointed out, it is possible to obey the rules of languages without being conscious of them. (Many readers will be aware that they can say more about the grammar of a language they learned at school than they can about the grammar of their native language, acquired from infancy.)
Thus DA helps us to understand both the language used and the social dynamic between the conversational participants, whose status and role are reflected in language6 or indeed ‘interactively achieved’7 (see also ten Have8). Thus, a secretary will talk differently to the chief executive officer (CEO) of a big company than she might to her sister, reflecting the differences in social roles, but the CEO (and only the CEO, presumably) can offer to engage on first name terms at some point, thus achieving a new social relationship. The type of asymmetry in this case, in which the CEO is recognised as the more powerful, both in the language he uses and in his right to change the rules (‘Call me Peter’) is typical of professional settings, including that of medicine.1,9,10
In most studies on institutional interactions, the person with the more senior role (or the professional role when the other speaker is a lay person) is conversationally more dominant, at least where the traditional ‘professions’ are concerned.10 In doctor–patient consultations, this means the doctor; in student–teacher conversations, it means the teacher. There are several widely used linguistic markers of conversational dominance, which we will discuss more thoroughly in the Methods section. There is evidence that a conversationally more dominant person is more likely to open the conversation, initiate new topics, talk more, ask more questions, interrupt his or her conversational partner, and close the conversation. As is always the case in studies of discourse, this kind of statement needs to be offered tentatively: all language games have very flexible rules.
The study described here used a DA approach to examine the language of medical role-play. Simulated patients (SPs), represented by persons who portray specific patient types with (a range of) symptoms in an educational setting within a medical school, are frequently used in the teaching and testing of clinical communication. Although the language, structure and content of the interactions between SPs and medical students have been studied,11–13 work in this area has been limited. This is a particular concern given the extent to which SPs are used in summative testing of both students and qualified doctors.
The simulated consultation between a student and an SP is an institutional interaction, but, as we will argue, it is a complex one. Seale et al.12 see simulations as ‘hybrid forms’ (see Roberts et al.11) in which the educational and medical settings coexist. Both conversational participants have dual roles. The SP is both assessor and patient; the student is both testee and doctor (Hanna and Fins14 provide an excellent analysis). In SP–student consultations, therefore, the SP is both a more powerful assessor and a less powerful patient: the student is both a more powerful doctor and a less powerful subject of assessment. Two games, with two sets of rules, are being played simultaneously. This paper aims to look more closely at simulated interactions in order to find out how the different sets of rules manifest in the conversation.
The total population consisted of 317 Year 3 students at Birmingham Medical School, who were assessed in 2003–2004 on a single consultation that was video-recorded to identify students who might be in need of additional teaching support. Thus the assessment did not represent a pass/fail hurdle. Year 3 students have had a little previous contact with patients in a primary care setting, in which they will have observed consultations and will have interviewed volunteer patients on topics such as smoking habits. They have had a limited amount of role-play training (2–3 hours each, in groups of up to 18 students). One hundred of these assessments were selected retrospectively (in 2007) through a process of stratified sampling (for grade, gender and scenario) as the dataset for this study. All consultations were observed as they happened by an examiner who was present throughout and awarded the mark. The simulated consultations were transcribed and the students and SPs anonymised in the transcripts. Transcription conventions (Table 1) were those adopted for previous studies, using a simplified version of standard techniques.15 This was advantageous because it enabled the transcripts to be analysed using Atlas.ti (Scientific Software Development GmbH, Berlin, Germany). Appropriate ethical approval was granted by the Medical Education Unit at the University of Birmingham, UK.
Table 1. An abridged transcription key
|[FN]||First name medical student|
|[LN]||Last name medical student|
| Underlined ||A lot lot louder, stressed|
|||Seconds of silence|
|-||Sudden interruption of speech|
|Sud[denly]||Non-finished word, within brackets only if it is clear what the word was|
|((xxxxx))||Description of non-verbal behaviour. For example:|
|((laughs))||Laugh accompanied by sound|
| ((shakes SP’s hand))||Handshake|
The SPs used for this assessment were experienced and had attended additional training for this intervention. The assessments took place at different hospitals in and around Birmingham. Scenarios (Table 2) were devised by the Interactive Studies Unit (ISU), University of Birmingham and created by clinical and non-clinical staff members. Further details have been published previously.13
Table 2. Scenarios used in the Year 3 communication skills assessment in 2003–2004
|Scenario 1: Cancelled operation|
|The student is asked to tell a patient (who is an auxiliary nurse) that his or her operation has been cancelled. This has happened to the patient once before|
|Scenario 2: Patient complaint about colleague|
|A patient is in hospital recovering from an appendectomy, having been rushed in the night before. The patient wants to discuss the behaviour of the doctor the night before as the patient feels he or she was mistreated and is considering filing a complaint|
|Scenario 3: HIV fears|
|A patient who is in hospital for specialist tests for stomach pain is worried that the underlying cause of his or her symptoms is HIV infection and wants to discuss this with a student to avoid notes being made on his or her records|
|Scenario 4: An embarrassing lump|
|A patient who is in hospital for the treatment of an infected toenail wants to talk about a lump he or she has found in his or her groin. The patient is anxious about being in hospital because of the death of a parent of cancer|
|Scenario 5: Alcohol abuse|
|A patient who is in hospital after having collapsed outside a nightclub as a result of alcohol consumption wants to talk to a student about studying medicine|
For the purposes of this study, we identified standard linguistic markers of conversational dominance, searched for them in the texts, and analysed the contexts in which they appeared.10,16–21 These were:
- • controlling the flow by: (i): asking more questions, and (ii) initiating topics;
- • managing the fringes of the conversation by: (i) opening the consultation, and (ii) closing the consultation, and
- • taking the floor by: (i) talking more, and (ii) interrupting more (our definition of interruption: overlapping speech resulting in a change of speaker).
The last category has been previously published.13 We found that SPs talked and interrupted more than medical students. The current paper will focus on the first and second categories.
This study shows that SPs are more likely to ask direct questions and students to ask indirect questions. Simulated patients are more likely to initiate topics and to follow topics poorly. There is uncertainty at the points at which the game of simulation starts and finishes. This game is formally opened by the examiner, but the opening, devoid of social preliminaries, is unusual. The SP is likely to close the consultation, sometimes through the addition of what we describe here as a trigger to the pre-closing sequence that immediately follows it.
This suggests two things. Firstly, that the simulation game is characterised by dominance by the ostensibly lay participant, the SP. The SP – and beyond the SP, the examiner, whose representative the SP is – is of course the dominant person in the game of education; thus, another way to consider this is to conclude that the rules of the game of education trump those of the game of medical consultation. Secondly, these findings suggest that this interaction is subject to difficulties related to authenticity, partly consequent on this departure from what we know about dominance in professional–lay encounters, and partly because the fringes of the consultation are clumsy: it is, in other words, hard to switch from one game to the other naturally.
There are a number of caveats here, however. One is that the student within the game of simulation is in fact ‘role-playing’ a student. He or she is bringing a real persona to a simulated situation, and in that sense is not role-playing in the same sense that the SP is. In addition, a student is not a doctor and is therefore only partly a professional. Another is that the very general characteristics of conversational dominance that are reflected in the literature are always constrained by particular circumstances. In our data, for example, as we observe above, Scenario 3 on HIV is such that the language characteristics are somewhat different. Interactions do not exist merely to establish or confirm social roles: professional interactions in particular are driven also by the need to give and receive information, tell the truth, manage uncertainty and the like. As we have noted, the development of DA was motivated by a desire to understand language in its social context, and the habit of DA is to concentrate on this. However, it is not the only dimension.
Nonetheless, there are more important points to be made. The SP in particular does a number of different things at the same time. The interaction must be guided by the SP in order to allow the student the opportunity to demonstrate competence. The SP can do this by asking questions, for example, partly because such questions are inherently a means of dominating and therefore steering the interaction, partly because the examiner needs to know how the student will respond, and partly because the simulation is designed precisely to be non-routine in order to gently test the limits of the student’s ability to handle complex situations. The same is true of the issue of topic nomination.
Therefore, in fact the relative dominance of the SP is a pre-condition of the test: if the student were dominant, the consistency among students would be lost. Partly, however, it is there precisely because it ought not to be, to offer the student the opportunity to demonstrate how to cope with problems. The structural necessity for the SP to shape the discussion is, to use the phraseology of the arts, given a rationale based in plot and character: for example, this is how angry people behave.
Being an SP in this sense requires complex skills. It relies on the ability to handle the interaction on both dimensions simultaneously, both as a plausible simulation and as an educational exercise. The SP must both ‘be the patient’ and monitor the student in order to shape the interaction. There is an absolute requirement for creative ambiguity.
The formulaic openings and sometimes clumsy feel to closings, together with the systematic differences (the lack of small talk and the trigger to the pre-closing sequence), demonstrate the limits to authenticity. The participants in the simulation game may need to, but cannot, refer to the world outside. They cannot be a general practitioner and patient who might begin by saying ‘How’s your mum keeping?’ or ‘Those road works outside are a nuisance.’
It seems clear, therefore, that when we say that simulations are ‘unreal’, we must say not merely that they are illusions, but that they are counter to what we know of real interactions. They are interactional games and are played by other rules. We would argue that this does not matter, but that we do need to be aware of it.
The language of the theatre, for example, is by many measures utterly unlike real language, most notably because a dramatic script does not contain the volume of hesitations and repetitions found in any transcript of an interaction. No-one, as Dr Johnson said, is confused by the theatre; we know it is a show: ‘Imitations produce pain or pleasure, not because they are mistaken for realities, but because they bring realities to mind.’30 The aim of simulation, then, is to provide sufficient authenticity to bring realities to mind.
To return to the sociological debates of an earlier era which informed CA in particular and which are still central, a lot of what is at stake here refers to the representation of the self in different circumstances. ‘Representation of the self’ is, of course, Goffman’s phrase.31 Of particular interest here is the concept of the ‘front stage’ (what the audience sees of the actor, or the patient of the doctor) and the ‘back stage’ (where the audience cannot go and which the patient therefore does not see).31 Monrouxe et al.,32 writing about bedside teaching encounters, see ‘the front stage as where students perform as doctors, and back stage as where teachers run through performances and correct actions’. The simulation game is the front stage. If students perform, so do doctors, and so do we all. Whether we regard a real-life ‘performance’ (the way Dr X talks to his real patients during a real clinic) as authentic has more to do with conventions than anything else. Thus, people from some cultures may find British people deceitful because they do not understand the rules of politeness at work when someone in the UK says ‘Well, it’s not perfect’, rather than ‘It’s really awful’. Some students will say that assessments by role-play are unrealistic, but will never consider how ‘realistic’ a multiple-choice question is. It is said (although this story sounds too good to be true) that Picasso was once upbraided by a man whose wife he had painted in one of his more idiosyncratic styles. ‘That’s nothing like my wife,’ the client complained. He produced a photograph. ‘Look!’‘Hmmn,’ said Picasso. ‘She seems rather small and flat.’ What we accept as a true representation is driven by culture, habit and convention.
The central educational point is that aiming for authenticity is useful, but does rather miss the point. The crux of the matter concerns whether a particular approach, or a particular scenario or a particular way of handling the ‘hybrid’ situation has educational value. In this respect, it is worth noting the points made by Seale et al., who reported that one doctor in their study used humour to ‘exploit the ambiguous realism in the role-playing frame’.12 It was this doctor who ‘was to show the most capacity for learning the new skills of elicitation, empathy, and so forth’.12
Understanding this point is at the heart of training using SPs. The artificial nature of the situation can be an advantage. For example, students can be invited to act in a manner completely different to the way they would normally behave; multiple students can consult with the same patient for a few minutes each to establish how they differ from one another; a stop-start method can be applied, as can more general other techniques from the tradition of applied theatre. Obviously, it is very important that SPs are trained to understand their dual role in the simulated consultation.
As regards testing, we are only at the beginning of our understanding. After all, most skills-based testing seeks to be authentic. Being tested on taking a blood pressure mostly involves taking a blood pressure. Of course, other issues pertain to the extent to which high-fidelity simulations will replace real tests, and the extent to which communication will continue to be tested as a set of skills, rather than within a wider context such as that of professional development.
With regard to research, issues such as the extent to which training has an impact on the overall language or linguistic detail of how students interact, the degree to which good performance in simulation correlates with good performance in the clinic, and how the performance of individuals might change over time remain to be investigated.
Finally, we hope to have demonstrated how a closer look at language can help us to make sense of situations in which communication takes place. Training people (whether they be medical educators, doctors or students) in the microanalysis of language is likely to improve observational and analytical skills, which, in turn, will benefit understanding of social interactions.