1.1 ‘Good’ Communication: Perspectives from Deontology, Consequentialism and Virtue Ethics

To set the scene for debates that will recur in later sections, we need to distinguish three approaches to deciding whether communication (or, indeed, any behaviour) is ‘good’ [Reference Duggan, Geller and Cooper2]. One is to identify specific communication behaviours as inherently good. We shall see this approach in the concept of ‘communication skills’ (Section 5), which denotes behaviours such as making eye contact, offering empathic statements, or summarizing what has been said as intrinsically valuable ones that practitioners should therefore be taught to use. This takes a philosophical position called deontology, whereby the quality of an action is inherent in the action. From this perspective practitioners might, for example, smile and greet patients to open a consultation because they have been taught that these are the correct behaviours to start an interaction. We shall see through the following sections that deontology has been the dominant approach in clinical communication; professional, governmental or institutional guidance tells practitioners how they should communicate in different situations, published curricula tell educators which communication skills they should teach practitioners to use, and communication rating scales evaluate practitioners’ performance of those skills. Even ‘pocket cards’ are available to tell practitioners how to communicate in specific situationsFootnote ¹. In Section 4, we shall see that some very influential frameworks of guidance for clinical communication – ‘shared decision-making’ and ‘patient-centred care’ – also take an essentially deontological position in providing instructions for how to communicate, procedures and materials to help practitioners follow those instructions, and assessment methods to measure how well they do so. However, we shall also see that it is very hard to create behavioural rules that cater for every situation that can arise in clinical care.

In the current era of evidence-based medicine, practitioners are expected to treat patients using methods justified by evidence of their effects. In clinical communication, too, educators and researchers therefore now often cite outcome evidence – showing, for instance, that when practitioners communicate in certain ways their patients are more satisfied or recover more quickly. This takes a consequentialist approach, whereby the consequences of a behaviour determine whether it is good. From this perspective, practitioners who smile and greet their patients might do so because they have seen evidence that it improves patient satisfaction or adherence to treatment recommendations.

A consequentialist approach to clinical communication is not straightforward, however, because any element of communication rarely has a single outcome. For instance, providing medication or treatment that patients request might enhance their satisfaction, but at the cost of iatrogenesis if the requested treatments are harmful, or of wasting scarce resources where they are pointless. This is not just a theoretical risk. Later, we will see surprising findings where doctors taught to communicate in ways that improved patients’ satisfaction achieved poorer clinical outcomes for those patients [Reference Kinmonth, Woodcock and Griffin3]. Unfortunately, communication researchers have prioritized ‘soft’ outcomes, such as patients’ satisfaction or mood, over ‘hard’ clinical outcomes. Indeed, the relatively small-N designs of most communication research, while powerful enough to detect large effects on soft outcomes, have insufficient power to detect the smaller effects that should be anticipated on more distal outcomes of morbidity and mortality. Even if outcome evidence from larger-scale studies were available, however, translation from such studies to practitioners’ moment-to-moment communication with individual patients leaves much to the practitioners’ interpretation and judgement (see Sections 5, 6). So good communication cannot simply be defined as that which achieves the best outcomes.

Just as in ordinary life, clinical communication can express values. For instance, we expect practitioners to be polite to patients, regardless of whether they have been told to be polite, and regardless of any evidence that politeness influences outcomes. Politeness indicates respect, and patients meeting rude or discourteous practitioners would feel devalued. And we expect practitioners, not only to respect their patients, but to be genuinely interested in them and motivated to discover what they need and how best to help, even in the absence of trials comparing outcomes between motivated and bored practitioners. Similarly, we expect practitioners to be knowledgeable and capable of good judgement, without waiting for evidence that knowledge and judgement improve care. Recognizing this dimension to clinical communication leads us to the third philosophical stance for identifying good communication: virtue ethics [Reference Gauthier4].

This approach, with origins traceable to Aristotle in the West, and Confucius in the East [Reference Van Norden5], means shifting our focus from communication behaviours and their consequences to communicators’ character and to the motivations, attitudes and knowledge that underlie their judgements about how to communicate [Reference Stiefel and Bourquin6]; that is, communication behaviour is good where it is chosen wisely, for good reasons. From this perspective, practitioners would smile and greet patients, not because they have been taught to or because they expect better outcomes, but because they respect and value them. The critiques in this Element will return us repeatedly to the conclusion that a virtue ethics perspective is a better fit to clinical communication than is the current emphasis on deontology and consequentialism. We shall see, though, that a virtue ethics perspective does not simplify educators’ or researchers’ tasks. Indeed, it exposes dilemmas that an exclusive focus on behaviours and their consequences does not. For instance, how we communicate indicates who we are as people – our personality and our values. So can we change practitioners’ communication without changing in some way who they are as people? Conversely, when practitioners are taught to communicate in a new way, are they no longer being ‘authentic’? We return to these dilemmas later (Sections 6, 7), when we examine how the authenticity that both patients and practitioners seek in clinical relationships can be reconciled with practitioners’ need to learn new ways to communicate with their patients.

1.2 What, and Who, Is This Element For?

The aim is that readers are informed and critical about this field. They need to be ready to evaluate claims they might read about communication in policy documents, research literature or educational material; and they need to be able to critique their own communication or communication that they experience or observe. So the Element is not just for academic readers. In informing practitioners and stimulating them to be critical, it can help them be more flexible and creative in finding solutions to challenges in communicating with their patients. The Element therefore takes a different stance from that of many existing textbooks on communication for doctors and other healthcare practitioners. Rather than a manual of communication skills, it is an introduction to ‘clinical communication science’ with the aim that clinical communication can sit alongside other clinical subjects in being founded on good science while also grounded in clinical reality [Reference Gulbrandsen, Gerwing and Landmark7].

The first sections therefore focus on basic principles that underlie the questions that clinical communication researchers ask or the goals that educationists choose, disentangling different influences that have shaped these principles. Reflecting the ethical dimension of clinical communication, Section 2 begins with principles that arise from ethical perspectives. Section 3 turns to principles derived from scientific models, and to the theory and research that allows us to critique these models. Section 4 critiques principles that have arisen from some communication ‘technologies’ that have shaped communication research and education and that have embedded in the field assumptions about how practitioners should communicate. Section 5 addresses how the technology of communication skills teaching has sought to implement those assumptions.

It is, though, of little practical use to think about principles in only general terms. Scholars applying virtue ethics to practitioners’ clinical interactions draw on Aristotle’s warning that the wisdom to make good judgements exists, not as theoretical or technical knowledge, but in making good judgements in practice [Reference Kinghorn8,Reference Kotzee, Paton and Conroy9]. Similarly, the medical anthropologist, Arthur Kleinman, argued that abstract principles remain vague generalizations with little real-world purchase unless they are tested and honed by studying what they mean in routine practice [Reference Kleinman10]. Section 6 therefore uses cancer care as a ‘case study’ to examine in detail what the principles identified in the preceding sections mean in clinical practice. Cancer care poignantly focuses the vulnerability and dependence that will emerge through this Element as central to the patient experience and to the psychological challenges of clinical communication. Moreover, cancer care has been an intense focus of communication research, guidance and education for more than fifty years. It therefore illustrates well the beliefs and assumptions that motivated and guided the pioneers in clinical communication research and education, and the ways in which these are challenged by more recent research. Finally, Section 7 identifies some pointers to how, responding to these challenges, the future of communication education might diverge from its past.

The field of clinical communication is vast, so this short Element is necessarily highly selective. It focuses on themes related to patients’ vulnerability and dependence, thereby arguably putting what defines being a patient at its centre. And it emphasizes the practitioner–patient relationship as the vehicle for responding to patients’ vulnerability and as the context within which the many functions of communication must be achieved. It is therefore a ‘primer’ for readers who will go on to consult literature and texts about those functions and who can read about them in light of the ideas introduced here.

Of all the health professions, doctors have historically been the main target of communication research, guidance and teaching; so most of the material cited here will be about doctors or aimed at them. Nevertheless, roles are changing, and other professions, such as nurses, physiotherapists or pharmacists, can also take clinical responsibility for patients’ care. Therefore, this Element, although informed mainly by literature concerning doctor–patient relationships, is potentially relevant more broadly to professionals consulted by patients seeking clinical expertise in response to feelings of vulnerability associated with physical disease. For clarity, the text will refer explicitly to doctors only when citing evidence or guidance that specifically concerns them. Elsewhere, the generic term ‘practitioner’ will indicate the potentially wider relevance of the material or ideas in play. Ultimately, it is for readers to make their own judgements about the extent to which ideas in this Element apply to their own context. The central thesis of the Element is that understanding ‘clinical communication science’ can equip readers to make good judgements, whether about communication in clinical practice, communication teaching, or communication research.

2.1 Patient Self-Determination: Legal Frameworks of Informed Consent

The beginning of legal constraint on doctors’ authority was marked by a judgement in a New York court in 1914 [Reference Murray15]. The case did not change law, but changed how existing law was interpreted in a new age of developing medical technology and changing societal expectations about individual rights. It arose because a patient sued a surgeon whose intervention had gone beyond what the patient had previously agreed to. Having accepted examination under anaesthesia the patient awoke to discover that the surgeon had identified and removed a tumour. Side effects of surgery led to circulatory problems that later necessitated amputation of some of the patient’s fingers. The eventual ruling stated that ‘every … adult … has a right to determine what shall be done with his [sic] own body; and a surgeon who performs an operation without his [sic] patient’s consent commits a battery’. In other words, it equated treatment without consent with being physically attacked. Since this landmark judgement, the principle of ‘informed consent’ has evolved, albeit at different speeds in different countries. In general, the criterion for ‘sufficient information’ evolved from the information that a ‘reasonable practitioner’ would give to what a ‘reasonable patient’ would expect. In the UK in 2015, a landmark legal case moved the criterion further; for a patient to be adequately informed of the risks of treatment, the practitioner must tell the patient, not only what a ‘reasonable patient’ would consider important but what that specific patient would think important [Reference Dunn, Fulford and Herring16]. Practitioners need, therefore, to be able to detect and respect the values and concerns of each patient, an injunction to which we will return repeatedly in this Element.

The trend towards respecting patients’ ‘right to determine what shall be done with his own body’, in the words of the 1914 judgement is, in practice, highly constrained by culture and politics. For instance, in seeking to reject treatment that prolongs life, or to request intervention to end life, there are intense contests when patients or families draw on evolving cultural views of individuals’ rights in the face of legal constraints and religious or political opposition [Reference Botti and Vaccari17]. Similarly, the United States illustrates how decades of extension of women’s rights over doctors in relation to contraception and abortion can rapidly be reversed when political power shifts. Aesthetic surgery, where patients seek surgical solutions for psychological reasons, remains highly contentious, with laws sometimes enacted to constrain patients’ access to certain procedures [Reference Turillazzi, Neri and Riezzo18]. Contest over individuals’ rights in relation to doctors arises also when states seek to mandate interventions. For instance, across Europe different legal frameworks around mandating or encouraging vaccinations against contagious diseases in children [Reference Bozzola, Spina and Russo19] and (in light of the COVID-19 pandemic) adults [Reference Gibelli, Ricci and Sirignano20] reflect contrasting cultural and political views of health as an individual or collective responsibility [Reference Chirico21,Reference Williams22]. Clearly, the ‘right to determine what shall be done with [one’s] own body’ cannot be understood simply according to a linear progression from medical paternalism to patient emancipation. Self-determination in practice has proved more nuanced than envisaged in the idealized statement from the New York court.

2.2 Patient Self-Determination: The Bioethics of Autonomy

The cultural trends that led to legal assertion of patients’ rights have been expressed in academic and professional perspectives on clinical relationships and we shall see that these, too, illustrate the practical complexity around respecting patients’ rights as autonomous individuals. In Section 3 we will see also how, within medicine itself, new ideas were arising that centred clinical care more on patients’ needs than on their doctors’ traditional priorities (Section 3). And in Section 4, we will see how social scientists took up the cause of patient empowerment, thereby aligning patients with other groups marginalized by powerful interests. First, though, we address how bioethicists’ ideas about ‘patient autonomy’ evolved as a moral foundation for those developments; and we shall see that the initial simplicity of an idealized view has evolved into a more nuanced picture.

Beauchamp and Childress’ famous statement of the principles of ethical clinical practice [Reference Beauchamp and Childress23] enshrined respect for patient autonomy as one of doctors’ canonical ethical obligations. Autonomy was equated with self-determination; patients need the information that equips them to decide for themselves about their treatment and care, and they should make such decisions freely, without coercion. By being enabled to make treatment decisions, patients were defended from the dangers associated with medical paternalism – that decisions would reflect doctors’ preferences or assumptions instead of individual patients’ values and needs. With hindsight, however, it became clear that asserting patients’ right to self-determination brought problems. For instance, where they seek interventions that would be pointless or harmful, or that might squander scarce healthcare resources, acquiescence adheres to the principle of autonomy but contravenes Beauchamp and Childress’ other fundamental principles: beneficence (provide what is in patients’ interests), maleficence (do not harm patients) and equity (treat all patients according to their need). Taken to the extreme, an over-riding emphasis on patients’ self-determination could replace paternalism by consumerism, whereby doctors are reduced to offering a cafeteria of services from which patient-customers can select [Reference Goldstein and Bowers24]. While the lens of consumerism might help illuminate limited areas of clinical practice such as aesthetic surgery, in which requests for clinical intervention can reflect social and cultural factors as well as personal needs [Reference Martinez, Dixit and White25,Reference Locatelli, Boccara and De Runz26], we will be concerned here with patients whose need for care is defined by feared or confirmed physical disease.

Equating patient autonomy with self-determination presents a further difficulty; it is rare that patients or their families are psychologically equipped to make good decisions. First, the worry or distress associated with being mortally ill, or with the fear of being so, are inimical to the rational and objective judgement that clinical decisions need. Second, it is implausible that patients could routinely be brought to the level of understanding and judgement necessary for complex clinical decisions, that practitioners take many years to reach. In Section 4 we will examine the concept of ‘shared decision-making’ (SDM), designed to help patients be self-determined in a context in which they lack technical knowledge and expertise or find it hard to reason objectively. However, more recent developments in bioethics suggest that aiming for self-determination as the way to respect patients’ autonomy is mistaken.

Concern with self-determination expresses a normative element of Western culture: the valuing of individual decision-making and responsibility. By contrast, in some other cultures, important decisions for individuals are made in groups of which they are a part, particularly the family [Reference Gilbar and Miola27]. Even in the West, the value attached to self-determination is not so robust that it can survive the threat of mortal illness, as a Canadian study showed strikingly more than three decades ago. Degner and Sloan [Reference Degner and Sloan28] asked two samples of people to rate the degree to which they would want to make decisions about treatment for cancer or have their doctors make decisions for them. The sample of healthy people, asked to imagine needing treatment, predominantly sought to be in control of decisions. By contrast, the sample of people recently diagnosed with cancer mostly wanted the doctor to be in control. The study’s message is that being mortally ill changes how we value personal control. It makes us look to practitioners to take care of us, rather than to our own sense of self-determination.

There is another reason to be cautious about taking too literally the value attached to individual responsibility and self-determination in the West. It probably tells us more about self-presentation than about people’s wish to be self-determined. That is, it is culturally valued in the West to appear – and feel – in control of one’s own life. Therefore, when patients are asked in questionnaires or interviews whether they want to be informed and to make decisions about their treatment for themselves, they typically say ‘yes’ [Reference Meredith, Symonds and Webster29,Reference Rankin, Newell and Sanson-Fisher30]. But this probably reflects what a New England Journal of Medicine editorial called an ‘information-as-power ethos’, rather than a true desire for information and choice [Reference Rosenbaum31]. In Section 6 we shall see that patients (or, where patients are children, their parents) who present themselves as wanting to be fully informed and to make decisions for themselves go on to describe needing their doctors to constrain information and take responsibility for decisions. Therefore we should be wary when self-report questionnaire studies appear to confirm patients’ desire for information and choice [Reference Salander and Henriksson32]. Some questionnaires simply tell us about language [Reference Skelton33]. For example, when patients endorse questionnaire items to indicate that they want ‘enough time’ to make decisions [Reference Rankin, Newell and Sanson-Fisher30] the simplest explanation is that, in English, it is impossible to say that one does not want enough of something! Practitioners are often advised to give patients ‘the information they want to have’, but asking patients what they want to know is clearly not straightforward. Responses to such a question are fundamentally uninformed; patients cannot make informed decisions about what information they want until they know what that information is [Reference Rosenbaum31].

Western individuals’ health-related decisions that, at first sight, seem to be expressions of self-determination based on individual preferences, such as the choice to be vaccinated for COVID-19, can on closer scrutiny be seen to have a strong relational element; individuals’ choices reflect social relationships and obligations and broader cultural influences [Reference Paul, Zimmermann and Corsico34]. Reflecting the need for an understanding of autonomy that does not simply recycle cultural norms around self-determination, Western bioethics literature has seen growing interest in ideas of ‘relational autonomy’ [Reference Carter, Entwistle and Little35–Reference Dove, Kelly and Lucivero38]. Originating in feminist critiques that equating autonomy with self-determination depicted individuals as unrealistically self-sufficient and isolated from a social network, relational approaches emphasize the social and cultural context within which people make choices. From a relational perspective, therefore, individuals’ autonomy can arise, not from making decisions for themselves, but from being able to rely on other people who respect, value and understand them and on whom they can depend to make decisions that are right for each individual.

A relational perspective has big implications for how practitioners might respect patients’ autonomy. Most strikingly, asking patients or their families to make choices could even reduce autonomy. In a report pointedly entitled ‘Autonomy gone awry’, Orfali and Gordon [Reference Orfali and Gordon39] compared parents’ experiences of neonatal care between France and the United States. In France, satisfaction with care, and a sense of personal autonomy, arose out of a strong doctor–patient relationship based on a traditional emphasis on medical authority and responsibility. By contrast, in the United States, with care characterized by greater emphasis on parental decision-making, parents were less satisfied, felt less resilient after the loss of a baby, and felt poorer rapport with doctors. Compared with an approach to autonomy that equates it with self-determined decisions, a relational view invites many other ways to respect patients’ autonomy. For instance, we shall see (Section 6) that, when doctors give reasons for their treatment decisions, patients can better ‘own’ those decisions [Reference Kukla40,Reference Mendick, Young and Holcombe41]. Moreover, a relational view takes us beyond decision-making for other ways to enhance autonomy. It points to mundane aspects of social interaction that communicate respect and dignity instead of disrespect and arrogance, such as being polite and well-prepared, acknowledging patients as equals, taking seriously what patients know or apologizing when late [Reference Beach, Branyon and Saha42–Reference Beach, Duggan and Cassel44]. From a relational perspective, patients’ autonomy does not, of course, just require patients to trust their practitioners to understand and look after their interests; it depends on practitioners genuinely being trustworthy in understanding and respecting each patient’s interests and in having the expertise to address them. As Pellegrino explained, ‘the vulnerability and dependence of the sick person forces him [sic] to trust not just in his rights but in the kind of person the physician is’ [Reference Pellegrino and Shelp45].

Whereas relational autonomy marks a change of direction for Western bioethics, it converges with long-established perspectives in less individualistic cultures [Reference Turoldo46]. For instance, Nakata et al. [Reference Nakata, Goto and Morita47] showed how the concept of ‘motenashi’, embedded in Japanese society, could inform bioethics of critical care, but their paper is relevant more broadly. According to motenashi, people who are respected should not be burdened with the expectation or obligation to make decisions to secure what they need. As Nakata et al. explain, ‘The respected person does not have to make any decisions because decision-making is always accompanied with risks and responsibilities.’ Rather, the obligation is on others to know what they need and to provide it. As a cultural practice, motenashi obliges hosts to discover and to respect what will satisfy and please their guests. As a bioethical principle, it obliges practitioners to understand and respect what matters to their patients. It therefore shows clearly why a relational approach to autonomy does not presage a return to paternalism, whereby decisions reflect what suits practitioners or what practitioners think best.

2.3 Conclusion

Redressing the imbalance of power between patients and practitioners from a legal perspective has proved more complex than a simple, linear process of patients taking ever more of practitioners’ power for themselves. Similarly, autonomy in bioethics is not a ‘zero sum’ whereby practitioners must simply cede authority for patients to gain it; the concept of relational autonomy is more nuanced than the earlier equation of autonomy with self-determination. Recognizing this complexity does complicate this Element’s task. Equated with self-determination, patient autonomy is relatively straightforward to deliver; practitioners give patients the information they need to make decisions about their care and allow them to make those decisions freely; educators can therefore promote patient autonomy by teaching practitioners the behaviours necessary to inform patients and share decision-making with them. Understood in this way, respect for autonomy would belong in deontology ethics; practitioners achieve ethical practice when they perform certain behaviours. By contrast, relational autonomy is not reducible to behavioural rules. Instead, it points to a virtue ethics perspective because it concerns the qualities that practitioners must have; not only clinical expertise and wise judgement but the motivation and ability to understand and respect what matters to each patient.

3.1 The ‘Psychotherapeutic Model’ of Clinical Communication

While the historical paternalism of medical care was being challenged in society and in bioethics, there were challenges from within medicine too. In 1977, a US psychiatrist, George Engel, advocated a revolution in medical thinking and practice [Reference Engel48]. Engel criticized medicine for being ‘doctor-centred’ or ‘disease-centred’, marginalizing patients’ emotional and social needs and their preferences and concerns. He blamed two fundamental principles of medical thinking. ‘Reductionism’ was the belief that everything that mattered in clinical care was biological and ultimately explicable at the molecular level. ‘Dualism’ was the assumption that the patient’s body and mind were separate and that doctors needed only to concern themselves with the body. Opposing these principles, he advocated a ‘biopsychosocial model’ of clinical practice, which recognized that both the causes and effects of illness often lie outside biology, being found instead in interactions between patients’ biology, psychological processes and social context. It followed, he argued, that doctors needed to be concerned, not just with their patients’ physical symptoms, but with the social and psychological context of those symptoms. They needed to understand, for instance, how patients’ emotional state, beliefs or social circumstances might be causing or exacerbating illness, and they needed to address the psychological and social consequences of the illness. Engel’s attempt to change the culture of medical care by centring it on the whole patient instead of the biological disease helped set the scene for subsequent development of practices to implement ‘patient-centred care’, to involve patients in decision-making, and to teach practitioners to be more holistic in their communication, to which we turn in Sections 4 and 5. Here, we examine how views about the nature of the clinical relationship evolved following Engel’s paper.

Key pioneers of clinical communication research and teaching were, like Engel, psychiatrists or with training in psychology or social science. They therefore brought a professional concern with patients’ psychosocial worlds which, when applied to care of patients with physical illness, offered to help implement the biopsychosocial model. From their perspective, the clinical relationship is an emotional connection. It develops over time as practitioner and patient get to know each other, being built through emotional talk just like relationships in psychiatry or counselling; the practitioner shows concern with the patient’s emotional feelings, the patient discloses them, and the practitioner in turn responds empathically [Reference Salmon and Young49]. On this reasoning, the strength of the relationship, and the degree to which it comforts distressed patients, depends on this emotional talk rather than on instrumental talk about the illness and treatments. This view of clinical relationships became very influential and continues to shape assumptions and practices in clinical communication education and research. Indeed, it is so influential that it is widely seen as axiomatic, and rarely identified in clinical communication literature as a contestable theory. However, we need to examine the validity of extending this model from the psychiatric or counselling clinic to the very different context of clinical relationships in physical illness.

The extent to which this ‘psychotherapeutic’ model of clinical relationships shaped researchers’ and educators’ expectations of practitioners’ communication is illustrated by a consensus statement in 2010 by European clinical communication experts [Reference Stiefel, Barth and Bensing50]. In explaining why oncologists should receive communication training, it states that they often disregard the emotional and social dimensions of their patients’ illness, and that their preference for focusing discussion on medical information can be a way to avoid such talk. Common research themes in clinical communication, and the language with which findings are reported, also illustrate how the psychotherapeutic model has become ‘taken for granted’. For instance, many papers have reported on the level of empathy that doctors show their patients, lamenting their failure to show enough. In one influential paper, Levinson et al. [Reference Levinson, Gorawara-Bhat and Lamb51] reported on the number of cues to personal issues or emotional feelings that were present in patients’ talk when they consulted surgeons or primary care doctors. By describing these cues as ‘opportunities’ for the doctor to ‘demonstrate understanding and empathy’ and thereby ‘deepen the therapeutic alliance’, they reveal how they think that those doctors should respond: in English, an ‘opportunity’ is not a neutral word, but indicates the expectation that it should be grasped. The key finding was that doctors responded empathically to 38 per cent of cues in surgical consultations and 21 per cent in primary care. The authors’ language again goes beyond scientific observation to indicate that the doctors had fallen short. For instance, ‘frequently [doctors] missed opportunities to adequately acknowledge patients’ feelings’. Words like ‘miss’, ‘opportunity’ and ‘adequately’ all add rhetorical power to the argument that doctors failed. But the paper does not specify to what extent they failed; it does not state whether, for instance, doctors should respond empathically to 100 per cent of empathic opportunities, 50 per cent or some other proportion. The reason is that there is no evidence to indicate what a desirable percentage should be. Their conclusion also disregards whether the doctors responded by addressing the clinical problem that elicited the cues [Reference Hsu, Saha and Korthuis52], for example, by providing relevant information [Reference Kuchinad, Park and Han53]. Similarly, many researchers have shown that communication training can increase practitioners’ demonstrations of empathy [Reference Stepien and Baernstein54], but without specifying a target level. Simply assuming that more is always better seems implausible, and closer to a slogan than a scientifically grounded prescription. In the limiting case, it might absurdly predict that the strongest clinical relationship would be achieved by exclusively emotional talk that disregards clinical care.

The apparent face validity of the psychotherapeutic model when applied to clinical relationships in physical healthcare therefore belies some conceptual difficulties. Scrutinizing the model further reveals several reasons why it does not clearly fit healthcare for physical illness. First, the suggestion that clinical relationships in this setting need an explicitly emotional connection between practitioner and patient is questionable. After all, practitioners need to be dispassionate and objective with their patients, not emotionally involved [Reference Stiefel55], and need effective clinical relationships even with patients whom they dislike. The performative language often used in educational or research literature seems to acknowledge this problem in its frequent reference to ‘displaying’ empathy, rather than ‘being’ empathic, thereby degrading the concept of an emotional connection to something enacted rather than felt. Moreover, the overwhelming emphasis on empathy risks encouraging practitioners merely to respond empathically to patients’ emotional cues without first identifying what gave rise to them and what practical responses might be needed [Reference Stiefel, Bourquin and Salmon56].

A further limitation of the psychotherapeutic model is its oversimplification of relationships [Reference Young, Ward and Forsey57]. First, emotional and instrumental communication are not as distinct as the model assumes. Emotional needs can be communicated and addressed by factual talk, and vice versa [Reference Brandes, van der Goot and Smit58,Reference Semmer, Elfering and Jacobshagen59]. For example, patients can secure general practitioners’ (GPs’) clinical engagement with their symptoms by emphasizing how worried or upset they feel [Reference Salmon, Humphris and Ring60], and patients with advanced cancer mentioned the death of a loved one after oncologists failed to respond to previous clues to their need to talk about their own impending death [Reference Salvadé, Stiefel and Bourquin61]. Second, in everyday life, relationships are multidimensional and often asymmetric; the different parties to a relationship can feel different levels of connection. Literature and drama poignantly exploit the divergence that can arise between the sense of relationship felt by each party (love unrequited, for instance), and it is obvious that a strong relationship can exist in the absence of explicitly emotional talk. By its ‘one-size-fits-all’ approach, the psychotherapeutic model simplifies relationships in another way. It has little to say about how they vary between different contexts; for example, how they differ between clinical settings, between surgeons and nurses, or amongst individual patients and individual practitioners.

Finally, and crucially, the psychotherapeutic model disregards the defining feature of clinical relationships in physical healthcare: the inherent asymmetry of vulnerability, knowledge and authority when patients are – or fear being – physically ill [Reference McWilliam, Brown and Stewart62]. Patients are typically suffering or frightened, and consult practitioners who they believe have the knowledge and authority to help them. In this context, the more plausible starting point for understanding clinical relationships is that patients’ main priority is not their emotional feelings, but the health problem provoking those feelings.

3.2 Attachment Theory and the Clinical Relationship

Instead of importing a psychiatric model of clinical relationships into physical medicine, we need a way to understand these relationships that is better suited to their very different context. Recently several writers have drawn on attachment theory, which has asymmetry of vulnerability and authority at its centre [Reference Salmon and Young49,Reference Palmer Kelly, Tsilimigras and Hyer63–Reference Maunder and Hunter65]. The theory emerged from studies of children with their parents [Reference Bowlby66,Reference Bowlby67] and centres on a simple observation that, when a child feels vulnerable, its overwhelming priority is to be close to a familiar caregiver – usually the parent. The parent is its ‘attachment figure’, or ‘secure base’, that helps the child feel safe and cared for. The child has an emotional bond with the parent, called an attachment. Crucially, attachment figures are not substitutable; if one is not available, someone else with whom the child has not formed an attachment would not suffice. Attachment theory is not just a theory of dependence; the security provided by an attachment figure who is available and attentive is the basis for the child feeling confident to explore and become independent. For instance, a young child taken to a nursery by a parent might run off and explore, but will periodically look back at, or return to, the parent, confirming that the world is still safe to explore further. The convergence with the earlier discussion of ethical theory is clear; autonomy when one is vulnerable depends on being able to trust in others’ authoritative and attentive care.

Although originating in studies of childhood, attachment theory has been applied also to close relationships in adulthood, particularly in the context of vulnerability as a powerful trigger for attachment processes. However, attachment relationships for adults differ in many ways from those seen in children [Reference Mikulincer and Shaver68]. With adulthood comes the ability of symbolic representations of the attachment figure to substitute for proximity; for example, the promise of an impending meeting with the attachment figure, or just knowing that he or she is available [Reference Mikulincer and Shaver69]. In some situations, adults can also need appropriate distance and separation to find a sense of safety [Reference Birtchnell70]. Most obviously, parents generally give way to other attachment figures – in particular, romantic partners. However, adults can compartmentalize needs for different kinds of support; different people might offer a ‘secure base’ but in relation to different sources of vulnerability. Moreover, whereas being comforted is usually the main indicator of safety for a child, adults seek more evidence of the authority of a potential attachment figure to ameliorate threats. Therefore, given the vulnerability inherent in physical illness, or the threat of it, patients can see a practitioner as having properties of an attachment figure in providing a ‘secure base’ in the context of the threat of illness [Reference Maunder and Hunter65]. Several implications for clinical relationships follow from this reasoning, and they diverge greatly from those of the psychotherapeutic model.

First, it is primarily practitioners’ authoritative and expert biomedical care, not their emotional engagement, that allows patients to see them as a ‘secure base’. Therefore, a practitioner who focused too much on patients’ worry would probably be harder for patients to see as a ‘secure base’ than one who focused authoritatively and expertly on the symptom or disease causing the worry. Therefore patients can prefer doctors to respond to their emotional cues with practical clinical information rather than with emotional talk [Reference Kuchinad, Park and Han53]. Of course, there will be situations where patients need practitioners who are explicitly empathic, and who spend time exploring their feelings and concerns, particularly in primary care where GPs often have to disentangle confusing presentations of physical and emotional symptoms. But, as the starting point for thinking about how practitioners comfort patients, attachment theory points in a very different direction. Consistent with an attachment perspective, we shall see in Section 6 that practitioner behaviours that allow patients to see them as attachment figures do not resemble the psychosocial exchanges advocated in the 2010 consensus statement described previously [Reference Stiefel, Barth and Bensing50]; instead patients identify behaviours indicating expertise, authority and commitment, and patients or family who feel in emotional turmoil can sometimes need practitioners who provide a secure base by being calm and rational instead of engaging with their distress [Reference Young, Hill and Gravenhorst71] (see Box 1).

From an attachment perspective, therefore, patients’ sense of emotional connection to practitioners does not need to arise over time from explicitly emotional talk. Rather, it arises out of patients’ vulnerability and associated attachment needs and their ability to see the doctor as a secure base. The multidimensionality of relationships (which the psychotherapeutic model does not easily accommodate) therefore becomes understandable; patients’ sense of emotional connection is not based on observable emotional engagement, or on reciprocal emotional feelings in the practitioner. Indeed, attachment theory helps us understand how a strong emotional connection for patients can arise from practitioners’ conscientious clinical care, whereas clinical communication educators and researchers have generally portrayed clinical care and emotional engagement as distinct [Reference Salmon and Young72]. Recognizing this exposes a danger for practitioners who, misled by not feeling that they have an emotional relationship with a patient, do not appreciate the intensity of connection that the patient can feel to them. They risk shattering patients’ ability to construct them as a secure base where they do not display conscientious and authoritative care, for example, by failing to honour promises to a patient, forgetting a patient’s name, not being well-informed about a patient or appearing uninterested in the patient or family [Reference McWilliam, Brown and Stewart62,Reference Wright, Holcombe and Salmon73].

A ‘secure base’ does not mean a practitioner who makes unrealistic promises to cure a patient or remove pain. Instead, it signifies the practitioner’s commitment to be with the patient through the challenges of illness and not to abandon the patient. Therefore, what makes a practitioner a ‘secure base’ depends on the context. For instance, a study of palliative care patients being cared for at home identified aspects of care that helped patients and their families feel they had a ‘secure base’ when cure was impossible [Reference Milberg, Wahlberg and Jakobsson74]: knowing that practitioners were available, seeing the same ones rather than unfamiliar faces, and feeling that the practitioners respected the patients and their families by listening and taking their concerns seriously. Even during the COVID-19 pandemic, when practitioners often had to provide support remotely, by telephone, it was possible for patients or family members to feel that a practitioner was ‘with’ them when the practitioner addressed clinical care conscientiously and authoritatively [Reference Ricco, Fiorani and Ferrara75,Reference Borghi76].

An attachment perspective also helps understand heterogeneity amongst patients in the way that clinical communication supports clinical relationships. In the acute crisis of a frightening illness or diagnosis, most patients’ concern will be with their practitioners’ ability to take the role of authoritative and conscientious attachment figures. Later, to learn effective and confident self-management as a long-term condition proceeds, patients will need practitioners who, while still being trusted to be authoritative and available, also respect and value patients’ own developing perspective and authority. Moreover, patients’ readiness to form trusting relationships will be influenced by their childhood experience of attachment [Reference Mikulincer and Shaver68]. In general, a child experiencing attentive and consistent care and affection will be better prepared as an adult to enter close, trusting relationships than one without that experience. More specifically, according to attachment theory childhood experiences can independently shape adults’ assumptions about other people and about themselves when in close relationships. These sets of assumptions are sometimes called ‘mental models’ [Reference Griffin and Bartholomew77,Reference Ciechanowski, Walker and Katon78], and each can be positive or negative to different extents. The model of ‘others’ is what the adult expects of other people: being trustworthy, caring and considerate where the model is positive, or untrustworthy, uncaring and rejecting where negative. The model of ‘self’ is what people feel about themselves. The positive end of this dimension denotes feeling deserving of care and affection; at the negative end people feel unworthy, even unlovable. Putting these two mental models, or dimensions, together, yields a classification of 4 attachment styles [Reference Ciechanowski, Worley and Russo79] (Box 2). Using questionnaires to categorize attachment style, around half the general population fall into the ‘secure’ category, defined by positive models of both self and other, the other half being distributed across the remaining styles [Reference Bartholomew and Horowitz80]. Differing communication needs between patients can reflect these attachment styles (Box 2). Patients whose experiences have left them distrustful of other people (negative model of ‘other’) will need sensitive prompting to disclose symptoms or concerns. Patients who have learned to feel unloved, or unworthy of care and affection (negative model of ‘self’), will need more reassurance than do others that their symptoms and concerns are important to practitioners and will be taken seriously. Patients hardest to reach will be those in the ‘cautious’ quadrant, where both models are negative, and some might need expert psychological help [Reference Tan, Zimmermann and Rodin64,Reference Ciechanowski and Katon81–Reference Hunter and Maunder84]. Evidence that patients with one of the ‘insecure’ attachment styles reported a poorer relationship with their doctors than did ‘secure’ patients [Reference Zaporowska-Stachowiak, Stachowiak and Stachnik85] is a warning about the additional help that these patients need to form trusting relationships.

Box 2 Adult attachment styles and clinical relationships

Adults’ mental models of self and others vary from negative to positive, defining four attachment styles. Names here are according to Ciechanowski [Reference Ciechanowski, Worley and Russo79], replacing traditional (and pejorative) terms in brackets. For each style, text illustrates implications for patients’ relationships with practitioners (see 64, 81, 84).

Attachment theory also alerts us to heterogeneity between practitioners. Because their capacity to be a ‘secure base’ depends on their power and authority in the face of patients’ vulnerability, attachment theory is most obviously relevant when patients see physicians or surgeons in the context of a physical health crisis. We should not expect clinical relationships with other professionals, or in other situations, to arise in the same way. Indeed, there is evidence from interviews with nurses and oncologists that they support patients emotionally in contrasting ways; the oncologists relied on clinical expertise and authority to comfort patients and families, while nurses engaged at a more explicitly emotional level [Reference Forsey, Salmon and Eden86]. Of course, heterogeneity amongst practitioners arises, not just from different professional roles, but from their own attachment styles and the experiences that shaped them. For instance, there are reports that medical students with ‘secure’ attachment style achieved higher scores in assessments of clinical communication [Reference Fletcher, McCallum and Peters87] than did other students, and were more likely to choose careers in primary care than in hospital specialities, perhaps indicating greater comfort with a specialty that emphasizes long-term relationships with patients [Reference Ciechanowski, Russo and Katon88]. There is evidence, too, that doctors’ attachment style predicts aspects of their responses to patients’ presentation [Reference Salmon, Wissow and Carroll89–Reference Cherry, Fletcher and O’Sullivan91], and even patients’ presentation itself [Reference Cherry, Fletcher and Berridge92]. For example, a doctor who has learned to be wary of emotional closeness might seek distance from patients who show dependence. However, the demands of professionalism mean that patients should not be at the mercy of practitioners’ different attachment styles. Instead, practitioners have the responsibility to appreciate how their attachment style influences their clinical relationships and, as part of learning their professional role, to become able to transcend this style when necessary. Of course, practitioners vary in ways that are not captured by formal assessments of attachment style, and patients can value idiosyncratic aspects of individual doctors’ presentation [Reference Wright, Holcombe and Salmon73]. After all, to offer a secure base a practitioner needs to be a committed individual, not just acting a role.

Attachment theory is, like any other psychological theory, an imperfect lens through which to see what might otherwise be obscure, and it is important not to use it inappropriately. While it might help understand some differences between practitioners in their response to patients, or how they cope with the challenges of clinical care [Reference Adshead93], it cannot explain why they care [Reference Salmon and Young49]. For instance, to claim that, because practitioners enjoy relationships with patients and can feel upset when patients suffer or die, they have attachments to their patients [Reference Granek, Shapira and Constantini94] is implausible. For a patient, the practitioner is non-substitutable as an attachment figure, but practitioners must be ready to move on to the next patient without the burden of grief-like responses when one patient suffers or dies. As a paediatric oncologist explained in a qualitative study, ‘We are sad at the time [of giving bad news of treatment resistance] and then we have to move on because we have other patients waiting for us’ [Reference Terrasson, Seigneur and Rault95]. Extending attachment theory to explain practitioners’ caring risks recycling the implausible assumption of the psychotherapeutic model: that clinical relationships are a symmetrical emotional connection between practitioner and patient.

Attachment theory is, of course, not the only theoretical lens available to view clinical relationships. From psychoanalytic theory, the concepts of ‘transference’ and ‘counter-transference’ have been used to describe how patients and practitioners, respectively, can transfer feelings or attitudes from previous relationships onto a current, clinical one. Similarly, ‘projective identification’ describes how one party to the relationship can take on emotional feelings, such as helplessness and hopelessness, emanating from the other. Kelly et al. [Reference Kelly, Varghese and Pelusi96] illustrated how these concepts can help to understand doctors’ clinical decisions in end of life care. A practically oriented application of psychoanalytic theory arose in the 1950s. Named after the psychoanalyst who first developed them for GPs, ‘Balint groups’ were small groups of doctors, facilitated by a leader but drawing on peer-supervision to help analyse and understand challenging clinical relationships [Reference Balint97]. Balint argued that the clinical relationship could even have a therapeutic function [Reference Balint98]. These groups continue, albeit outside the mainstream of clinical communication practice and research [Reference Sally99], and without clear evidence about outcomes [Reference Van Roy, Vanheule and Inslegers100]. Arguably, the language and theory of attachment will be more accessible than psychoanalytic theory to practitioners who are not psychological specialists.

3.3 Conclusion

This section, on the science of clinical relationships, has converged with the conclusion of the previous one, about the ethics of relationships. Each began with an idealized view that excluded the inherent asymmetry of expertise and vulnerability that defines clinical relationships. And each has concluded with a way of understanding clinical relationships that is built on that asymmetry, and that sees practitioners’ authority as the basis of the relationship. Both science and ethics therefore point to a serious dilemma: patients remain inherently vulnerable, regardless of the quality of practitioners’ communication. It is well known that vulnerable children or adults can be abused or exploited by powerful individuals whom they trust. Similarly, patients’ need to trust a ‘secure base’ leaves them open to abuse and exploitation. They can even become impassioned supporters of ostensibly authoritative practitioners whose clinical peers denounce them as fraudulent and exploitative [Reference Passalacqua, Campione and Caminiti101]. In the extreme case, practitioners whom patients trust as strong attachment figures can even be murderous. Patients described one GP in the UK as ‘going out of his way’ to look after them, but he murdered hundreds [Reference Soothill and Wilson102]. Because vulnerability and dependence are inherent in clinical relationships, ensuring that patients are protected from care that is paternalistic, or even exploitative or abusive, cannot be delegated to patients themselves by a deontological emphasis on practitioners’ performance of communication behaviours such as providing information and offering patients choices about treatment decisions; it remains the practitioners’ responsibility. This Section therefore points, once again, to the need for communication teachers and researchers to engage with practitioners’ character, including their knowledge, values and motives [Reference Stiefel and Bourquin6].

4.1 Patient Empowerment

While recognition of the persisting paternalism of medical care motivated bioethicists’ interest in patient autonomy, it also drew social scientists’ attention to patients’ role in healthcare. The broader context was the growing social and political concern with empowering marginalized groups in society. Here we examine related technologies that arose out of concern with patients’ empowerment in healthcare and with ensuring their involvement in healthcare decisions.

4.1.2 Shared Decision-Making

Also drawing on ethical ideas around autonomy, the technology of ‘shared decision-making’ (SDM) has provided more explicit guidance about how practitioners could contribute to patients’ empowerment by involving them in decisions about their own care. Although the principle of sharing decisions with patients had been advocated for decades, the technology of SDM began to take shape in the 1990s [Reference Charles, Gafni and Whelan129,Reference Charles, Gafni and Whelan130]. Its aims were to respect patients’ autonomy (understood from the ethical perspective of self-determination), and to ensure that decisions about treatment are consistent with each patient’s needs and values. SDM recognized that, for example, some patients might give a higher priority to certain side effects of a treatment than other patients – or doctors – would or, conversely, that some patients might be ready to tolerate treatments that others would consider too onerous. The aims of SDM could be delivered, it was argued, by following some general principles. First, patients and practitioners should share information – the patients about their symptoms, needs and priorities, doctors about treatments and their risks and benefits. Second, they should both participate in decision-making. Third, they should agree a final decision. Arguing that these principles were vague and open to variable interpretations in practice (for instance, around what exactly ‘participate’ means, or ‘sharing’ information), Elwyn et al. proposed more practical recommendations [Reference Elwyn, Frosch and Thomson131,Reference Elwyn and Miron-Shatz132]. First the patient should know that a choice exists, and that therefore a decision is needed. Second, he or she should know what the options are and what their effects are. Third, the patient should deliberate – that is, should think about the options and weigh them up. Turning to the practitioner, Elwyn therefore described several kinds of talk that are needed: explanation that a choice exists; description of the options; and support to deliberate and agree a decision [Reference Elwyn, Lloyd and May133].

SDM has been widely endorsed by healthcare organizations. Nevertheless, the objective features of shared decision-making remain elusive in routine practice [Reference Couët, Desroches and Robitaille134,Reference Gerwing and Gulbrandsen135]. Its failure to penetrate practice has prompted re-examination of the recommendations for SDM specifically, and of the concept of patient involvement more generally, identifying several difficulties [Reference Gerwing and Gulbrandsen135–Reference Thomas, Bass and Siminoff141]. First, to follow the guidelines can be laborious for practitioners, who might simply have insufficient time, and who can have multiple decisions to make in a single consultation [Reference Eirik Hugaas, Jan and Edvin142]. It can be burdensome for patients, too, who often do not want to deliberate about treatment options or feel unqualified to. While valuing being recognized by practitioners and healthcare institutions as individuals used to making decisions for themselves, the main decision patients make in a clinical context can be about trusting their clinical team rather than choosing treatment. Similarly, they can look to their practitioners’ personal judgement and experience as information resources more valuable than the objective information emphasized in SDM [Reference Sinding, Hudak and Wiernikowski125]. In this way, patients can feel involved, but having ‘reinterpreted’ involvement to mean following expert guidance [Reference Peerbhoy, Hall and Parker143]. Moreover, although Elwyn et al. specified their recommendations to achieve SDM in some detail, they remain imprecise. For instance, it remains unclear exactly what ‘deliberation’ means; it might even include thinking irrationally or inaccurately.

Accounts of SDM also lack recognition that it does not remove the reality that clinical relationships are asymmetric; practitioners inevitably usually have more knowledge and experience than they can share with patients, and patients’ vulnerability can leave them emotionally ill-equipped to make difficult decisions, to the extent of needing to trust experts to do so. Therefore, as we shall see in Section 6, patients can gain a sense of partnership with doctors from receiving information, even when they take no part in decisions (Box 4). Even where there are options, it is still the practitioner who constrains choice by deciding which ones to offer. In many situations, moreover, no real choice exists – there is only one clinically justified option – so SDM procedures would have little value. Moreover, the emphasis on agreeing the final decision does not address situations where patients need to be persuaded; for instance, where they seek treatment that will be ineffective or harmful, or where they reject potentially life-saving treatment out of concern about relatively minor side effects.

Box 4 Information supports partnership

This set of interview extracts on Healthtalk illustrates patients’ varied expectations and experiences of decision-making (https://healthtalk.org/experiences/shared-decision-making/what-shared-decision-making/). The sixth patient describes having a strong sense of ‘partnership’ with his cardiologist. But this is based, not on having shared decision-making with her, but on the cardiologist taking responsibility for decisions while explaining each one to the patient. Other patients explain the importance of doctors knowing what matters to the patient and of respecting patients’ own expertise, particularly in managing long-term conditions where expertise arises from the patients’ experience of self-management.

Finally, SDM shares the limitations of models of decision-making that were influenced by the assumption that people make decisions rationally and that there is therefore usually an objectively ‘good’ decision to be deduced [Reference Thomas, Bass and Siminoff141]. In reality, people routinely make important life decisions by using ‘heuristics’ which ‘short-cut’ decision-making [Reference Brown and Salmon137]. In particular, they often make choices by following the recommendations of someone whose expertise they trust, or they decide emotionally according to how different options ‘feel’. Decisions that depend on these ‘gut feelings’ and other heuristics can be efficient and accurate, perhaps partly because the decision-maker avoids information overload [Reference Gigerenzer and Gaissmaier144]. Nevertheless, people’s habitual use of heuristic decision-making presents problems for SDM. First, patients are liable to be unprepared to deliberate about options in a novel, rational way [Reference Sinding, Hudak and Wiernikowski125]. Second, identifying what is an objectively ‘good’ decision becomes elusive because non-rational, heuristically based decisions are apt to be idiosyncratic and contextualized [Reference Brown and Salmon137].

The development of rating scales to measure the extent to which SDM is present in consultations fuelled research which identified another difficulty with the concept. High inter-rater reliability can be achieved between observers trained to use the same instrument [Reference Keshtgar, Cunningham and Jones145]. However, scores from different instruments are sometimes unrelated [Reference Scholl, Kriston and Dirmaier146], and, when patients and practitioners rate their consultations for SDM, their scores do not reliably correlate with observers’ ratings, or with each other’s [Reference Keshtgar, Cunningham and Jones145]. It seems that SDM can mean different things to different people [Reference Saba, Wong and Schillinger147] (Box 5). Given these difficulties with the concept, it is unsurprising that reviews of patient outcomes when practitioners have been trained to follow SDM guidance have not, in general, provided strong evidence of tangible benefits for patients [Reference Légaré, Adekpedjou and Stacey148–Reference Bruch, Khazen and Mahmic-Kaknjo152].

Box 5 Objective and subjective SDM diverge

Saba et al. [Reference Saba, Wong and Schillinger147] (paper available at https://www.annfammed.org/content/4/1/54.short) categorized primary care consultations in a US clinic according to whether SDM was (i) present objectively in the doctor–patient dialogue and (ii) present in their subjective experience. The table shows that these two ways of classifying consultations did not correlate. Text shows some of the reasons for divergence.

While SDM has not yet appreciably changed the patterns of dialogue in routine consultations, a very tangible expression of SDM is the burgeoning availability of ‘decision aids’ [Reference Elwyn, Frosch and Volandes153]. These are instruments – on paper or on-line – designed to involve patients in care decisions by providing information about treatment options in ways that patients can understand and make use of, and by structuring patients’ deliberation about the available choices. They differ greatly in format and complexity (Box 6). Some use pictograms to depict statistical concepts liable to confuse patients, for example, the difference between relative and absolute risk reduction when taking a statin drug to reduce the risk of a serious cardiovascular event. Others rely on verbal, rather than pictorial presentations. Patients who use them are better informed, have more accurate understanding of risk, feel less conflicted about their eventual decision and take a more active part in decisions [Reference Leinweber, Columbo and Kang154,Reference Stacey, Lewis and Smith155]. Although it is not clear that, in general, decision aids change the decision, their use in older people might help avoid invasive life-sustaining procedures in favour of comfort care [Reference Gans, van Mun and de Groot156]. Nevertheless, like SDM dialogue in consultations, decision aids remain rare in routine practice. One practical difficulty is that, in the effort to encompass more and more of the information potentially relevant to a decision, they become longer, more complex and less easily readable, putting them beyond the reach of many patients – thereby repeating the trend seen in patient consent forms [Reference Eltorai, Naqvi and Ghanian157,Reference Duong, Mandrekar and Winham158]. That some healthcare organizations or professional bodies mandate or encourage the use of decision aids to protect practitioners from accusations that they did not adequately inform patients about a procedure [Reference Matlock, Fukunaga and Tan159,Reference Joseph-Williams, Elwyn and Edwards160] points to the danger that we saw with patient activation; a technology inspired by the need to empower patients can be co-opted to serve practitioners’ or institutions’ interests.

Box 6 Patient decision aids

Developers of decision aids (DAs) must confront the tension between making them easier to use (shorter, simpler language, fewer statistics) and accurately conveying the necessary information.

At low levels of complexity, DAs simply portray risks and benefits of treatment in ways more accessible than verbal statements of probability. For instance, this brief DA (https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/976877/CovidStats_07-04-21-final.pdf) used pictograms to show benefits and risks of vaccination against COVID-19 for people of different ages. This DA (https://statindecisionaid.mayoclinic.org) uses similar pictograms to explain the risks and benefits of a statin drug.

During this webinar (minutes 27–38) https://www.youtube.com/watch?v=agjce7TEDJc the developer explains a DA for people offered lung cancer screening. It aims to be comprehensive, but in trying to encompass all relevant information it has become longer and more complex.

This site https://decisionaid.ohri.ca/Azsumm.php?ID=1851 links to DAs that take different communication approaches, some going beyond simple pictorial depiction of risk probabilities. One (on genetic testing for Alzheimer’s disease) tries to help undecided users by using ‘vignettes’ of patients making the decision.

In the face of these practical and conceptual problems with SDM, there have been attempts to reformulate the concept to better fit the reality of clinical relationships [Reference Beach and Sugarman136,Reference Brown and Salmon137,Reference Landry139,Reference Entwistle, Carter and Trevena161]. These reformulations recognize the asymmetry of expertise between practitioners and patients and accept that, in many cases, there is no real choice – just one clinically appropriate option. Nevertheless, they try to achieve the aims of SDM: to respect patients’ autonomy and to ensure that clinical decisions are consistent with patients’ needs and values [Reference Beach and Sugarman136]. Entwistle et al. [Reference Entwistle, Carter and Trevena161] suggested a procedure to help people decide about health screening decisions, such as when women are offered breast mammography or smokers are offered lung cancer screening. They sought a middle way between paternalism, whereby experts simply tell people to accept screening, and consumerism whereby individuals make their choice based on acquiring sufficient information for themselves. The starting point for their approach is a relationship with a trusted practitioner, such as a GP, who provides a recommendation to accept a specific screening offer and explains which experts have recommended it – for example, a professional body or government-appointed scientists. The practitioner then describes why those experts made the recommendation, and why they can be trusted, before explaining how anyone who wants more information can access it. Entwistle et al.’s approach recognizes that most people want to trust experts about such a decision and that, for these people, the responsibility for ensuring that the decision to be screened is the right one must rest with those experts and cannot be devolved to the patient. It also recognizes that some people will want to be more informed.

A second proposal addressed major treatment decisions such as in cancer care [Reference Brown and Salmon137]. Again, it emphasizes that good decisions require a clinical relationship with a trusted practitioner, who should evaluate the patient’s ability and motivation to make what the practitioner would regard as a ‘reasonable’ decision for that patient. For example, if life-saving intervention is available with minimal risk, then the reasonable decision would normally be to accept it. If the patient is a Jehovah’s Witness, however, and the intervention is a blood transfusion, the reasonable decision might be to reject it because this religion eschews transfusion. That is, the reasonableness of a decision must be judged from the patient’s perspective, not the practitioner’s. Practitioners should therefore assess and engage with patients’ decision-making by exposing and, where necessary, challenging the propositions that guide their heuristic reasoning. They should then take the responsibility to lead decision-making, from gentle prompts through to persuasion, to the extent that patients are unable or unwilling to reach ‘reasonable’ decisions.

Both these approaches clearly rely on practitioners’ sensitive judgement. They cannot be implemented simply by specifying behavioural rules for them to follow. Similarly, we saw that previous formulations of SDM also defied attempts to specify objectively agreed and measurable procedures. Instead, the reformulations emphasize the essential values of SDM – respecting patients’ autonomy and ensuring that decisions serve patients’ own priorities and values [Reference Galasiński, Ziółkowska and Elwyn162]. This, in turn, points to the qualities that practitioners need to develop if they are to be ready to detect and respect patients’ priorities and values, including humility, curiosity and flexibility; as Piertese et al. observed, ‘how can a doctor … explore “patient preferences” without being curious about that specific person and his/her life’ [Reference Pieterse, Gulbrandsen and Ofstad163].

4.2 Patient-Centred Care

There have been calls to centre care on patients rather than their diseases since the beginnings of recorded medicine. However, systematic use of the term ‘patient-centred care’ or, more recently, ‘person-centred care’ [Reference Pilnick13], dates from the middle of the last century; Balint, a psychoanalyst, advocated ‘patient-centred care’ in urging GPs to concern themselves with the ‘whole patient’ [Reference Balint97,Reference Balint98] (Section 3). Current popularity of the term owes much to the work of a Canadian group of primary care researchers in the 1980s and 1990s, and to their advocacy of a ‘Patient-Centred Clinical Method’ [Reference Levenstein, McCracken and McWhinney164–Reference Stewart, Brown and Weston166]. Essentially, this guides practitioners to explore and be concerned with patients’ experience of illness, and to collaborate with patients in creating a shared understanding of the illness. In inviting patients’ own perspectives into the consultation, practitioners can learn from their patients and can arrive at options for care that reflect each patient’s needs and values. In its emphasis on bringing patients’ emotional experience and psycho-social world into the consultation, this guidance clearly follows the lead of Engel’s earlier description of the ’biopsychosocial model’ [Reference Engel48] (Section 3), and overlaps with the aims and practices of SDM. However, over recent years, both empirical and theoretical difficulties with the concept have become clear. In linking the clinical relationship to explicitly psychosocial talk, patient-centred care has less to say about the role of practitioners’ expertise, authority and conscientious clinical care in sustaining clinical relationships. Moreover, continued refinement and reinterpretation has resulted in many different definitions of patient-centredness [Reference Pilnick13,Reference Mead and Bower167]. For instance, one influential UK healthcare ‘think tank’ defined it as care that is coordinated, tailored to the individual, underpinned by dignity, compassion and respect, and that enables a fulfilling life for the patient [168]. Essentially, this is a prescription for good quality care, and the term ‘patient-centred’ is in practice now widely used to mean ‘good’ care.

Empirically, although training practitioners to communicate in ways that characterize patient-centred care can change their communication, there have not been clear benefits for patients’ clinical outcomes [Reference Dwamena, Holmes-Rovner and Gaulden169]. Indeed, one evaluation of training became famous in the field as a warning that communication that looks patient-centred might not always be in patients’ best interests [Reference Kinmonth, Woodcock and Griffin3]. GPs and specialist nurses were trained in patient-centred communication, focusing specifically on care of patients with diabetes. Patients seen by trained practitioners felt more satisfied with practitioners’ communication and treatment than did those seen by an untrained control group. Despite this evidence that training made consultations more patient-centred, two clinical markers (blood triglycerides and body mass index) were poorer in patients seen by trained practitioners, as was patients’ knowledge about diabetes. These findings point to the danger that emphasizing the importance of a clinical relationship based on engaging explicitly with patients’ emotional experience might constrain doctors’ clinical care [Reference Chew-Graham, May and Roland170,Reference Miller and Fritz171]. Moreover, there is a risk that learners reject the concept of patient-centredness where it is taught as behavioural skills without provision for them to understand the concept and to relate it to real experiences of care [Reference Bansal, Greenley and Mitchell172].

Once again, we see that a technology with moral aims cannot readily be sustained at the level of behavioural rules that practitioners can be taught to follow, because those rules turn out not to be inherently and consistently in patients’ best interests. However, the literature and guidance that has grown around the concept of patient-centred care is important in helping practitioners to define their role more holistically than a focus on diseases decontextualized from patients’ lives. That is, the value of this technology lies most clearly in the domain of values. Being truly patient-centred means having the attitudes and beliefs that define patient-centeredness: respecting and valuing patients as unique individuals. As Duggan et al. suggested, being genuinely patient-centred is ‘just the right thing to do’ [Reference Duggan, Geller and Cooper2].

4.3 Conclusion

Each of the technologies in this section is a morally grounded attempt to deliver aspects of the biopsychosocial model: to ensure that practitioners respect patients as autonomous individuals with needs that extend beyond the biology of their disease to encompass its psychosocial context. Each technology offers practical solutions at the level of behaviours that practitioners can adopt: they can use specific consultation strategies to involve patients or to elicit and respond to their psychosocial needs, and they can use questionnaires to audit patients’ empowerment. However, each technology does not clearly work at this deontological level; practitioners who are taught to use those consultation strategies do not always help their patients by doing so, and the questionnaires do not always detect what matters to patients. Instead, our analysis points again in the direction of virtue ethics: that involving patients and meeting their psychosocial needs depends on practitioners who respect their patients as autonomous individuals and who understand and respect their needs.

5.1 Assessing Communication and Communication Skills

Curricula often include some classroom teaching, but the emphasis is on supervised role-play and expert and peer feedback, often with actors playing the part of patients. To assess outcomes of CST the usual method is for observers to score learners’ display of taught skills with real patients or actors playing patients. Many clinical communication researchers use the same approach to study communication skills in routine consultations. Although several observational coding schemes are available, they share a similar, reductionist approach in breaking down streams of communication into elements that can be quantified. The pioneering scheme was Bales’ ‘Interaction Process Analysis’ [Reference Bales193,Reference Bales194] reported in 1950 and applied in clinical and other settings. To apply his method, an observer identified segments of speech or non-verbal behaviour that could be allocated to one of twelve broad categories; six concerned task-oriented communication (such as seeking suggestions or giving opinions) and six assessed social-emotional communication (such as showing ‘solidarity’ or ‘antagonism’). Crucially, the observer’s task was interpretive: that is, to impute the meaning, or function, of each segment of dialogue in the context of the whole interaction and of the observer’s knowledge of that type of interaction. Subsequent clinical communication coding schemes have mostly followed only one of these two defining features of Bales’ system. Reflecting the emphasis on overt emotional engagement with patients that we have already seen in communication literature, they continue to distinguish clinical, or instrumental, communication from communication that is emotional in content and overtly oriented to the relationship (e.g. [Reference Zimmermann, Del Piccolo and Bensing195-Reference Heaven, Clegg and Maguire199]). The second feature of Bales’ scheme, interpretive coding, has been rare in subsequent coding schemes. Instead, most have been designed to be as objective as possible in distinguishing different elements of communication according to what can be directly observed; that is, according to their form rather than imputed function. Nevertheless, accounts of these coding schemes then typically conflate form and function by equating emotional talk with emotional function and regarding communication that is task-oriented as having little relational value [Reference Roter and Larson197]. The pursuit of objectivity in measurement perhaps reflects the continued attraction of the natural and biological sciences as models for communication research. Sometimes this lure is explicit; for instance, when the radiological term ‘thin-slice’ is used to refer to coding of brief segments of clinical dialogue [Reference Roter, Hall and Blanch-Hartigan200].

5.2 Critiquing the Concept of Communication Skills

Beginning in the 1980s, when the view of communication as made up of objective and teachable skills was taking shape, writers have questioned the validity of an educational model based on reducing communication to objectively definable elements that can be taught by communication experts [Reference Gulbrandsen, Gerwing and Landmark7,Reference Skelton33,Reference Langewitz176,Reference Stiefel and Bourquin191,Reference Plum201–Reference Salmon and Young206]. First, communication behaviours do not, in practice, have universal meaning that can be objectively defined. For instance, the same behaviour from a practitioner can be perceived as caring or uncaring by different people or in different situations [Reference Quirk, Mazor and Haley207]. Behaviour that objectively provides only instrumental support can be experienced emotionally [Reference Brandes, van der Goot and Smit58,Reference Semmer, Elfering and Jacobshagen59] and, when patients talk about symptoms or treatment, they can mean many different things depending on the relational and clinical context [Reference Bourquin and Stiefel208,Reference Salvade, Stiefel and Bourquin209]. The reason is that communication is inherently subjective and contextual; what a listener hears depends, not only on what the speaker has said, but on the subjective and social context, including the interaction to that point. This problem is typically hidden in plain sight by communication curricula or coding schemes; when behaviours, such as eye contact, are listed as skills to be taught or counted they are usually qualified by words like ‘appropriate’, thereby implicitly acknowledging their contextual dependence [Reference Skelton33]. But gauging the ‘appropriate’ moment to make or avert eye contact, to speak or be silent or to touch a patient depends on sensing the ‘atmosphere’ of a consultation that defies reduction to individual elements and which the practitioner is simultaneously both sensing and shaping [Reference Langewitz176,Reference Boudreau, Wykretowicz and Kinsella210]. Moreover, many listed ‘skills’ are global aspects of relationships, such as ‘empathy’ or ‘partnership’, which are clearly not objectively definable behaviours. Other aspects of communication that defy an attempt at behavioural description are simply absent from curricula or coding schemes. For instance, we shall see in Section 6 that qualitative researchers who examine clinical communication in cancer care inductively have identified potentially valuable communication practices that are missing from published lists of skills. One influential framework for communication teaching and assessment does therefore emphasize the attainment of communication goals (SEGUE: Set the stage, Elicit information, Give information, Understand the patient’s perspective, and End the encounter), leaving learners to decide how to achieve each [Reference Makoul211]. This approach arguably shifts reductionism to another level; the function of the whole consultation is understood as the aggregate of these predefined component functions. However, goals in consultation can be complex and idiosyncratic, changing rapidly from moment to moment [Reference Hulsman212,Reference Veldhuijzen, Mogendorff and Ram213].

An important consequence follows from recognizing the invalidity of the assumption of objectivity in communication. Communication skills research, and the evaluation of CST, rely heavily on quantitative research designs; correlational studies link the quantity of specific types of communication to patient outcomes, while randomized trials test effects of training on practitioners’ performance of skills. However, because the same behaviour in different contexts can have different meanings, designs that aggregate measurements of communication behaviours across consultations or patients can be of only limited help to practitioners. For instance, many studies quantify practitioners’ empathic behaviour, correlating it with patient outcomes like satisfaction, or showing that CST increases empathy. But the value of empathic behaviour depends, not on its quantity, but on its nature and timing, so counts of empathy tell us little. Similarly, a single well-chosen utterance can change the mood of a whole consultation – the ‘lucky punch’ that Langewitz described [Reference Langewitz, Nübling and Weber214]. For several decades, there have been warnings that, because of this inherent variability and subjectivity in communication, the findings that quantitative outcome research can deliver are inevitably limited to bland generalizations [Reference Skelton33,Reference Stiles215]. Consistent with these warnings, reviews of the outcomes of CST have consistently shown little clear evidence that patients benefit – even when practitioners’ behaviour changes in line with what was taught [Reference Moore, Rivera and Bravo-Soto216–Reference Gilligan, Powell and Lynagh220].

An even more fundamental problem with the concept of communication as made up of skills is rarely acknowledged in academic literature, although poignantly illustrated by a BBC news report of the launch of communication training for doctors in the UK NHS.Footnote ² In drafting the headline ‘Doctors to get lessons on being nice’, the journalist pointed with humour to the tension between learning caring behaviours and being authentically caring. That is, while patients and practitioners value authenticity [Reference Salmon, Mendick and Young221] (Section 6), learning to display caring or empathic behaviours is not the same as being authentically caring or empathic. Educationists and ethicists identify the conundrum; predetermining objectives for students’ learning means that, when students deliver these, their performance cannot be regarded as self-determined, but self-determination is necessary to view behaviour as authentic [Reference Kinghorn8,Reference Alexander222]. Some clinical communication research continues to disregard this challenge (e.g. [Reference Hillen, de Haes and Stalpers223]), and communication skills literature has not yet offered a theoretical framework within which learning skills can be reconciled with authenticity (Box 7).

Despite these difficulties with the concept of communication skills, it still shapes many educators’ goals. The educationist, Eisner, echoed Winston Churchill’s statement that ‘we make our buildings and then our buildings make us’ when he warned that ‘we make our curriculum and then our curriculum makes us’ [Reference Eisner224]. That is, behaviours defined as ‘skills’ come to define good communication. After all, the word ‘skill’ itself implies something to be valued intrinsically. Similarly, the availability of communication coding schemes, particularly where there is a technologically sophisticated interface that makes it easy and attractive to use,Footnote ³ shapes and constrains the questions that researchers ask. Therefore, in evaluations of CST, learners’ performance of taught skills is typically presented as sufficient evidence of successful training. Papers that become widely cited as demonstrating the value of CST show increases in the quantity of skills displayed in real or simulated consultations without any evidence from patients’ perspective [Reference Fallowfield, Jenkins and Farewell225,Reference Fallowfield, Jenkins and Farewell226]; increased performance of skills is sufficient for communication to be described as ‘improved’. In this way, researchers and educators, rather than practitioners or patients, define what is valued in clinical communication, even to the extent of berating practitioners’ routine communication and asserting their continued need to be trained to use expert-defined communication skills [Reference Fallowfield and Jenkins227].

This problem matters because, as we shall see (Section 6), patients can value aspects of communication other than the skills that communication experts value and measure [Reference Kuchinad, Park and Han53,Reference Hatem, Mazor and Fischer228,Reference Epstein229]. Therefore, when patients evaluate consultations, their evaluations do not correlate appreciably with those of communication experts based on practitioners’ performance of skills [Reference Burt, Abel and Elmore230,Reference Hulsman, Ros and Winnubst231]. There has long been suspicion that the focus on expert-defined skills might ill-serve patients’ needs. Algedahl et al. [Reference Agledahl, Gulbrandsen and Førde232] studied video-recordings of consultations with hospital doctors in Norway, concluding that the doctors were ‘courteous but not curious’. In other words, they had learned to perform the skills of courteous, polite interaction with patients, but were not authentically interested in them and in what made them, and their healthcare needs, unique. The authors warned that more CST would not solve this problem. We saw in Section 4 that training for diabetes practitioners that focused on patient-centredness skills left patients more satisfied with consultations but with poorer biomarkers [Reference Kinmonth, Woodcock and Griffin3]. Similarly, the only statistically significant outcome of one CST programme for practitioners caring for end-of-life patients was that their patients were more depressed than those cared for by untrained practitioners [Reference Curtis, Back and Ford233]. While unexpected findings like these might be ‘Type 1 errors’, or ‘false positives’, they are warnings that CST might even sometimes have deleterious effects by focusing practitioners on their own performance of skills at the expense of their patients’ needs.

In shaping what educators and researchers value in clinical communication, the communication skills model also risks narrowing their gaze. By attributing failures in communication to lack of skills, it promotes a ‘deficit’ model; that practitioners lack certain skills which they need to be taught. But this stance can blind researchers and educators to other reasons for apparent communication failures. After all, most practitioners probably already have learned much about communication and relationships before they start clinical training, and might have other reasons for not drawing on what they know. For instance, they can decide that certain ‘skills’ are clinically inappropriate or unnecessary in some situations [Reference Essers, van Dulmen and van Weel234], or do not fit their feelings about the patient or their own mood [Reference Cocksedge and May235] or their view of their professional role [Reference Robinson, Priest and Susman236]. Patients can trigger emotional reactions in practitioners that stop them using skills they have or that even lead to collusion with patients to avoid certain issues [Reference Salander and Sandström237–Reference Stiefel, Nakamura and Terui240]. Many common complaints about practitioners’ communication potentially indicate attitudes that signify lack of respect, or arrogance [Reference Entwistle, Cribb and Mitchell43,Reference Beach, Duggan and Cassel44]. Perhaps attributing such behaviours to ‘lack of ‘skills’ helps avoid the more brutal implication that some practitioners do not respect their patients and feel superior to them.

5.3 Breaking Bad News

Breaking bad news repays detailed attention because it has been a major focus of teaching and guidance over decades and illustrates poignantly the main elements of the critique in this Section. Reflecting its importance, and many practitioners’ nervousness about the task, several protocols have been published to guide practitioners. The most influential is known by the acronym SPIKES [Reference Baile, Buckman and Lenzi241], signifying six broad sequential tasks (Setting up, Perception, Invitation, Knowledge, Emotions with Empathy, and Strategy or Summary). It provides detailed instructions for how to achieve each task, and scripted illustrations of the necessary communication skills, including making empathic statements. However, several critics have argued that such protocols are unrealistic [Reference Hatem, Mazor and Fischer228,Reference Langewitz242–Reference Salander245]. They define bad news as objective information and assume that consultations containing it can therefore be predicted in advance and standardized or controlled by the practitioner. However, the subjectivity of bad news means that practitioners cannot always anticipate it. For instance, some patients with persistent and troubling symptoms find being told that tests or investigations found no serious abnormality distressing – the ‘bad news that nothing is wrong’ [Reference Nettleton246]. Moreover, bad news is rarely a single piece of information in the way that the protocols assume. It often arises as one more event in a sequence: doctors tell the patient that they are concerned; they explain that further tests are needed; then one of those tests shows a need for more tests; then perhaps there is a diagnosis; then a proposal for treatment; but then perhaps the treatment cannot start for a few days or weeks; then there are blood tests or a scan that show an initial effect of treatment; then there are further results, with implications for changing treatment and so on. Each point in the trajectory of information can reveal what feels like bad news to the patient; so bad news can be a continuous and subjective process rather than a single objective event. Therefore, Eggly [Reference Eggly, Penner and Albrecht243] argued that patients’ interpretation of information as bad news should be seen, not as an objective fact instigating consultation, but as a subjective inference emerging from it, and from the relationship within which the consultation occurs. Moreover, Langewitz [Reference Langewitz242] warned that bad news can confront both practitioners and patients with ‘existential questions that push both persons to the limits of what they can bear’. If consultations about bad news are not so amenable to practitioners’ control as protocols suppose, it is unrealistic to assume that they can be programmed to unfold in a consistent way.

Protocols for breaking bad news also illustrate the danger of scripting practitioners’ emotional support. Illustrations in SPIKES of empathic statements include explicitly referring to the news as ‘bad’, and citing the practitioner’s own emotions, for example in being ‘sorry’ to have to give the news, or finding ‘it very difficult for me also’ [Reference Baile, Buckman and Lenzi241]. We shall see in Section 6 that these are strategies that some cancer doctors have told researchers that they would never use [Reference Mendick, Young and Holcombe247,Reference Mendick, Young and Holcombe248]. Indeed, Langewitz [Reference Langewitz242] suggested that a list of what not to say might be more helpful than attempts to specify what should be said. Moreover, the focus on emotional talk and empathic statements could be at the expense of clinical information that, as we saw in Section 3, patients might find more comforting [Reference Salander244]. Box 8 points to a consultation acted according to guidance contained in protocols such as SPIKES, prompting the reader to question whether such a scripted approach is necessarily what patients need. In contrast, Box 9 points to an account of a patient who gained an immediate sense of having a ‘secure base’ from the four words with which her surgeon introduced the news of a cancer diagnosis.

Despite these difficulties, once protocols such as SPIKES become widely adopted, the skills and practices that they advocate become ‘taken-for-granted’, illustrating how expert-defined communication skills become the ‘gold standard’ for communication. Therefore, there remains very little evidence as to whether patients benefit when practitioners follow such protocols because, as with communication skills training generally, researchers have been content to show that, unsurprisingly, training practitioners to follow a protocol increases their adherence to the protocol [Reference Mahendiran, Yeung and Rossi249,Reference Paul, Clinton-McHarg and Sanson-Fisher250]. Frameworks such as SPIKES might be helpful to new or uncertain practitioners. However, their ‘gold standard’ status is hard to justify.

5.4 Methodological Shortcomings of Quantitative Communication Skills Research

It seems that training in communication skills in general, and breaking bad news specifically, continues on a path that it is not well grounded empirically or theoretically, a phenomenon that would be recognized as a serious failure of evidence-based medicine in other clinical specialties. The reason may lie in the way in which ethical and scientific perspectives are confounded in this field [Reference Salmon and Young205]. Communication skill educators typically appeal to scientific justification of effectiveness when they refer to evidence that training ‘improves’ skills, and the appearance that CST is scientifically justified has perhaps deterred ethical challenge. Conversely, the belief that CST is inherently important and morally unassailable might explain why its scientific scrutiny has not reached the levels of rigour that are now standard across clinical medicine [Reference Salmon and Young251].

For instance, evaluations of CST typically fail to specify the primary outcome variable, instead testing several outcomes. One seminal study included eight outcome measures (counts of different communication skills, each analysed in multiple ways). Although only half showed effects significant at p<.05, the authors concluded that training was successful [Reference Fallowfield, Jenkins and Farewell225] (Moreover, their later conclusion that improvements were maintained at one-year follow-up was based, incorrectly, on not rejecting the null hypothesis of no deterioration rather than on rejecting a null hypothesis of deterioration [Reference Fallowfield, Jenkins and Farewell226].) In a more recent evaluation that the authors entitled ‘rigorous’, the main research question was assessed with over 100 separate outcomes of which, again, only around half showed effects favouring training [Reference Bylund, Banerjee and Bialer252]. In areas of clinical medicine that take evaluation more seriously, primary outcomes are routinely specified in advance in published or registered protocols. This protects against the risk that authors ‘fish’ for positive results amongst a large pool of variables, and it avoids the danger that, without statistical compensation for the increased risk of Type 1 errors (false positives) that arises from multiple testing, reported positive findings are random variation. Multiple outcomes present another problem. It is rarely stated how many, or which ones, should be significant for the training to be regarded as successful. In the studies cited previously, the authors clearly regarded around half as sufficient. But should half be sufficient? And does it matter that, for example, ‘empathy’ improved but ‘checking understanding’ did not [Reference Fallowfield, Jenkins and Farewell225]? Would the opposite result have been just as good (or poor)? The problem here is the absence of a theoretical rationale that allows evaluators to make specific predictions. Similarly, most evaluations report ‘improvements’ in outcome measures. But few emulate evaluations of interventions in clinical medicine by specifying in advance the level of change that would be clinically significant (for a notable exception, see [Reference Girgis, Cockburn and Butow253]). In a recent study which included patients’ ratings of satisfaction with aspects of doctors’ communication as outcomes, baseline satisfaction already exceeded 4.5 on a 5-point scale for many ratings, so was probably adequate at the start [Reference Bylund, Banerjee and Bialer252]. Baseline measures of skills are hard to interpret because communication skills literature provides few benchmarks for adequate levels. Therefore, instead of specifying clinically significant changes in advance, communication skills trialists typically aim simply to increase skill performance from the starting level. This implicitly assumes that ‘the more the better’. Taken to the limit, the target consultation style would therefore be one filled with talk that displays the measured skills, to the potential exclusion of important clinical tasks. Similarly, in several trials, learners’ confidence in their ability to perform learned skills is an outcome measure, e.g. [Reference Bylund, Banerjee and Bialer252,Reference Pehrson, Banerjee and Manna254]; the implicit theory here is that total confidence is better than a level that might leave learners inclined to reflect on their performance (Section 7).

There is another serious problem with quantitative communication research. Its purpose is typically to generalize to a wider population from the sample being studied. For example, the famous finding, from a sample of seventy-four consultations in a US university primary care clinic, that doctors interrupted 69 per cent of patients’ opening statements and that the average time of interruption was 18s after patients started speaking [Reference Beckman and Frankel255] is widely cited as showing that ‘doctors tend to interrupt patients’ opening statements’, or that ‘doctors interrupt patients after an average of 18s’. However, generalization relies on sampling principles that are hard to achieve in clinical communication. Researchers can rarely sample randomly from the population to which they hope to generalize; as in this instance, samples usually arise from a specific country, city and clinic. Moreover, clinical communication is sensitive to cultural changes so generalizability over time cannot be assumed, whereas findings typically are published a year or more after they are made and are cited after even longer intervals. This is not to argue that quantitative methods have no place in clinical communication research. Rather, we need to be very clear about what they can and cannot achieve, a question which we address in Section 7.

5.5 Conclusion

Assumptions and practices that have become ‘taken-for-granted’ in CST and its evaluation reflect advocates’ commitment to helping practitioners communicate holistically with patients; but they also reflect a climate which has protected those assumptions and practices from robust scientific challenge. When examined critically, the view that good communication depends on deploying skills that are objectively definable – and teachable – is hard to sustain. Once again, we see that the deontological approach that locates quality of communication in reproducible behaviours cannot alone ensure quality. Neither can a consequentialist perspective offer appreciable support for the value of teaching the structured protocols or scripted skills that we have examined in this section. In Section 7, we return to the topic of CST from the perspective that quality of communication depends, not so much on the ability to reproduce skills as on good judgement about when and how to use them. That is, our analysis points again to a virtue ethics perspective; good communication relies on practitioners having the knowledge and motives that allow them to adapt communication flexibly to the needs of individual patients at specific moments in the consultation [Reference Stiefel and Bourquin6].

6.1 The Emotional Challenge of Cancer

The experience of cancer exemplifies well the biopsychosocial model in extending far beyond the physical dimensions of the disease. Patients with cancer, like those with other physical diseases, are more likely to be depressed or anxious than are healthy members of the normal population. Prevalence figures depend on the assessment procedure and population studied but, in general, ‘clinical’ levels of depression or anxiety are found in around 15–25 per cent of people treated for cancer [Reference Thom, Silbersweig and Boland256–Reference Brandenbarg, Maass and Geerse258]; post-traumatic stress disorder is present in 5–25 per cent [Reference Swartzman, Booth and Munro259]. Fatigue is also common, affecting a third to a half of patients or survivors, with prevalence increasing in more advanced cancers [Reference Al Maqbali, Al Sinani and Al Naamani260,Reference Jones, Olson and Catton261]. While fatigue is often a direct result of treatments, particularly chemotherapy and radiotherapy, it can also have an emotional or motivational component [Reference Jones, Olson and Catton261,Reference Reuter and Harter262].

Physiological mechanisms might link emotional distress to cancer [Reference Lee and Giuliani263,Reference Politynska, Pokorska and Wojtukiewicz264]. However, there are also obvious psychological reasons why cancer should be distressing. There is the sheer burden of illness and treatment: experiencing pain or discomfort, coping with restrictions on life, managing disclosures of illness to friends and family, and learning to navigate a complex healthcare system in which many different professionals and clinics can be involved. Some patients also feel ashamed of cancer or blame themselves for it or for their ‘failure’ to cope with it. Cancer can re-evoke previous life traumas; women with breast cancer who recalled being abused as children were more likely than others to be distressed, to feel ashamed of their disease and to blame themselves for it [Reference Salmon, Hill and Krespi265]. There is also an existential threat: the ‘biographical disruption’ that interrupts the expectations that we have for the future, and that signals the prospect of death [Reference Locock, Ziébland and Collyer266]. Denying the reality or extent of the threat associated with cancer offers some emotional protection, and the tension between denial and the need to learn to accept the illness and its implications can itself be a source of distress [Reference Brennan267].

On this analysis, therefore, distress is a normal part of psychological adjustment to cancer, and we should be wary of medicalizing it as a pathology to be treated. Nevertheless, distress can be damaging. Depression, in particular, is associated with poorer treatment adherence and poorer prognosis [Reference DiMatteo, Lepper and Croghan268–Reference Wang, Wang and Zhong270]. It also seems that patients with cancer or other physical illnesses who are depressed or anxious use more health service resources, and so cost more to look after [Reference Carlson and Bultz271,Reference Gaulin, Simard and Candas272]. Clearly, patients with cancer can have intense emotional needs that practitioners must address along with the cancer itself.

6.2 Attachment in Clinical Relationships in Cancer Care

In health policy and guidance documents and in professional publications we see the assumptions introduced in previous sections about how practitioners should address cancer patients’ emotional needs; in particular, that they should use emotional talk to build clinical relationships and provide emotional support [Reference Stiefel, Barth and Bensing50,Reference Epstein and Street273–Reference Jefford and Tattersall275]. Over recent years, the expectations on practitioners around emotional support have extended to include identifying patients who are distressed at a ‘clinical’ level and therefore need additional help. Distress has even been described as patients’ ‘6^th vital sign’, after temperature, respiration, heart rate, blood pressure and pain [Reference Bultz and Carlson276]. Cancer services are therefore now widely encouraged or mandated to screen patients regularly for depression and anxiety [Reference Ownby277,Reference Smith, Loscalzo and Mayer278], although the validity of this practice has been questioned [Reference Dekker, Graves and Badger279,Reference Salmon, Clark and McGrath280]. Policy and professional guidance in cancer care also illustrates the second big theme that we have seen in the clinical communication field: practitioners should seek to empower patients by giving them information and involving them in choices over treatment decisions [Reference Epstein and Street273,274].

There is a paradox in this guidance; it depicts patients as emotionally vulnerable and needing their practitioners’ support, while being sufficiently robust to assimilate clinical information and make treatment decisions. This apparent contradiction arises from the way that the biopsychosocial model has shaped expectations on practitioners (Section 3). Engaging with patients at an emotional level and empowering them to make treatment decisions were both morally inspired ways to humanize clinical practice and respect patients as individuals. Now, however, as we saw in in previous sections, a more evidence-based approach is possible based on understanding more about the nature of clinical relationships.

The theoretical background for understanding findings to be described in this section is the attachment perspective on clinical relationships introduced in Section 3; patients’ sense of relationship can arise out of attachment needs rather than being built by practitioners’ use of communication skills to engage empathically with patients. A study of newly diagnosed breast cancer patients illustrated how quickly this process can operate in the cancer clinic, without appreciable emotional talk [Reference Beesley, Goodfellow and Holcombe281]. These patients had just seen their breast surgeon for the first time, after several hours in a rapid diagnosis clinic, in a consultation of around 20 minutes that focused on diagnosis and treatment planning. Before leaving the clinic, patients completed a questionnaire that measures the strength of the alliance patients feel they have with their practitioner, and that previously had mostly been used in counselling or psychotherapy settings – typically after many weeks or months of weekly talk about emotions and feelings. Even after such a brief, clinically focused, interaction, these newly diagnosed patients rated their alliance with their surgeon as stronger than in almost all the published reports in which patients had rated their alliance with psychological practitioners.

Of course, such a relationship does not emerge from patients’ attachment needs irrespective of practitioners’ behaviour. For patients to feel they have a ‘secure base’, they need to see practitioners as having the expertise, authority and conscientiousness to provide the security the patient needs [Reference Wright, Holcombe and Salmon73]; even endorsements of practitioners’ expertise by other staff or patients can help [Reference Wright, Holcombe and Salmon73,Reference Salmon, Mendick and Young221]. Because attachment figures are not substitutable, continuity of care by the same practitioner is also important [Reference Milberg, Wahlberg and Jakobsson74]. However, cancer clinics are often organized so that patients see different practitioners each time they attend. There is some evidence that a well-functioning clinic can act as a secure base in its own right, but this depends on effective and timely communication between the component parts; without this, patients can feel lost, with nothing to ‘hold on to’, and lacking confidence in their care [Reference Atherton, Young and Kalakonda282,Reference Lilliehorn, Hamberg and Kero283] (Box 10). Conversely, knowing that a single senior clinician oversees their care and is available to them can be comforting, whether or not patients take advantage of that availability [Reference Salmon, Mendick and Young221,Reference Salander244].

In complex clinical care such as for cancer, patients are inevitably sometimes confronted by confusing situations or apparent failures in communication or in the care system. Often patients or, when patients are children, their parents can compensate in actively constructing their ‘secure base’. Individual practitioners’ abruptness might be attributed, for example, to their being stressed, or perhaps to their knowing what their patients need even better than the patients do [Reference Wright, Holcombe and Salmon73]. In paediatric cancer care, parents could often ‘contain’ problems arising in the functioning of the wider team by attributing them to causes, such as work pressure or the demands of healthcare bureaucracy, that protected their need to see the senior clinician as a secure base [Reference Wright, Holcombe and Salmon73,Reference Davies, Salmon and Young284]. This is, of course, not to justify failures in care and communication on the grounds that patients or family might find a way to ‘contain’ them emotionally.

In constructing their secure base, patients or family have further challenges that require mental ingenuity and effort. Being bounded by the constraints of a professional relationship, clinical relationships necessarily diverge from attachments in everyday life. Practitioners cannot be burdened with emotions of affection or grief that would be natural within family and social relationships; similarly, patients’ entitlement to practitioners’ time and support cannot extend beyond their professional role. In breast cancer patients’ consultations with surgeons, both parties worked to create a relationship that was authentically personal and intimate, while being firmly circumscribed by professional boundaries [Reference Salmon, Mendick and Young221]. Surgeons described keeping their own emotions out of consultations, for example, by avoiding being tearful when they felt sad or never saying they were ‘sorry’ when giving bad news (Section 5). Conversely, they described deliberately trying to make each patient feel special, even when they did not recognize her, for example by engaging in social talk; that is, they admitted to acting a role. Similarly, patients were aware of the reality of being a ‘stranger’ to surgeons who saw ‘hundreds and hundreds’ of patients and who could therefore not know them individually. Nevertheless, patients could feel ‘special’. As one explained, ‘She’s [Surgeon] got to make me feel I’m the only one, although she’s doing it constantly. But when I’m there, I’m the only person.’ For both parties, therefore, knowing that the surgeon was acting a role was compatible with feeling that the relationship was sincere and authentic. Authenticity lay in patients recognizing surgeons’ conscientious execution of their role.

Clinical relationships therefore depend, not just on practitioners’ performance of their role, but on patients’ capacity to perform their own complementary role, including the necessary attachment work of constructing the ‘secure base’ and being able to feel ‘special’ to the practitioner. Some patients will find this easier than others. We saw in Section 3 how the concept of attachment style can help understand heterogeneity in patients’ ability to form trusting clinical relationships and in how they need practitioners to help them. Studies in breast cancer patients have shown one way in which such heterogeneity can arise. Patients who recalled being abused as children felt less supported by their practitioners and by other people in general [Reference Salmon, Holcombe and Clark285,Reference Fagundes, Lindgren and Shapiro286]. In one study, only a quarter of women recalling abuse completely trusted their surgeon, compared to two thirds of those with no history of abuse; abuse had left them with a negative ‘mental model‘ of themselves; they could not feel fully supported because they did not feel worthy of others’ support [Reference Clark, Beesley and Holcombe287].

6.3 The Role of Inductive Research in Clinical Communication

We have seen in previous sections that much of the evidence-base for ‘evidence-based practice’ in clinical communication is weak or tangential. The problem arises because the predominant research paradigm has been quantitative and deductive, whereby researchers’ theory and assumptions frame the questions they ask and the methods they use to answer them, thereby tending to perpetuate the ways of thinking from which those questions and methods arose. For example, when researchers use communication coding schemes to investigate whether empathic statements correlate with patient satisfaction, they are not immediately exposed to the possibility that other aspects of practitioners’ talk are more important than empathic statements. Questionnaire methods share the same problem; they quantify findings about things that a questionnaire asks about, providing few clues to ones that might be more important.

By contrast, inductive research derives insights from detailed observation, constrained as little as possible by researchers’ assumptions. Indeed, good inductive research surprises researchers, changing their views. In clinical communication, inductive research generally uses qualitative methods, and the example in Box 11 illustrates their importance. Careful interviewing that allowed participants to go beyond their initial, normative responses showed that interviewees who first said that they ‘want to know everything’ went on to complain of having been given too much information [Reference Mendick, Young and Holcombe247,Reference Salmon, Hill and Ward288]. Of course, qualitative research lacks generalizability. However, we saw in Section 5 that, given the inherent subjectivity and contextual dependence of communication, generalizability even of quantitative findings is more limited than typically assumed. Recall Kleinman’s warning (Section 1) in relation to medical ethics that general principles are of limited help to practitioners, and that the main work of ensuring ethical practice takes place in practitioners’ judgements about how to interpret and apply those principles in their work [Reference Kleinman10]. Similarly, it is in their routine interactions with patients that practitioners must learn the ‘practical wisdom’ of implementing general moral principles in specific cases [Reference Kinghorn8,Reference Kotzee, Paton and Conroy9]. Therefore, inductive research into the solutions they find is essential so that these can be understood, critiqued and, potentially, learned from; ‘practice-based evidence’ is the precursor to ‘evidence-based practice’ [Reference Salmon and Young289].

Critical context for communication encompasses not just observable features of a consultation and of the clinical setting and broader culture in which it is embedded, but also participants’ subjective world: what each knows, and what each is seeking and experiencing. Therefore, even qualitative research that is confined to observations of communication will be restricted in the inferences that it can support. To understand and evaluate fully any instance of communication, we also need to know the practitioner’s aims and the patient’s experience. At the least, practitioners and patients could be debriefed as part of an evaluation of practitioners’ communication. In more formal research designs, studying three streams of data simultaneously will mean transcending the methodological ‘brands’ that have been built around specific types of qualitative data [Reference Salmon, Mendick and Young221,Reference Salmon and Young290]. For instance, the way in which practitioners and patients influence each other through their use of speech and language has been studied by discourse analysts or conversation analysts, whereas patients’ accounts of their experiences of healthcare have typically lent themselves to thematic or phenomenological approaches. Of course, practical constraints of resources, skills or access often limit researchers to only one type of data or one methodological perspective. However, where the aim is to evaluate communication, researchers need to understand that inferences drawn from only one type of data will be conditional on what is known or unknown from the missing perspectives [Reference Salmon and Young290].

In the rest of this section, we see how qualitative research that has become available over the last two decades, and particularly research that has integrated observer, practitioner and patient perspectives, has identified ways in which practitioners and their patients have implemented principles around providing emotional support and involving patients. Most of this work has concerned doctors, reflecting the continued emphasis of clinical communication research, and attachment theory warns us not to apply these findings unquestioningly to relationships with other groups of practitioners (Section 3).

6.4 How Can Cancer Practitioners Provide Emotional Support?

As we saw in Section 3, the starting point for deductive research into patients’ support needs has been the principle that practitioners should engage with patients at an explicitly emotional level. By contrast, inductive, qualitative research has begun to illuminate the complexity of practitioners’ task in providing support in the context of an asymmetric clinical relationship.

In a detailed study of clinical consultations of parents of children being treated for leukaemia we audio-recorded and analysed dialogue between parents and haematologists, but we also interviewed both parties to the consultation to identify what they wanted – and experienced – from consultations [Reference Young, Ward and Forsey57,Reference Young, Hill and Gravenhorst71,Reference Forsey, Salmon and Eden86,Reference Salmon, Hill and Ward288]. The parents were emotionally needy, devastated by the fear of losing a child. However, the haematologists wanted to keep their distance and remain objective, and not to engage at an emotional level with the parents. Consultations therefore contained little talk about anything other than clinical results and treatment. However, subsequent interviews with the parents showed that those consultations had comforted them emotionally; comfort came from haematologists being conscientious and authoritative in their clinical care and dialogue with parents. This finding was so striking that we asked parents directly whether they would want to talk to their child’s doctor about their fears and feelings. Their response was a clear ‘no’. From an attachment perspective, this is unsurprising; in the context of their acute sense of vulnerability, emotional comfort arose from conscientious clinical care. Moreover, given the emotional turmoil associated with diagnosis of cancer, parents needed practitioners who responded to their distress by remaining rational and unemotional [Reference Young, Hill and Gravenhorst71]. These findings warn us not to regard emotional cues necessarily as pleas for overtly empathic responses, or doctors’ failure to provide such responses as always indicating failure of empathy. Emotion can communicate instrumental needs, and doctors’ feelings of empathy with patients could therefore be manifest in instrumental responses that address those needs [Reference Brandes, van der Goot and Smit58,Reference Bourquin and Stiefel208,Reference Salmon, Mendick and Young221]. It is salutary that, in a sample of cancer patients attending community care who were emotionally distressed, more wanted to speak with a dietician than with a psychologist or social worker [Reference Kendall, Glaze and Oakland291]!

Patients and parents can actively collaborate in constraining dialogue in consultations to clinical matters, preferring not to introduce psychosocial issues into consultations with practitioners responsible for clinical care [Reference Atherton, Young and Kalakonda282]. Where clinical teams include a specialist nurse, patients or family can prefer to direct psychosocial problems to them [Reference Young, Hill and Gravenhorst71,Reference Salmon, Mendick and Young221]. Indeed, many cancer teams make this arrangement explicit. Consistent with this division of responsibilities, there is some evidence that nurses and oncologists see their supportive roles as needing very different communication strategies; oncologists emphasize talk about clinical care while nurses describe relying on overtly psychosocial engagement [Reference Forsey, Salmon and Eden86].

Collaboration on constraining what is openly discussed can, of course, signify collusion, whereby patients and practitioners avoid consciously addressing issues, such as those associated with mortality, that present intolerable emotional challenges to both parties [Reference Stiefel, Nakamura and Ishitani238,Reference Stiefel, Nakamura and Terui240]. Collusion is a risk when practitioners’ judgements about when to collaborate with patients’ wish to constrain what is voiced in consultation reflect their own emotional needs and values, for instance concerning death [Reference Helft292,Reference Rodenbach, Rodenbach and Tejani293], so practitioners need to become aware of how their own psychology might shape their communication [Reference Stiefel55,Reference Grignoli, Wullschleger and Di Bernardo239]. Observation of consultations in patients with advanced cancer, and patients’ own accounts, have identified oncologists’ practices that might be collusive in avoiding emotionally challenging subjects such as death and dying [Reference Salvade, Stiefel and Bourquin209,Reference The, Hak and Koeter294,Reference Byrne, Ellershaw and Holcombe295]. However, the line between collaboration and collusion is hard to define, and future research will need to explore the practitioners’ and patients’ perspectives also.

Another manifestation of the pervasive assumption that emotional distress in cancer care necessarily indicates emotional needs that must be addressed at an emotional level is the priority currently placed on screening for distress throughout the treatment trajectory [Reference Ownby277,296]. Enthusiasm for psychological screening in cancer probably owes much to the analogy with biomedical screening, such as routine mammography to detect breast cancer. However, whereas biomedical screening reveals something otherwise hidden from the patient, psychological screening that relies on responses to transparent questionnaire items can hardly reveal feelings or thoughts of which the patient is unaware [Reference Salander297]. Moreover, screening does not, alone, consistently identify patients who want psychological help [Reference Salmon, Clark and McGrath280,Reference Salander297,Reference Meggiolaro, De Padova and Ruffilli298]. When patients who screen positive for emotional distress are asked if they want such help, around half or more say ‘no’ in some studies. Conversely up to a half of those in whom screening identifies no distress do want help. Stigma associated with psychological intervention might explain some distressed patients’ reluctance to seek help, and some who are seriously depressed or anxious might not appreciate that services exist to help them. However, a qualitative study indicated another reason [Reference Baker, Beesley and Dinwoodie299]. Soon after diagnosis, patients did not want specialist psychological support to help cope with their feelings. Rather, they were coping by focusing on treatment. It was only later in the treatment trajectory that patients were more comfortable engaging with formal emotional support. Once again, we see that, from patients’ perspective, clinical management and emotional support are not distinct. Unlike a positive finding from mammography screening, the significance of a positive screen for distress is highly contextual. The challenge for practitioners and services is therefore to help patients identify when they need help, but also to spot those who are not ready to ask but who are ‘stuck’ in their distress or whose distress is damaging them, for example by reducing adherence to treatment [Reference Dekker, Graves and Badger279,Reference Salmon, Clark and McGrath280].

6.5 How Can Cancer Practitioners Manage Information and Involvement

We saw in Section 2 that guidance for practitioners to give ‘as much information as possible’, or ‘the information that the patient wants to have’, is unhelpful in practice. Information is essentially infinite and comprises many different types, from disease biology to treatment side effects, and patients cannot make informed decisions about what information they want until they know what the available information shows. Inductive, qualitative research has shown, instead, that patients need practitioners to manage information for them carefully, rather than provide it and, moreover, that they value knowing that information is being managed for them [Reference Atherton, Young and Kalakonda282,Reference Atherton, Young and Salmon300]. However, like clinical aspects of cancer care, managing information is complex, given its many functions and consequences for patients and their families.

6.6 Conclusion

In this section, we have drawn heavily on inductive research in which practitioners and patients agreed to cooperate with social scientists, to be interviewed by them and to have their discussions recorded and observed. Many practitioners will be cautious about research into their communication, particularly given the criticism they receive from communication researchers and educators. Therefore, those willing to open their consultations to researchers’ scrutiny are a selected subgroup, and the communication strategies that this section described might not be widespread. Conversely, there are probably many more strategies awaiting exposure by future inductive research. However, the point of such research is not to describe what routinely happens, but to identify strategies in use, examine them critically, and make them known so that other researchers, educators and, ultimately, practitioners can learn from them.

7.1 Pedagogy for Clinical Communication Teaching

7.1.1 Educational Pedagogy: A Framework for Setting Objectives

First published in 1956 but revised subsequently [Reference Krathwohl, Bloom and Masia317–Reference Simpson319], Bloom’s ‘taxonomy of educational objectives’ is still widely used in education to help craft learning objectives tailored to the subject (Box 14). It prompts educators, first, to distinguish three distinct domains in which they might create objectives: behavioural actions, or skills; knowledge and cognition; and affect, or emotion, including the attitudes and values that underlie emotional reactions. For each domain, Bloom’s framework further prompts educators to decide the level of change to be sought in learners, from very limited levels through progressively more complex ones that demand more of both learners and educators. The domain of knowledge and cognition has received most attention from educationists, but each domain is potentially relevant to clinical communication.

Box 14 Bloom’s taxonomy of educational objectives

Educators can create learning objectives in each of three domains and, within each domain, at progressively more complex levels [Reference Krathwohl, Bloom and Masia317-Reference Simpson319]. (See https://www.simplypsychology.org/blooms-taxonomy.html#Cognitive-Domain-1956.)

The Behavioural Action Domain

As we saw in Section 5, the term ‘skill’ has been applied widely to diverse aspects of clinical communication, such as ‘appropriate eye contact’ or ‘empathy’, which defy behavioural definition. Nevertheless, potentially important objectives for clinical communication might be created in this domain, where learners need to learn behavioural repertoires that they might not have acquired before clinical training. At the lowest levels, objectives would be simple scripted behaviours such as checking a patient’s identity to start a consultation. At more complex levels, they would encompass interactive skills, which cannot be fully scripted because they depend on patients’ responses; for example, organizing, signposting and sequencing information, or ‘de-escalation’ protocols to respond to challenging patients. Learning specific frameworks to structure or sequence a consultation, like SPIKES (Section 5), might be valuable for new practitioners, or ones in training, who lack the experience to improvise. Learning linguistic techniques of argumentation could help doctors lead patients’ decision-making (Section 4) [Reference Labrie and Schulz320–Reference Bigi322]. However, given the unpredictability of conversation with another individual, high-level objectives in this domain will emphasize, not the performance of reproducible skills, but learners’ flexibility and creativity in adapting them to individual patients and situations [Reference Stiefel and Bourquin6,Reference Salmon and Young206].

The Affective, or Emotional, Domain

While CST places a high priority on patients’ emotions, and on practitioner–patient dialogue at an emotional level, it has less to say about whether and how educators can engage with practitioners’ own character and emotional experience [Reference Stiefel and Bourquin6]. By contrast, the analysis in this Element has pointed repeatedly to the need to recognize that quality of clinical communication depends on practitioners’ character and motives. Therefore, there is much scope for creating objectives in this domain. At the basic level, learners need to be aware of their values and emotions in communication with patients [Reference Stiefel, Bourquin and Salmon56,Reference Stiefel, Kiss and Salmon192]. At higher levels of complexity, they will be able to question and, where necessary, modify these. At the highest, they will evolve a value system that reflects their own individual character and experience as well as the demands of their profession [Reference Mendick, Young and Holcombe248].

The Knowledge and Cognition Domain

Surprisingly, CST has not emphasized learning objectives in the knowledge domain, in striking contrast with other clinical subjects, like physiology or pharmacology, in which learners acquire a sound knowledge base before deploying relevant skills. For instance, even after two decades during which its relevance to clinical practice has been described [Reference Salmon and Young49,Reference Ciechanowski82,Reference Wilkinson323], attachment theory is missing from major textbooks on clinical communication [Reference Salmon and Young324]. Other relevant areas of knowledge are also absent from communication textbooks and curricula. Without being taught about the psychological effects of the fear or reality of serious disease it will be harder for practitioners to understand their patients’ experience and to judge how best to support them [Reference Stiefel and Bourquin6]. Similarly, practitioners will need to know how patients explain, or make sense of, illness so that they can craft explanations that find common ground between biomedical and lay knowledge systems [Reference Salmon325]. Educators could also disseminate knowledge of communication strategies that inductive research has discovered practitioners using, such as those in Section 6. Without a strong knowledge base, learners are placed in the role of technicians rather than scientifically grounded practitioners; that is, following rules without the understanding that would allow them to make good judgements about when and how to apply – and, where necessary, break – those rules.

At the lowest levels in this domain, educators’ objectives will be for learners simply to remember what they have been taught, perhaps reproducing it for written exams. Therefore, didactic, classroom teaching will be important, just as for other clinical subjects. At progressively more complex levels, objectives will be for learners to evaluate knowledge critically, and to be creative and imaginative in how they use it to make judgements in specific situations with specific patients. That practitioners are curious about their patients will be an objective, too, so they can learn from them.

The knowledge and cognition domain prompts educators to look beyond the kinds of knowledge that allow judgements only about abstract or technical matters. Dating from Aristotle’s accounts of the virtues that underlie good judgement, the concept of ‘practical wisdom’ refers to knowing how to make judgements that are sensitive to the nature and circumstances of specific situations. In the present context, it means the practical know-how whereby practitioners with good motivations weigh up what goals are important in any clinical interaction and judge how to achieve them; that is, to do the right thing at a specific moment with a specific patient [Reference Kotzee, Paton and Conroy9,Reference Plews-Ogan, Sharpe, Brown, Veen and Finn326]. It therefore depends, not just on having the right motives and knowledge, but on the experience of working with these in practice [Reference Boudreau, Wykretowicz and Kinsella210,Reference Kaldjian, Yoon and Ark327].

In each domain, therefore, we see that the highest levels of learning emphasize creativity and originality. Several educationists and researchers in clinical communication have already argued that the quality of communication depends on these levels of learning; that is, not so much on practitioners’ ability to reproduce skills as on their good judgements about when and how to use them, and on their ability to be creative and imaginative in adapting to the unpredictability of clinical practice[Reference Stiefel, Bourquin and Salmon56,Reference Stiefel, Kiss and Salmon192,Reference Salmon and Young206,Reference Hatem, Mazor and Fischer228,Reference Haidet328]. Experienced educators will already work towards those higher-level objectives, particularly when they meet with experienced practitioners. However, to the extent that the scientific and educational literature that informs CST neglects this level of objective it will not inform their work, which will therefore remain a type of ‘craft’ learned through experience or passed on from one teacher to another.

7.2 Pointers to the Next Generation of Communication Teaching

A recent consensus statement set the scene for renewing communication teaching in ways that diverge in important respects from the past [Reference Stiefel, Kiss and Salmon192]. It was concerned specifically with cancer care which, as we have seen, has long been in the lead in communication teaching. It revised the consensus of European experts reported ten years earlier [Reference Stiefel, Barth and Bensing50], advising educators to look beyond standardized skills to the attitudes and emotional feelings that shape practitioners’ communication. To help practitioners understand the relational context of communication, educators should teach about relationships but also help learners to achieve and maintain the emotional self-awareness that would help foster effective relationships. Implementing this guidance will clearly need educational methods that go beyond training communication skills, to include fostering the competencies and qualities that inform and motivate their judgements about how to communicate in practice [Reference Stiefel and Bourquin6]. That is, the current deontological approach will need to give way to one closer to virtue ethics. Reports over recent decades provide some pointers to approaches that go beyond the teaching of skills to engage with learners also at the levels of knowledge and cognition, and emotions and attitudes.

Several reports have focused on empathy, reflecting its central importance in communication literature. Recognizing that practitioners’ empathy for their patients should start with understanding them and feeling empathic towards them rather than with empathic behaviours, some educators have explored narrative and experiential approaches to sensitize learners to patients’ experience [Reference Doukas, Ozar and Darragh345]. In one account of teaching about ‘bad news’ consultations, junior doctors saw video-recordings of clinical communication informed, unusually, not by assumptions about what patients would value, but by the accounts of patients who had watched those recordings [Reference Hatem, Mazor and Fischer228]. Reflecting on real or fictional patient narratives or poignant medically related stories from personal experience, popular literature or films has also been advocated as a way to inform empathy in doctors in training or practice [Reference Wear and Varley334,Reference Shaw, McQuade and Reilley346]. Medical students have accompanied patients through a series of medical encounters [Reference Plews-Ogan, Sharpe, Brown, Veen and Finn326], or have even been hospitalized [Reference Wilkes, Milgrom and Hoffman347] so that they could reflect on the experience of being a patient. Some educators have suggested that theory about acting could help teach empathy. They argue that teaching skills promotes ‘surface acting’, whereby learners perform behaviours associated with empathy regardless of whether or not they feel empathic. By contrast, ‘deep acting’ means learning to feel empathy and responding to that feeling [Reference Larson and Yao348,Reference Finestone and Conter349]; that is, instead of creating a task focus in learners, they could be helped to use their own feelings to become attuned to the ‘atmosphere’ of a consultation [Reference Langewitz176]. While actors are widely used in CST to play the part of patients, expertise in acting might inform teaching itself.

When learners are experienced practitioners, they bring their own experience and expertise – and the authority that arises from that experience and expertise. Although experienced communication skills teachers will already draw on that background in teaching sessions, they are not helped by the theoretical underpinning of CST which emphasizes expert-defined skills, giving little status to communication strategies, such as those in Section 6, that practitioners have developed informally [Reference Rollnick, Kinnersley and Butler344]. Some educators have described approaches to communication education that explicitly put practitioners’ expertise at the centre. Rollnick et al. [Reference Rollnick, Kinnersley and Butler344] contrasted traditional ‘workshop’ training, typically held outside practitioners’ place of work and with objectives and content directed by educators, with an approach they called ‘context-bound’ training that explicitly rejected the assumption that the practitioners lacked skills that they needed experts to teach. To ground training in their learners’ (GPs) everyday experience rather than the educators’ knowledge of communication skills, training was in the GPs’ premises and shaped by their own preferences. Participating GPs consulted with a simulated patient shortly before meeting with colleagues in groups facilitated by the educators. Crucially, each GP received a transcript of their consultation, so that group discussion could be informed by GPs’ reflection on their own experience. The educators accepted that, as the experts on what they needed, each GP would take something different from the training. As one GP explained, ‘It wasn’t about communication skills training. It was about getting better at what we do every day.’

Expertise is not, though, a ‘zero sum’, in which the educators need to claim less to make space for that of learners. Some recent reports describe initiatives anchored in practitioners’ clinical settings, and in their everyday experience and expertise, in which educators provide specialist knowledge to inform practitioners’ own judgements about communication. Stiefel et al. described how individual supervision from a psychological expert could foster oncologists’ awareness of ways in which their own personal history and emotional responses complicated their relationships with patients [Reference Stiefel, Nakamura and Terui350,Reference Stiefel, Bourquin and Wild351]. Realistically, resources for individual supervision will be scarce in many services. Therefore a more cost-effective approach could be that described by Salander [Reference Salander and Sandström237]; the psychological expert facilitated a group of oncology practitioners meeting periodically to analyse communication and relational challenges in their work. As well as using scarce psychological expertise efficiently, this format allowed peers to contribute their own insights to solutions that were clinically realistic for the context. Many communication dilemmas for experienced practitioners in oncology and other specialties contain ethical dilemmas, where the ‘right thing’ to do or say is unclear. Prompted by emotional and moral challenges created by restrictions associated with the COVID-19 pandemic, Delany et al. [Reference Delany, Benhamu and McDougall352] described ‘reflective ethics discussions’, in which the source of external expertise to facilitate reflective groups was bioethics rather than psychology. Delany et al. were explicit that the function of external expertise was not to tell practitioners what to do, but to help practitioners reflect on what they were experiencing so they could find their own responses.

A programme of teaching that the originators called ‘Program to Enhance Relational and Communication Skills (PERCS)’ [Reference Browning, Meyer and Truog353–Reference Bell, Pascucci and Fancy356] illustrates how the term ‘skills’ is now widely used simply to mean ‘good communication’, because this programme goes far beyond a concern with behavioural skills to include many of the curriculum elements described previously. It starts with an explicit statement of philosophy and pedagogy. This grounds teaching in the uniqueness of clinical encounters and the complexity of clinical practice in which good communication depends, not on reproducible skills, but on practitioners’ attitudes, values and relational capacities including curiosity about patients, and in which there is typically no one ‘correct’ way to communicate [Reference Lamiani, Meyer and Leone355]. The curriculum format, centred on scenarios played by actors, allows educators to contribute expertise, while eliciting and valuing participants’ own expertise and insights. Rather than play standardized patients, the actors improvise so that scenarios replicate the unpredictability of routine practice. Arising from a paediatric setting, PERCS has been extended to other specialties, and has included multiple disciplines and both students and experienced practitioners.

These examples illustrate features that will characterize the next generation of communication education: an emphasis on relational and contextual aspects of communication, recognition of patients’ and practitioners’ individuality, respect for learners’ expertise, and provision for learners to understand and reflect on what they are learning. They also illustrate that, just as there is typically no one way to communicate in any situation, there will be no single way to teach communication; experimentation will be needed. Approaching communication teaching from a perspective closer to virtue ethics than to a deontological emphasis on behavioural skills brings further possibilities. Communication teaching will overlap considerably with teaching of medical ethics and professionalism, which focus explicitly on learner qualities that influence their care of, and communication with, patients [Reference Kinghorn8,Reference Doukas, Ozar and Darragh345,Reference Lyon357]. Therefore, expertise and insights from teaching those subjects can inform communication teaching. For instance, communication skills teachers have long been concerned about limited transfer of skills from teaching settings into clinical practice. However, education that emphasizes learners’ qualities will refocus this concern with generalization. Learners need to transfer, not predefined skills, but the ability to adapt communication to the context [Reference van den Eertwegh, van Dulmen and van Dalen358]. An emphasis on the underlying qualities that guide this adaptability – including knowledge, curiosity and conscientiousness – would provide the continuity that is needed between formal teaching and informal practice.

Communication research and teaching can also engage with current debates around virtues and their relationship to practitioners’ behaviour, that have been fuelled by disenchantment with the deontological approach of behavioural guidelines and protocols as ways to help clinicians with morally ambiguous judgements in clinical practice [Reference Kinghorn8]. These debates emphasize that, while understanding why moral considerations such as patient-centredness are important will be necessary for learners to embrace the principles, or virtues, that should guide their relationships with patients [Reference Duggan, Geller and Cooper2], merely talking about virtues will be insufficient; habit-forming practice or observation of skilled practitioners is also necessary [Reference Lyon357]. Therefore, there will be much more concern than now with the communication expertise of educators and supervisors. Technical knowledge about communication will be insufficient, and learners will also need exposure to practitioners who are recognized as having acquired the ‘practical wisdom’ of excellent clinical communicators [Reference Kinghorn8,Reference Plews-Ogan, Sharpe, Brown, Veen and Finn326]. Social scientists or practitioners who lack clinical experience in the specific clinical context of their learners – for example when psychologists, nurses, GPs or psychiatrists deliver training for cancer clinicians – are therefore necessarily constrained in what they can achieve alone. Ethicists applying virtue ethics ideas to clinical education also warn that merely observing experienced practitioners is insufficient. Learners need to be helped to reflect on what they observe and on their own behaviour [Reference Kaldjian, Yoon and Ark327]. Therefore, for example, observing practices associated with patient centredness could be a stage in the formation of authentically patient-centred practitioners, provided that learners acquire in parallel the capacity for self-reflection through which they can internalize the values of authentic patient-centredness [Reference Duggan, Geller and Cooper2]. The authors of one educational program explicitly designed to facilitate practical wisdom in clinical interactions chose to embed their medical students in continuous longitudinal relationships – with patients, peers and senior doctors – in the belief that this would promote real reflection better than more fleeting encounters would [Reference Plews-Ogan, Sharpe, Brown, Veen and Finn326].

A virtue ethics perspective does not mean seeing practitioners as saintly and infallible, or on trajectories to becoming so. Evidence that human behaviour is heavily determined by situational factors that can outweigh the influence of people’s virtues [Reference de Bruin, Zaal and Jeurissen359] is a warning that the qualities which clinical training will seek to inculcate have to jostle with practitioners’ other attributes and priorities, and with the reality of clinical care and professional life. In the present context the value of a virtue ethics perspective is therefore not as a theory about how character might outweigh situational influences [Reference Upton360], but as a pragmatic way to address the reality that good communication relies on practitioners making good judgements for good reasons. Moreover, the capacity to make good judgements is not an end-state to be reached by sufficient training and experience; rather, it is a continued process of learning and reflection [Reference Plews-Ogan, Sharpe, Brown, Veen and Finn326,Reference de Bruin, Zaal and Jeurissen359].

7.3 Evaluation of Communication Teaching

7.4 Conclusion

While the portrayal of clinical communication as made up of skills perhaps helped facilitate the subject’s incorporation into clinical curricula alongside other ‘clinical skills’, educators and researchers need a richer conceptual background for developing and evaluating the next generation of clinical communication education. This section has pointed to ways for communication education to escape the constraints of the skills model. Doing so will help those educators whose teaching already goes beyond skills to articulate and reflect on what they are doing; and it will help to build the pedagogical base for the next generation of communication education to become as intellectually lively as it should be, given its position at the interface of ethics, theory and practice.