Medical Care as Flea Market Bargaining? An International Interdisciplinary Study of Varieties of Shared Decision Making in Physician–Patient Interactions

Abstract

Phenomenon: Shared decision making (SDM) is a core ideal in the interaction between healthcare providers and patients, but the implementation of the SDM ideal in clinical routines has been a relatively slow process. Approach: In a sociological study, 71 interactions between physicians and simulated patients enacting chronic heart failure were video-recorded in China, Germany, the Netherlands, and Turkey as part of a quasi-experimental research design. Participating physicians varied in specialty and level of experience. The secondary analysis presented in this article used content analysis to study core components of SDM in all of the 71 interactions and a grounded theory approach to observe how physicians responded actively to patients even though they did not actively employ the SDM ideal. Findings: Full realization of the SDM ideal remains an exception, but various aspects of SDM in physician–patient interaction were observed in all four locations. Analyses of longer interactions show dynamic processes of interaction that sometimes surprised both patient and physician. We observed varieties of SDM that differ from the SDM ideal but arguably achieve what the SDM ideal is intended to achieve. Our analysis suggests a need to revisit the SDM ideal—to consider whether varieties of SDM may be acceptable, even valuable, in their own right. Insights: The gap between the SDM ideal and SDM as implemented in clinical practice may in part be explained by the tendency of medicine to define and teach SDM through a narrow lens of checklist evaluations. The authors support the argument that SDM defies a checklist approach. SDM is not uniform, but nuanced, dependent on circumstances and setting. As SDM is co-produced by patients and physicians in a dynamic process of interaction, medical researchers should consider and medical learners should be exposed to varieties of SDM-related practice rather than a single idealized model. Observing and discussing worked examples contributes to the physician’s development of realistic expectations and personal professional growth.

Keywords:

• Physician–patient communication
• shared decision making
• physician–patient interaction
• international comparison
• patient contribution

Phenomenon

Clear communication is a vital element of a physician’s skill set.^1,2 Integrating the patient’s perspective into treatment decisions and the decision-making process is a core goal of medical education.³ The principle of shared decision making (SDM) in patient medical care challenges historical paternalistic notions of the physician–patient relationship. In the ideal shared decision-making process (the SDM ideal), information sharing is a prerequisite of good medical care: at least two persons participate in this process, and the ideal outcome is a mutually accepted treatment decision.⁴ Within the SDM ideal, the patient becomes a competent and active partner not only in the decision-making process but also in the therapeutic process. Many scholars have studied the obstacles to SDM in real-life medical encounters and developed practical guidelines for improved practice.^5–12

Despite wide acceptance of the SDM ideal, empirical studies show that implementation of SDM in clinical routines has been relatively slow.^13,14 In a recent cross-sectional survey in a Dutch hospital, a majority of physicians claimed a preference for the SDM ideal but reported paternalistic decision-making choices and actions when asked in more detail about their clinical practice.¹⁵ This gap between ideal and practice can be explained in part by various restricting factors.¹² Employing the SDM ideal depends on the patient having access to a range of treatment options and both physician and patient having ample time to consider these options together. Elwyn pointed to cognitive, emotional, and relational work that has to be done in the context of inevitable power asymmetries, unequal knowledge, and differences in status: “There is no one right way to achieve shared decision making. Each interaction will bring different decisions, contexts, and prior relationships into the mix.”^14(p1593) Cultural differences between physician and patient add to the communication challenge.¹⁶

Arguably, lack of attention paid to the patient’s context is an underestimated cause of medical errors during the development of care plans.^13,17 It is therefore vitally important not only to identify and understand the barriers to implementation of the SDM ideal in clinical practice but to suggest ways these barriers might be overcome. We hold that prevalent research designs are limited in that they usually follow a checklist approach to assessing the implementation of the SDM ideal, with few exceptions.^18,19 Experienced practitioners are less often selected for participation in studies of SDM as compared to medical learners.^20–23 These limitations may in part explain why barriers to the implementation of the SDM ideal are not yet well comprehended.

The present study uses physician–patient interactions observed in an earlier sociological study. The original study asked whether and how medical professional knowledge differed across the world^24–27; it therefore created a geographically dispersed and culturally heterogenous sample that included experienced and specialized physicians besides residents and medical learners. The current secondary study aims to gain deeper insight into the complex dynamics of physician–patient communication by studying varieties of SDM-related practice.

Accepting the value of varieties of SDM may be seen as problematic for the implementation of textbook ideals of medical communication. For example, the international and interdisciplinary team of physicians, medical educators, and sociologists authoring this article identified several processes of physician–patient interaction as examples of best-practice SDM, even though physician and patient did not reach full agreement. In one case, physician and patient negotiated a mutually acceptable compromise,^28,29 an event that some might argue violates ethics in the field of healthcare. Is it justifiable to view a compromise negotiated in a physician–patient interaction—like a final price haggled in a flea market—as an example of best practice in medical care? And can we view a physician who exerts authority in a physician–patient interaction as responsive to a patient when that authority is exercised in circumstances where the patient does not fully acknowledge their medical condition?

Acknowledging these normative debates, we emphasize that acceptance of variation in SDM-related best practice may not only be more appropriate in clinical settings but also allows for good communication and decision making despite the differences between physician and patient that are likely to persist.^30,31 Patients often have to deal with existential challenges that physicians cannot solve for them.³² Physicians follow professional and institutional standards that can be implicit or complex; even physicians who follow SDM principles may be unable to communicate all of their concerns during a dynamic interaction.³³ The concept of negotiation recognizes that physicians and patients need to take a decision together while respecting their existentially different perspectives.

Educational strategies that promote SDM-related practice may be more effective if medical learners are exposed to varieties of SDM to enhance their personal professional growth. This approach can help them develop realistic expectations of what is possible in a decision-making process. Accepting varieties of SDM-related practice might also aid clinicians in overcoming barriers to the implementation of SDM in their interactions with patients.

Approach

Original study

In a sociological study on the globalization of medical professional knowledge, a representative sampling of physicians worldwide was deemed to be unrealistic.²⁷ The study therefore created a “quasi-naturalistic experiment.”³⁴ Between March 2019 and January 2020, 71 physicians at four university hospitals situated in Ankara (Turkey), Beijing (PR China), Groningen (The Netherlands), and Wuerzburg (Germany) were video-recorded during first contact with a simulated patient (SP) reporting symptoms of chronic heart failure. The original study argued that similarities observed despite significant differences between the sites of observation would point to a universal medical professional culture.^24–27

To maximize contrast, research collaborators in each setting had recruited participants with various specialties—cardiologists, internists, or family practitioners—and with a range of expertise, from interns to residents to experienced hospital staff to professors (see Table 1). This participant range is of benefit to the present study, because research on SDM profits from studying video-taped practice of experts.^20–23 In particular, in the Wuerzburg and Groningen samples, we observed 16 experienced cardiologists, including nine professors or heads of academic departments.

Table 1. Level of experience in the sample

	Ankara (Turkey)	Beijing (PR China)	Groningen (Netherlands)	Wuerzburg (Germany)	Total
Total	20	20	18	13	71
Professor/head of department			5	4	9
Staff/experienced physician			4	3	7
In residency	11	14	4	5	34
Intern	3	6	4	1	14
Missing values	6		1		7

Physicians knew that the patient was an SP. They were instructed to treat the patient as they would in their usual clinical practice. Because SPs are often used in educational settings³⁵ and because physicians were aware of being observed for a project studying professional knowledge and practice, participating physicians were motivated to demonstrate their best level of clinical practice despite the instruction. The original study made no reference to SDM. All participating physicians gave their informed written and oral consent prior to participation in the study. We anonymized names and details of all participants.

A core team of five sociologists, a cardiologist, a health economist, and a medical educator in Germany designed the original study. Collaborators at sites of observation were physicians (Germany) or medical educators (Turkey) or both combined (PR China, Netherlands). The core team’s cardiologist and medical educator created a patient profile of a 59-year-old man with chronic systolic heart failure (n = 39) for which clear treatment guidelines existed.³⁶ A female profile with unclear treatment options was used as a contrast (n = 32). Profiles were complemented with a professional and family biography. The German medical educator (Germany, Netherlands) or a local medical educator (Turkey, PR China) trained locally recruited SPs, who were encouraged not to act as “engaged patients”³⁷ but to present as friendly and reserved, answering the physician’s questions comprehensively but appearing reluctant to take initiative. During the consultation, SPs provided the physician with a file that included a case history, lab reports, electrocardiogram (ECG), and an ultrasound report.

Most consultations took place in the local language. In the Netherlands, physicians and patients were asked to use their second working language (English) rather than the local Dutch to avoid inaccuracies related to translation. Transcripts were prepared in the language of interaction; in addition, consultations conducted in Mandarin and Turkish were translated into English.

Because the original project searched for location-related differences in medical knowledge and practice, we invited local collaborators to adjust the SP profile. Collaborators in Ankara, for example, changed the profession of the male SP from taxi driver to administrator. SPs were permitted to incorporate their own experience and imagination into their interactions with the physicians.^38,39 Consequently, though SPs provided similar information to physicians in all four locations of the study, the individual physician–patient interactions in each scenario varied.

The original study focused on the practical implementation of technical medical knowledge rather than communication skills. Contrary to our expectations, we found an unexpected degree of convergence concerning the application of treatment guidelines and the use of gestures.^25–27 The focus of the current analysis therefore shifted to a topic with more obvious variation: the dynamics of physician–patient interaction.

Secondary study

The present study is a secondary interdisciplinary analysis of the prevalence and varieties of SDM-related practices. According to Heath and Luff, “quasi-naturalistic experiments” are “not primarily concerned with the evaluation of theory or identification of causal explanation but rather with clarifying and discovering knowledge, practice, and reasoning that inform the interactional production of everyday organizational activities.”^34(467) Referencing their distinction between “experiment” and “naturalism,”³⁴ our observations can be seen as “naturalistic” because they took place in clinical settings; patients’ symptoms and the problems that physicians were challenged to solve were typical of those that physicians encounter in their daily work. At the same time, our observations were “experimental”: by using SPs, we could present physicians with similar clinical problems across four locations.

Research comparing simulated and authentic medical consultations found differences between the two types of consultation. In simulated consultations, physicians tend to overexaggerate formulaic phrases and behavior that conform to communication guidelines such as the Calgary–Cambridge model.⁴⁰ Most studies reporting differences between simulations and real practice were conducted with medical students in the role of physicians.^41–43 The sample in our analysis included a significant number of senior physicians in addition to residents and interns (see Table 1). We agree with Grießhaber’s argument that agents experienced in a particular type of interaction are more likely to perform a situation “naturalistically” during simulations, at least with regard to informal routines and in response to spontaneous challenges by the patient.⁴⁴

For the current analysis, further medical educators were invited to join the team, including a linguist studying simulated patient practice (C.B.). Thanks to the continued interaction within the original project, we recognized that the medical educators in our team expected a standardized checklist approach to assess whether a specific interaction conforms to the SDM ideal, whereas the sociologists on the team favored exploratory qualitative methods.⁴⁵ This is suggestive of how different fields use different methods for practicing (and teaching) SDM—one of the insights yielded by our research.

We responded by using two methodological approaches. The first approach was favored by the medical educators on the team: three collaborating coders used content analysis to code the entire sample of 71 cases in accordance with the ideal of medical communicator as envisaged in CanMEDS 2015.^1,46# Because exploring the patient’s experience is not always part of clinical practice, particular attention was given to the two core components of SDM,¹ active listening and deliberation of alternatives.^11,47

The second approach was put forward by the sociologists on the team: we used the grounded theory method proposed by Glaser and Strauss⁴⁸ in line with Han et al. stating that “[c]urrent communication skills assessment practice should be revisited as it itemizes physicians’ communication skills as distinct and separate constructs rather than mutually affecting dynamics. Rather than imposing a theoretical rubric, assessment criteria should evolve through naturalistic observations of physician-patient communication.”^18(p33),45

The grounded theory analysis focused on a problem for some SPs and physicians that we had not foreseen when constructing the script. In a subsample of 29 cases, the male SP portrayed a taxi driver who heads his own business. He works long hours and hopes in the long run to pass the business on to one of his three children. This script results in existential challenges;³² the patient was torn between the competing desires to provide for his family and to make life changes in order to take care of his health. As a consequence, physicians’ responses and the dynamic process of physician–patient interaction showed a high level of variation.

In line with the grounded theory method, our analyses reconstructed line by line how the problem was introduced by the SP, whether and how physicians responded, and vice versa.⁴⁹ Due to our interest in medical learning, analysis focused on the content of mutual suggestions and decisions, not on the formal organization of interaction that is the domain of conversation analysts.^20,21,23 The analysis was started by a discussion of selected transcribed and translated videos during a 2-day online workshop. Here, the multidisciplinary team of authors debated what could be seen as SDM-related practice. On this basis, two sociologists (I.S., A.W.) conducted a line-by-line analysis of all 29 cases in Beijing, Groningen, and Wuerzburg in which the male SP portrays a taxi driver and treatment guidelines are clear.^* Varieties of the core category of “negotiation” are presented as case studies to show how subtle differences change the dynamics of physician-patient interaction and produce varieties of SDM-related practice.

Findings

Prevalence of SDM

In the present study, coronary angiogram and CT scan are alternative diagnostic options. Content analysis shows that during first contact with the SP, most physicians postponed this decision by referring to “further diagnostics” or by transferring the patient to a specialized unit. But 23 of 71 physicians, mostly specialists, mentioned either angiogram (n = 14) or CT (n = 4) as the next step in diagnosis, sometimes explaining the advantages of the option the physician had chosen.^† Only 5 physicians explained both options, and only one of these had the SP make an informed decision for themselves.

Content analysis also shows that some aspects of the SDM ideal were common across the sample. For example, all physicians comprehensively informed SPs about diagnosis and treatment. Elwyn et al. see active listening as one of two core components of the SDM ideal and define it as “paying close attention and responding accurately.”¹¹ In our sample, 32 of 71 physicians (45%, distributed across all four locations) checked to see whether the patient had understood, often at the end of the interaction: “something you want to ask me. I mean, is there any point you didn’t understand?” (Ertekin [anonymized name]: 326 [paragraph number in full transcript]; verbatim translated from Turkish). The same number of physicians (32 physicians, or 45%), although not the same physicians in all cases, anticipate the anxiety of the patient: “Don’t worry too much. This problem is common to many people” (Dai: 162). Some physicians’ versions of active listing are pronounced. Dr. Ou, for example, not only repeats what the SP has expressed earlier in the conversation but also expresses empathy: “I can see that the burden on your shoulder is very heavy. I can understand that you are the head of the family, you are the pillar of your family. Not only must you consider your own problems and your body but also consider what to do for the future” (Ou: 270). Content analysis suggests that though we did not find textbook examples of the SDM ideal, a total of 56 out of 71 physicians (79%) aspire to explore the patient’s perspective by integrating aspects of the SDM ideal into their observed clinical practice.

Varieties of SDM

Our grounded theory analysis shows that physicians reacted with significant variation to the existential challenges faced by SPs, who, in 29 cases, expressed concern for their health and business prospects. The early stages of heart failure rarely necessitate an immediate cessation of driving a car, but some physicians expressed concerns about the side effects of a new medication, which could lead to fainting. A heightened risk of heart attack or major rhythm disturbances arguably incurs a legal responsibility to protect customers and the public. Some physicians thus recommended a long-term or short-term cessation of driving or, in few cases, immediate hospitalization. In other cases, discussion of existential challenges did not occur: several experienced cardiologists postponed discussion of foreseeable problems that might result from implantation of a cardioverter-defibrillator to the next stages of treatment; in debriefing, these physicians told us that it is important not to scare a patient in the first physician–patient encounter but, rather, to motivate the patient to undertake further treatment. Less experienced physicians tended to focus on the SP’s options for short-term lifestyle changes and stress reduction.

These general insights overlap across individual cases, which supports the first and most important result of the grounded theory analysis: The complex dynamics of physician–patient interaction elude an itemized approach and can best be understood as a dynamic process of negotiation. The following four case analyses exemplify significant variations found in the subsample of taxi driving male SPs (n = 29). The first two cases show physicians who are sensitive to the SP’s cues, but interactional dynamics prompt them to become less responsive to the SP’s expressed concerns in their final recommendations. The last two cases show physician–patient interactions that our team of authors identified as best practice even though they deviate from aspects of the SDM-ideal.

Case 1: Active listening produces unexpected outcome

The first case was chosen from several similar cases to demonstrate how the SP changes the physician’s mind. Subsequently, the interaction moves in a direction that is problematic for the SP. Experienced cardiologist Dr. Jung treats SP Mr. Herbst in Wuerzburg (see Supplemental Online Appendix 1. Please see the Supplemental Online Appendices for the full transcript of all cases. Line numbers shown in parentheses here correspond to lines in the Supplemental Online Appendices). We might assume from the consultation that Mr. Herbst has learned from previous interactions with physicians, because he brings up the matter of taxi driving and heart disease four times. He first mentions concentration problems during work (line 2), to which Dr. Jung responds with active listing. A little bit later, the patient asks whether he will be able to work as usual while doing a 24-hour Holter ECG (11–13), and Dr. Jung confirms that he will be able to do so. To the third mention of the problem (15–16) the physician responds by explaining further diagnostics. At the very end of the consultation, Dr. Jung asks an open-ended question, as discussed above:

D: Are there then other questions from your side? (2)

P: no, I was-, I was simply a bit taken aback that this so (.)

D: Suddenly, yes,

[…]

P: and that is of course very bad eventually if one sits somehow in the car all day long and has the feeling uhm (.) don’t know but I am not really well that’s of course not nice

D: Well, I would in the moment actually advise against you driving a car and in particular driving a taxi (.) until we have clarified the whole thing (34–43)

Mr. Herbst responds to the open-ended question by expressing his emotions. Dr. Jung shows empathy and Mr. Herbst talks again about not feeling well all day. Only at this point does Dr. Jung advise Mr. Herbst to stop driving.

In this case, the outcome is paternalistic, but the physician clearly explores the patient’s perspective and responds to it. In fact, the SP seems to nudge the physician toward prohibition by repeatedly mentioning the problem. The decision is not discussed further, even though the SP had emphasized earlier that continued driving is a top priority for him. In this case, the physician was led by a process of active listening to the unilateral conclusion that the SP should stop driving.

Case 2: Attentive listening consolidates physician’s position

The second case is presented here as an informative contrast to the first case. Dr. van Dijk, an experienced cardiologist in Groningen, responds to SP Mr. de Boer with an immediate prohibition of driving, and he never changes his mind (see Supplemental Online Appendix 2). Upon closer analysis, a similarity to the first case also becomes apparent: Like Dr. Jung, Dr. van Dijk responds to his patient’s cues. He becomes more adamant in his advice because he notes during the interaction that the SP seems not to take his condition seriously.

In the course of the interaction, Dr. van Dijk first appeals to the SP’s sense of responsibility:

D: […] it would be = uh important not to = uh not to to drive = uh erm professionally in th in the first (unclear: two) days after after medication is = uh (.) is started, an an increased (.) to see if if uh if if you tolerate it well (.) Cause obviously an only have to (.) think about yourself, but you al = ys you = s yo also have to think about your customers. So

P: Yeah, tha that = s (right)

D: Yeah, yeah

P: Yeah, tha-that = s (right)

D: Yeah, yeah

P: but but (.) I er I, well, we have not so many drivers so I schedule myself every day, so tha that = s (.) That will be (.) hard, but when when should that (1) maybe I sh (.) when can mou should I start with that medication because the weekend

D: No, we star

P: is always very busy so then (.) maybe after the weekend it = s easier to do but

D: Yeah. No, w-we can start with the medication today (.) basically (6–17)

In contrast to the first case, this interaction shows a physician who is initially inclined to share a responsible decision with the SP. In lines 4–8, the physician suggests that Mr. de Boer should cease driving for 2 days after beginning a course of new medication and watch for side effects for some weeks afterwards. In response to Dr. van Dijk’s appeal, the SP offers to start medication after the weekend but he also mentions several difficulties. Dr. van Dijk does not respond to the SP’s offer to start medication after the weekend. After listening to the SP attentively, the physician exerts authority in order to protect his patient’s customers. He informs the patient that he must stop driving immediately.

In both cases, the active role taken by patients and the dynamics of the physician–patient interaction are salient. Both cases conclude with the physician taking a decision, even though interaction evolves through subtle negotiations as both patient and physician listen and react to one another.

Case 3: Emotional resonance

Other cases better reflected a more broadly defined ideal of SDM.^19,50,51 We selected case 3 to demonstrate how the sharing of decision making can remain implicit, but the emotional resonance in the interaction between physician and patient still leads to a mutually accepted conclusion.

Dr. Ou, a junior resident in internal medicine, treats SP Mr. Wang in Beijing (see Supplemental Online Appendix 3). Directly after diagnosis, the SP expresses his questions and sorrows and asks how long it will be before his life is normal again. Most of all, he expresses worry about his company. He says that although he has tried to cope, recently he has not been able to do so (4). The video shows that Dr. Ou makes eye contact with Mr. Wang and listens carefully, indicated by listening-confirming response tokens such as “mmh,” nodding, and keeping her head slightly at an angle. Mr. Wang then mentions his concern that he will die in his 50s like his father (7–8). As quoted above (see Findings: Prevalence of SDM), Dr. Ou expresses her understanding that, as the head of his family, Mr. Wang must think about the future (9–11). She reassures him by saying that Mr. Wang’s situation is not comparable to that of his father, because medical treatment has progressed (13–17). Then she emphasizes that it is necessary for the SP to rest:

D: […] If it were my own father, I would also think that it is, of course, the most important thing to have a healthy body. Everything else is in second place.

P: Right, right (.)

D: At the very least, there are still children (unclear: to help you). Don’t you think so?

P: Yes, you are right (22–27)

The focus of this interaction is not decision making but, rather, understanding, empathy, generating an atmosphere of trust, and encouraging the SP to follow sound medical advice. Mr. Wang seems not to be searching primarily for the objectively best decision but for compassion and understanding from his physician. Dr. Ou validates this mode of dialogue and responds empathetically that his children might want to help him. She offers him hope and encourages him to care for his health and take seriously the necessity to rest. Emotional resonance, as one important aspect of active listening, can facilitate physician–patient communication and promote trust in clinical decision making.⁵¹

Case 4: Negotiated compromise

Another experienced cardiologist in Groningen, Dr. Nijhof, gives us a salient example of decisions that are not “shared” in the strict sense of the word but explicitly “negotiated” by physician and patient (see Supplemental Online Appendix 4). Dr. Nijhof does not directly suggest massive changes, but she asks whether Mr. de Boer—portrayed by the same actor as in case 2—can take a few days off from his work as a taxi driver in connection with the prescription of new drugs (7). After the SP expresses reluctance with this plan (8), the physician emphasizes that this is a short-term preventive measure in case his blood pressure drops and he becomes dizzy (9–13). The SP again shows hesitation (14–15). Then Dr. Nijhof takes a different approach:

D: (.) When are your days off?

P: (1) Oh (.) normally, in the weekend I do less, but not always. It depends a little bit.

D: So the day after tomorrow is the weekend.

P: Yeah. (.)

D: So it (.) if

P: I think on Saturday I have no schedule in the moment.

D: Okay. Then start tomorrow evening (.) with (1) an I’ll uh give you the recipe in a moment (16–23).

A similar orientation toward compromise through explicit negotiation is observed in later stages of the interaction, too. When Mr. de Boer asks how severe the condition is (35–36), the physician repeats that the SP should reduce his activity, while expressing her understanding that as a company owner (42–46) the SP cannot cease working altogether: “but for example, if you would have been a teacher (1) then I would say reduce (.) your working load” (46–47). With this comparison, Dr. Nijhof reiterates her perspective and recommendation while at the same time showing her understanding of the SP’s existential challenges. By validating the SP’s existential challenges and acknowledging the need to find a compromise between economic and health interests, the physician engages the SP in creating a relationship of trust, where different perspectives are expressed and understood and a compromise is negotiated.

Interaction in this fourth case was classified as optimal during the analysis workshop of the interdisciplinary research group. Observers noted that the two participants negotiate adeptly, almost as if they were bargaining in a market. As with the emotional resonance of the third case, we observed physician and patient establishing the foundations of a trust relationship. This relationship was strengthened over the course of the interaction through a process of give and take but, in contrast to the third case, Dr. Nijhof and Mr. de Boer discussed explicitly the steps to follow. Note that their negotiation did not focus on absolute alternatives (continue working as usual versus stop working altogether). Instead, the negotiation was about “working less”; the question of how much less is not discussed in this first encounter. This variety of SDM shows the importance of deliberation as a complement to active listening.¹¹

Insights

This study confirms a gap between the ideal of SDM and its implementation in clinical practice as well as other more specific research findings, such as physicians’ preference for asking whether a patient had any further concerns at the end of the interaction.⁴⁷ The sample includes experienced experts in addition to physicians early in their career. Physician–patient interaction was observed in diverse settings, similar in that they closely resembled natural healthcare environments but different in terms of educational, cultural, and institutional context. Though our sample is not representative of physicians in the world, our observations expand the geographic scope of research noting an implementation gap for SDM.^13–15

This finding may be discouraging, but it does not mean that SDM in clinical practice is an unrealistic ideal. Many aspects and variations of the SDM ideal were evident in the physician–patient interactions we observed. For example, content analysis showed that many physicians were clearly familiar with the CanMEDS roles of communicator and collaborator and used these roles to explore the patient’s perspective. They asked questions designed to improve their understanding or they suggested explanations for the anxieties they felt the patient was struggling with. They often did so in a routine and not entirely personal manner, but interactional dynamics exemplified in the first case illustrate that even a routine invitation to ask further questions can prompt patients to voice their concerns. This aligns with fine-grained conversation analytical research that points to the importance of conversational gaps and gaze for eliciting discussion of patient concerns.¹⁹

The grounded theory analysis of a subsample of cases in which demands of a family company contradicted health concerns showed much variation and dynamic processes of (non)negotiation. The first case was notable in that while the physician adjusted the care approach based on the responses of the patient, the medical interaction came to a conclusion that the patient did not desire. This suggests that even a mutually participatory decision-making process can lead to an outcome that is inconclusive or even unacceptable to the patient. We can understand the same from a contrasting case in which the physician takes a position in response to unsuccessful attempts at negotiation.

Our findings contribute to discussions about the “true” meaning of SDM. The SDM ideal can be interpreted in an individualistic manner:⁵² Physicians motivated by normative commitments and even a fear of litigation decide to implement SDM to enable patients to make well-informed decisions. But in essence, SDM is a shared process in which both physician and patient contribute. As Kunneman et al. hold, “SDM is … a conversation between clinicians and patients in which they think, talk, and feel through the situation of each patient.”^50(p1323) SDM aspires to contextualize patient situations. In the cases under discussion, contextualization concerned the personal and not only the clinical context of the patient. Also, physician–patient interaction was characterized by unpredictable and unintended social dynamics.

We therefore posit that the core of SDM-related practice must be identified on the basis of observed and realistic physician–patient interactions. With this in mind, we used the multiple perspectives of an international and interdisciplinary team to debate observed varieties of SDM that integrate a patient’s perspective into decision making but that differ from a textbook approach to the SDM ideal. In the third case, Dr. Ou and Mr. Wang interacted in a way that held emotional resonance. This case deviated from the SDM ideal in that alternatives were not openly discussed and decisions were deferred, but the physician was able to “share” the patient’s perspective in a non-cognitive manner. Interaction in the fourth case was deemed to be optimal by our interdisciplinary team. By negotiating a compromise, the physician and patient engaged in a joint process of collaborative dialogue. In both best-practice cases, disagreement between physician and patient remains observable, but the two participants find a mutually acceptable solution.

Limitations

This article presents a secondary analysis on data collected for an earlier sociological study on the globalization of medical knowledge, where analysis had yielded unexpected insights for medical education. We therefore could not control for exposure to SDM education; however, our content analysis together with contextual information and expert interviews suggest that most of the participating physicians were aware of the need to consider the patient’s perspective.

Observing practice with simulated rather than real patients reduces ethical concerns but also has limitations. As SPs learn during reiterations and are experienced in medical education, they may be influenced by educational goals as well as personal ideas about how physicians should interact with a patient. This may explain Mr. Herbst’s repeated questions about continued driving in the first case. That physicians had time and knew they were being observed likely improved their emulation of the SDM ideal.¹⁵ That physicians were aware of the patient simulating the disease might have reduced their efforts to accommodate the patient’s concerns.

Interactions may vary in accordance with the research setting.⁴⁰ Clinicians in our research team viewed recorded interactions and confirmed that the participating physicians took more time in the simulated consultations than is typical for that type of consultation in real life. The cases in the Netherlands may have lost nuance because physicians and patients were asked to interact in English rather than their native language. However, the clinicians of our research team also judged that the style of recorded interactions closely resembles that of nonsimulated practice.

The original study had expected significant cultural difference across the four research settings. As our grounded theory analysis shows, variation in interactional dynamics is a salient finding that may have eclipsed more subtle cultural differences. Studies in very different healthcare settings (poorly funded hospitals, for example) might yield different results and point to different conclusions. Additionally, in each case in our study, the physician and the SP were from similar cultures. Demands for culturally sensitive SDM are not contradicted by our study.^16,53

The quasi-experimental approach of the original study allowed for unexpected findings such as the ones presented in this article. This is in line with other research that rejects an itemized approach to evaluating SDM.^18,19,50 Findings from our grounded theory approach cannot be quantified or standardized, because complex interaction dynamics are, by definition, situated practice.^54,55 Further quasi-experiments could expand the depth and scope of our findings; for example, by systematically relating physicians’ specialty, experience, and education to SDM-related practice. Replications could work with unannounced SPs or simply observe clinical practice.^22,56

Implications for medical education

As standardized patient education and OSCE exams have become more common in the world over the past several decades, their advantages over bedside teaching were groundbreaking. Today, however, some researchers are calling for more holistic methods.^57,58 Our results do not call into question the necessity and merits of skills-focused and standardized approaches to medical education and practice, but we invite educators to integrate some aspects of the comprehensive approach essential to bedside teaching into more advanced methods of medical education.

In a now classic text, Dreyfus and Dreyfus argued that novices follow general rules, whereas experts intuitively recognize patterns given their large, internalized personal library of situation-specific experiences and scripts.⁵⁹ Training medical students to explore the patient’s perspective is an important first step toward implementation of the SDM ideal in practice. In our study, we observed physicians in multiple locations asking questions designed to increase their understanding of a patient’s perspective.

Encouraging students to engage with worked examples of experienced practitioners may be important as a next step toward a more comprehensive approach to SDM. Students may not be able to achieve expert levels of resonance or negotiation right away, but they can discuss and develop a vision of physician–patient interactions that they cannot necessarily control but can likely improve.

Medical learners may benefit from worked examples showing how other physicians assess and respond to a patient’s perspective. Comparing and contrasting the practice of diverse experienced senior and specialized physicians^$ illustrates the fact that even experts apply standards with a great deal of variation. This is in line with the work of Welink et al., who suggested that supervisors and trainees should take turns conducting consultations, discussing their mutual observations afterwards.⁶⁰ Starting with what is actually done, without judgment for the deviation from the textbook, may compromise the normative clarity of the SDM ideal, but it also enables physicians to address the unique complexity of specific interactions. Arguably, in all educational settings, comparative discussion of observed practice enhances feedback literacy, which is important for continued learning.

Medical educators should invite learners to develop realistic expectations:

• Itemized communication techniques, such as questions to enhance understanding, are relevant, but only as a first step toward employing a more mature process of shared decision making.⁵⁷

• The dynamics of a physician–patient interaction cannot always be controlled or predicted; this does not indicate that the physician lacks communication skills.

Our proposals are in line with approaches to medical education that focus on professional socialization and subjectification rather than qualifications alone.^58,61 A realistic expectation of professional maturity encourages physicians to continually work toward decisions that, over time, become more shared or resonant, or more reflective of negotiation.

SDM in practice differs from the textbook approach to SDM. It happens within complex and dynamic interactions in which even reserved patients play an active role. SDM education should be more accepting of improvization and variation:

• Negotiating solutions communicates the ideal of compromising between different positions and perspectives.

• Physicians and health educators should encourage patients to express all concerns and to engage with the physician as a partner in a shared process of negotiation.

• Outcomes that are not mutually accepted can still be satisfactory to both parties, because satisfaction can result from a process of negotiation in which both sides articulate their knowledge, experiences, needs, and interests as well as listening and responding with empathy to each other.

We have learned in our study that SDM is healthcare par excellence.⁵⁰ And a kind of healthcare that is co-created by patients and physicians worldwide.^62,63

Ethical approval

The research project was reviewed by the Ethics Committee of the Medical Faculty of the University of Duisburg-Essen. The committee concluded that there are no ethical or legal objections to the project.

Author contributors

Conceptualization of the study: AnW, TaL, BeQ, and IlS. Construction of SP scenario: StM, TiN. Participant recruitment and infrastructure: AlM, CaM, ChW, GöW, MeE, and StS. Data collection: AnW, IlS, SaW, BeQ, and TaL. Interpretation and analysis during several data sessions: all authors. Drafting the manuscript: IlS, AnW. Revision and editing: AnW, BeQ, GöW, CaM, SoA, AlM, TiN, CaB, StM, and StS. Funding: TaL and AnW.

Reflexivity statement

The research was characterized by professional, postcolonial, and migration-related hierarchies. The impact of professional expertise and the diverging perspectives of medical and sociological professionals have been detailed in the main text. Postcolonial and migration-related hierarchies might have subtly influenced the observations and interpretations. Within the core team, one author speaks Mandarin, and one speaks Dutch and Turkish in addition to English and German. In the complete team, most authors are White; one core team member and one collaborator are read as Asian and two as (other) ethnic minorities. Three authors had migrated to the country of employment.

Data availablity statement

We confirm that all personal identifiers have been removed or disguised so the persons described are not identifiable and cannot be identified through the details of the story. Within the limits of ensuring confidentiality, our data are available for further research at Qualiservice Bremen.

Acknowledgments

We thank four anonymous reviewers and the editor for valuable arguments and comments. We are grateful for the collaboration of Margret Breunig, Dan Gao, Valentin Gendolla, Guo Fan, Alexander Auth, Elif Kandemir, Ewa Łączkowska, Liu Wenting, Franziska Loos, Gül Ayse Öcal Schiwek, and Bilge Tuncel. Last but not least, we thank the 11 simulated patients and 71 physicians who participated in this study.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research was funded by the German Research Foundation (DFG), Grant Number 394996186.

Notes

# Note that, in 2023, the CanMEDS website includes SDM-related ideals in the role of collaborator, too.

* Because collaborators in Ankara had changed the profession of the male SP from taxi driver to administrator, the subsample (n = 29) did not include interactions with the male SP in Ankara (n = 10).

† We know that both options exist in all of the hospitals, but tendencies vary between locations, with Ankara and Wuerzburg taking angio as the standard and Groningen favoring CT. The less experienced physicians in Beijing mention a variety of next steps but do not focus on angio versus CT.

$ Data from our study are available at QualiService Bremen^27,64–67 for secondary analysis and university education, because 64 of 71 physicians consented to make their data available. Readers are invited to use these or other existing data sets to create worked examples of varieties of SDM.