What Is Shared Decision Making? (and What It Is Not)

Both the practice of medicine and the expectations of patients regarding their care are changing. A point of confluence is in the need for medicine to be more patient centered and in the need for patients to be more involved in their care.1, 2 This confluence is particularly pertinent when more than one reasonable approach is available to manage the patient's situation and when those approaches differ in ways that matter to patients.3 In shared decision making (SDM), clinicians and patients work together to understand the patient's situation and to determine how best to address it. Emergency medicine is not exempt from these trends. In this paper we seek to define SDM and its role in contemporary healthcare. Our goal is to set the stage for the active exploration of SDM in the care of patients in the emergency department (ED). Shared decision making is the preferred approach for making fateful healthcare decisions when more than one reasonable option is available. While elements of SDM can be often seen in routine consultations, formal implementation of SDM is likely hampered by the lack of clarity on what constitutes SDM.4, 5 To help participants in the 2016 AEM Consensus Conference clarify what they meant when they talk about SDM, some alternative definitions were offered for debate. Participants quickly discarded SDM as a variation on the informed consent process. Attendees also rejected the idea that SDM was a formality involving listing options only for the clinician to recommend or “order” one. Presenting evidence-based information to the patients so that they can decide what to do was considered neither sufficient nor desirable, as this would lead to unrealistic workload and responsibility for patients. In some cases, “it's your decision” may make patients feel abandoned, rather than cared or empowered. On the other hand, patients' role should not be limited to voicing their preferences, only for the clinician to decide what to do. Indeed, it is helpful to remember that preferences evolve in the deliberation process, in considering the situation and the options available. Thus, they usually do not exist ahead of the consultation and cannot be assumed to be similar to that of other patients or to be stable over time. The implementation of patient decision aids, tools that patients use on their own to form preferences or make decisions, was not considered to be equivalent to SDM by the attendees; these don't necessarily include the “dance” inherent in SDM. Instead, attendees reached consensus that SDM is best described as a conversation between the clinician and the patient in which they figure out together what to do to address the patient's situation. Ways to improve the care of patients include new treatments tested in randomized trials, practice guidelines based on this evidence, and quality or performance measures that narrow variation. Although these efforts are helpful in caring for people like this patient, they could fail in caring for this unique patient. That is because a specific patient is particular in terms of the applicability of the research evidence to their situation (i.e., this patient should expect different rates of beneficial or harmful effects compared to patients in published studies); the values, goals, experiences, and preferences the patient has at the time of determining what to do; and the personal and social context (e.g., mental and physical comorbidity, poverty, isolation) in which the patient experiences disease and implements treatments. SDM is a strategy to care for this patient. All these elements contribute to arriving at the best approach to address the patient's situation. These are not items on a checklist or instructions on a recipe, but rather iterative and interactive steps in a conversational dance. Patients do not consult clinicians just seeking information or choice. Patients seek care. What is needed to care for patients is that clinicians and patients think, talk, and feel through the situation of the patient in a diagnostic conversation.7 In this conversation, the parties determine the aspect of the patient's situation that demands action and the actions that the situation demands. When reasonable actions are identified, they represent hypotheses that should be tested against the situation in the conversational dance between the patient and clinician. We illustrate the diagnostic conversation with the following example. A 55-year-old male developed chest pain after lunch today at 12:30 pm shortly after returning to his construction job. He described the pain as a pressure sensation that was central in location; nonradiating; not associated with shortness of breath, nausea, or diaphoresis; and spontaneously resolved within 30 minutes. He says the pain did not get worse when exerting himself on the job. He denies fever or cough, and the pain is not worse with deep breathing. He has no history of diabetes or coronary artery disease (CAD), although his father did have a heart attack at 45 years of age. He is a nonsmoker, active, and healthy. Given that he did not want to lose money by taking off of work, he presents to the ED at 7 pm in the evening after his work day. He does not appear to be acutely ill; his vital signs are within normal limits; he has no abnormal findings on physical examination; the initial electrocardiogram (ECG) shows a normal sinus rhythm with nonspecific ST changes; and the initial troponin, chest x-ray, complete blood count, and electrolyte panel are all within normal limits. Your assessment is that the patient's nontraumatic chest pressure could potentially be due to an acute coronary syndrome (ACS) but that the symptoms are not classic and his initial workup is negative. However, given his family history of premature CAD, age, and sex, your gestalt is that he is at low risk for ACS. You calculate his HEART score8 at 3 based on the nonspecific ST changes on ECG, his age, and positive family history for CAD, putting him at 0.9%–1.7% risk for a cardiac event at 6 weeks. In your hospital, one troponin at least 6 hours after pain onset is required, and 6.5 hours have passed since the onset of the patient's chest pain, so no additional troponin testing is needed to rule out acute myocardial infarction. You download a decision aid developed for this scenario, Chest Pain Choice,9, 10 and discuss with him his risk of a heart attack or pre–heart attack in the next 6 weeks, the options of staying for stress testing, which would require an overnight stay at that time of day in your hospital, or following up with a cardiologist at your hospital or his family physician for consideration of further testing within 72 hours. You are comfortable with either option, given your risk assessment. He chooses to follow up with his primary care physician, as he is concerned about the cost of an overnight stay and does not want to miss work tomorrow. You arrange outpatient cardiology follow-up as a back-up in case he is unable to see his family physician within 3 days and discharge him home from the ED. To support SDM, an increasing number of decision aids have been developed. In most cases, these aids are used outside the clinical encounter and target either the patient or the clinician.11 These tools can be valuable to prepare patients and clinicians for the encounter and the decision to be made, but their design does not necessarily support the conversation between patients and clinicians. Over the past decade, our group has developed conversation aids designed primarily to support the dialogue patients and clinicians must establish to determine the patient situation and the actions required to address this situation.12 Their emphasis has been on supporting the identification of available reasonable approaches and, from them, the one that best addresses the patient's situation. This journey started with the Statin Choice tool, which features a risk calculator and renders a 100-patient pictograph that conveys the 10-year risk of cardiovascular events without and with statins.13 This tool is carefully designed to avoid framing effects (e.g., presenting only gains or losses) and to offer graphical, numerical, and verbal forms of risk communication. A randomized trial demonstrated its efficacy in clinical encounters,14 but, more importantly, video recording of intervention and control encounters revealed an unintended difference between them. In intervention encounters, the conversations created went well beyond the notion of normalizing or optimizing cholesterol levels or reducing cardiovascular risk, instead leading to moments of deep human connection. These moments often resembled a dance in which patients and clinicians synchronized their nonverbal communication while discussing what was best for the patient (see Figures 1A and 1B). A few years later we developed another form of conversation aid, which focused on selecting from a large number of diabetes medications the best one for the patient's situation.15 The tool facilitated communication by helping clinicians ask patients for the most salient issue for them that distinguished the available options. Then, patients navigated two or three of these issues, presented as cards, to arrive at the best option for their situation. Again, a randomized trial in practice demonstrated their efficacy and, again, video recordings found an unintended consequence of supporting these otherwise technical or even mundane conversations:16 patients and clinicians were more likely to connect emotionally with each other and even laugh together. A particular patient remains salient in our mind. He was very elderly, had recently lost his wife, and had moved to an assisted living facility. As he was considering the options available to control his diabetes, he reported being interested in losing weight and using an injectable agent to do so. The reasons: “that place is filled with women” and “the nurse comes twice a day to give you the injection.” Patient and clinician, who knew each other for over a decade, went from dejected to cheery to laughing out loud. More recently we have developed a new generation of tools that continue our commitment to support diagnostic conversations in the clinical encounter by combining the above-mentioned risk communication and issue cards. This is the case of our anticoagulation in atrial fibrillation tool. While this tool is just now undergoing empirical evaluation in a randomized trial,17 early testing reveals that it can produce conversations in which the care is not just for people like this patient. Rather, by bringing the situation of this patient into sharp focus and detail, the care rendered is the best for this patient. When we started our work a decade ago there were some almost axiomatic beliefs about SDM: that this will be an approach used exclusively by specialty clinicians with long consultations with well-off and relatively young patients. It was also believed that SDM and the use of conversation aids would face clinician opposition, require extensive training, and consume substantial consultation time; in other words, that SDM tools would not be used. Our experience has been quite different. Using user-centered design, we developed tools and approaches with the users and in the context of their consultations. This led to the development of interventions that fit well in the busy contexts of specialty and primary care, hospital and ambulatory care, and even emergency medicine.9 Most of the work18-21 involved older patients (mean patient age 65 years, range = 20–95 years) and, as hoped, 74%–90% of clinicians, depending on the clinical context, preferred to use the conversation aids they were testing with their next eligible patient. This approach on average added 3 to 4 minutes to the consultation time, but the range was wide. In the context of ED care, investing this time may not always be feasible given concurrent high-acuity patients, but doing so when possible provides the opportunity to personalize care, improve the patient experience of and satisfaction with care,22 and potentially obviate an overnight observation unit stay,9 thus decreasing the downstream workload of the entire clinical care team. Clinicians, on average, correctly used 6 of 10 key features of conversation tools without extensive training. Also, the use of these tools promoted the acquisition of 15%–20% more knowledge about the options and similar increases in patient involvement in the consultation, all factors associated with better SDM (see Figures 2A and 2B). We were concerned about the effect on vulnerable populations, but we found that patients with limited education and low socioeconomic status benefit to the same or a greater extent than more fortunate patients. The effects on clinical outcomes, health care utilization, adherence to medications, and costs have been variable. SDM has been touted as way of improving these outcomes and the value of healthcare. In turn, this expectation has supported policies promoting SDM. Our findings, however, suggest that these justifications for SDM are not consistently supported by the extant evidence. Widespread belief that SDM tools often sit on shelves has also been weakened by our experience: more than 12,000 uses of our aids are logged online every month, with accesses from every continent in the world, and every state of the United States. Decision and conversation aids can be valuable in facilitating SDM, but they are neither necessary nor sufficient for choosing an approach to address each patient's situation. The purpose of SDM is not to merely implement elements as choice awareness, information giving, and preference elicitation in a mechanical way, to formally “tick the SDM box.” The underlying goal of SDM is to fundamentally care for the patient in a manner that resolves each person's situation by virtue of its effectiveness, consistency with what matters to this person, and fit with the contexts in which treatment and condition play out. In our view, SDM is a process, a conversation between clinicians and patients in which they think, talk, and feel through the situation of each patient. Evidence-based options are hypotheses to address the situation, which are tested in the conversation until the best solution becomes clear. In essence, SDM is a human expression of care that is careful and kind.