• unannounced standardized patient
• doctor-patient interaction
• direct observation
• assessment and feedback
• activation and empowerment

Evidence suggests that the quality of a doctor’s communication, including non-verbal interaction, data-gathering skills, levels of empathy, ability to summarise and clarify, information sharing and interactive patient educational strategies, is associated with positive patient health outcomes.1–4 In this issue, Amelung et al 5 contribute to this evidence using observational data combined with in-depth qualitative analysis to explore how misalignment and misunderstanding in the doctor–patient interaction can lead to negative ‘interim’ outcomes critical to patient safety. This accumulation of evidence provides even clearer targets for the education of physicians. Our medical education research group has called for the identification and definition of Educationally Sensitive Patient Outcomes (ESPOs)—those interim outcomes that can be maximised through education and training of physicians and that are critical to ultimate health outcomes.6 7

Having the skills to ensure that a patient is fully informed and activated to act in his or her own best interest is an ESPO—an outcome directly attributable to physician practices, at least in part, that we as educators can teach and measure. Amelung and colleagues found that a failure to achieve consensus at the end of the care visit often manifested as a ‘false’ sense of concordance between physician and patient, leading to lack of patient follow-through and/or dissatisfaction.5 This finding illustrates the critical importance of patient education, an essential aspect of the Calgary/Cambridge model8 that often gets short shrift in the broader communication literature. Teach-back is the simplest and most commonly used core skill in patient education. It is effective in creating dialogue that facilitates trust, shared understanding, accurate information gathering and most importantly patient activation—patients who are empowered to and engaged in actively managing their health.6 9

Our work, using standardised performance-based assessment of medical students and residents, has consistently shown that (1) patient education skills such as clear explanations and teach-back are much weaker than communication skills in other domains (eg, relationship development and information gathering), (2) that they improve with medical school training (more so than the other communication domains)10 and (3) they may decline in the pressured atmosphere of actual practice during the early years of postgraduate medical training residency (ie, postgraduate years 1 and 2 in North America, and foundation years 1 and 2 in the UK).11 And while the ability to educate patients is necessary to achieve positive health outcomes, we hypothesise that it is not sufficient—physicians should also strive to ensure that patients are activated in their own self-care. Our preliminary research, using patient activation as an interim outcome in standardised patient assessments, suggests that although a trainee’s ability to activate patients is moderately correlated with communication skills, it is also a distinct capacity.12 13 And we believe patient activation is key—knowing that at the end of the visit, the patient understands their situation, knows what to do about it, feels confident about what they can do (perceived self-efficacy) and has established a partnership with the physician. These end-of-visit practices are essential but are the most challenging aspect of core communication skills to master.14 Asking the patient to explain next steps or what they understood, also known as ‘teach-back’ or ‘ask-tell-ask’ is frequently left out, even by learners trained in programmes who pride themselves on their communication curriculum.

Patient empowerment, or increasing the capacity for patients to make choices about their health and act on these choices in an effective way, is also a key to successful health outcomes.15 Amelung and colleagues vividly illustrate that physician–patient interaction is a two-way, complex interaction that demonstrates the fluidity and complexity of the interaction between two people. Educating patients to understand physician perspectives, the structure of the interview and why they need to be honest and provide all information, and that they have every right to ask all their questions and state preferences, has the potential to improve health outcomes in synergy with and beyond even what we can do with physicians’ education.16

So, how do we integrate this growing understanding of what must happen during physician–patient communication to ensure quality care? First and foremost, we need to understand the underlying mechanisms of the patient–physician interaction. Second, we must train physicians to recognise when they are not achieving alignment or trust and how to adapt to this knowledge and improve. Most importantly, we need to expand our assessment of communication skills to focus on educating, empowering and activating patients. Physicians need more training in and feedback on how to explore and influence patient beliefs, expectations and understandings. What the Amelung article does not tell us is what the physicians’ reactions were to the results. Would they be surprised that their patients did not feel heard? Did they sense the lack of concordance? Do the findings help them understand why their patients might fail to adhere to their recommendations or how this would lead to negative outcomes? What are they going to do with this information?

This study makes an important contribution by suggesting a link between poor outcomes (ie, missed or delayed diagnosis, failure to adhere to treatment, or failure to provide accurate information about severity of symptoms) and the physician–patient interaction. We are now challenged to prevent continued misalignment. The authors suggest eliciting patients’ beliefs and expectations prior to the visit. However, the study showed that those beliefs and expectations are dynamic, suggesting that it is not sufficient to assess them once. Focusing on how to continuously, efficiently and dynamically elicit and influence patient beliefs throughout the interaction should be a fundamental component of patient safety and outcomes research. Physicians must be attentive to contextual cues and assess misunderstanding, disagreement and mistrust.17 Teach-back is one such strategy.

Right now, most physicians do not get sufficient training in patient education or activation (except perhaps in targeted motivational counselling efforts), and in most cases, training ends at completion of medical school or residency. And since physicians are often unaware that there is a misalignment in goals or expectations with their patients without ongoing feedback, there is no way to improve communication, ensure congruent goals and therefore maximise care quality and safety.

This article demonstrates, and we agree, that the best way to assess physician skills is through direct observation in the clinical microsystems where they occur.11 Our team and others have developed sophisticated methodologies of in situ direct observation, assessment and feedback using highly trained unannounced standardised patients (USPs), or ‘secret shoppers’.11–14 18 USPs are skilled actors who consistently portray a standardised clinical scenario ‘incognito’ in healthcare settings and have been used, for example, to characterise prescribing practices.19 The USP method overcomes many of the limitations of commonly used assessment methods (eg, patient surveys, chart reviews, billing data) including the observational data used in the Amelung study by capturing the full experience of the doctor–patient encounter/partnership in the real-world context using highly trained and calibrated assessors with structured measurement tools and controlling for the patient and clinical characteristics that can confound accurate assessment.20–22 USPs are thus a compelling approach to systematically assessing physician skills in the context of challenging, real-world patient interactions and authentic contexts.

The data and insights provided by USPs allow us to provide in-depth, highly specific feedback on physician practices and aspects of the patient–physician interaction that can lead to poor quality care and negative patient outcomes. We design USP cases to be representative of the patients cared for in that local system and tailored to provide appropriate clinical challenges—and then give this feedback to the physician. While an actual patient might have left the encounter with an incomplete understanding of her condition and no intention of following up on the diagnostic referral, the USP identifies where the communication breakdown happened and what the clinician can do differently to avoid this outcome in the future—from the patient’s perspective. This approach reinforces the critical importance of an educated and activated patient to physicians and provides actionable feedback.

A learning healthcare system is defined as one in which there are cycles of assessment (problem definition), implementing solutions and determining the effectiveness of those solutions. USPs can play a critical role in fully realising the benefits of a learning healthcare system by ensuring that those aspects of the patient–physician interaction and team interactions that are critical to patient safety are assessed in a meaningful way and routinely fed back to all the individuals responsible for clinical system improvement.23 When done with respect and subtlety, data from USPs become core to conducting a root cause analysis when a clinical process needs to be improved. Work such as Amelung and colleagues’ provides important targets for quality and safety in primary care and other ambulatory settings because it establishes the importance of foundational communication skills to achieve safe care and best possible outcomes. We advocate for building ongoing assessment and training cycles around these essential skills as a critical component of every institution’s patient safety agenda.

Remember—this study involved physicians and patients who knew they were being observed and assessed and therefore were presumably demonstrating their best practice. Imagine how the rates of misalignment might be different for typical encounters in a busy practice.