Teaching the science of uncertainty

Introduction

Uncertainty is ubiquitous in clinical practice [1, 2], so clinician educators should strive to better define this entity [3, 4] and design, implement, and evaluate curricula addressing clinical uncertainty. Although frequently encountered, uncertainty is often implicitly referred to rather than explicitly discussed [5]. Increasing explicit discussion of – and comfort with – uncertainty has the potential to improve diagnostic reasoning and accuracy and improve patient care, for example, by explicitly discussing differential diagnosis, discussing probabilities during handoffs, and implementing inductive reasoning while teaching [6]. Discussion of both diagnostic and prognostic uncertainty with patients is central to shared decision-making in many contexts as well [7], from the outpatient setting to the inpatient setting, and from undergraduate medical education (UME) trainees to graduate medical education (GME) trainees. In this article, we explore how the science of uncertainty is taught within UME and GME curricula and we describe strategies to explicitly discuss uncertainty with all levels of learners.

Teaching the science of uncertainty at the UME level

As learners transition from the pre-clinical environment to the clinical environment, they gain skills and knowledge through clinical exposure. The transition from the “classic” cases depicted in textbooks to grappling with the nuances of real-life clinical uncertainty may be jarring to novice learners. When medical students face uncertainty in their clinical environment, it may lead to increased stress, particularly given their desire to avoid mistakes [8, 9]. The desire towards perfectionism may provoke anxiety and detract from learning and improvement [9].

Medical sociologist Dr. R.C. Fox encountered and studied three types of medical student uncertainty [10]:

1. Acknowledging limitations in current medical knowledge

2. Incomplete or imperfect mastery of available knowledge

3. Difficulty in distinguishing between personal ignorance or ineptitude, and the limitations of present medical knowledge.

According to Fox’s observational studies, tolerance for uncertainty develops for medical students throughout the entire training process [11]. Lee et al., suggest clinical uncertainty is defined as a dynamic interplay between sources of uncertainty, subjective influencers of uncertainty, and one’s response to uncertainty [3]. Defining uncertainty in this manner allows for educators to contextualize uncertainty to learners across learning environments and enforce the dynamic and iterative nature of clinical uncertainty.

While understandable, student discomfort with the concept of uncertainty in clinical care has consequences. Difficulties tolerating uncertainty can impair medical students’ clinical skill development by multiple mechanisms, including poor clinical decision-making [12] or overuse of low-value care [8]. Merrill and colleagues measured medical students’ aversion to uncertainty while assessing students intolerance to ambiguity [13]. Intolerance to ambiguity is defined as “the tendency to perceive situations that are novel, complex, or insoluble as sources of threat [14].” Merrill noted an aversion ot uncertainty in clinical medicine as the best predictor of negative attitudes toward hypochondriac, geriatric, and chronic pain patients [13]. Another study suggested that medical students with decreasing intolerance to ambiguity impacted willingness to work with underserved communities [15]. Furthermore, clinicians’ inability to tolerate uncertainty may lead to poor communication, especially during conversations about shared decision-making [16]. Lastly, medical students who display personality traits suggestive of an intolerance of ambiguity and are overly concerned about making mistakes may demonstrate maladaptive perfectionism, making them vulnerable to stress and poor coping [9], which could ultimately lead to worsened clinician burnout.

Although this research demonstrates the need for UME curricula to introduce clinical uncertainty with learners and teach strategies to tolerate uncertainty, the optimum timing and type of curricula is also undetermined. An effective curriculum should ideally address students’ needs, developmental and emotional maturity levels, prior experiences, current knowledge, and current clinical skill levels.

So far, educational initiatives to address clinical uncertainty with medical students have focused on recognition of uncertainty, its impact on clinical reasoning, and how to communicate uncertainty. For students to recognize uncertainty, cognitive interventions have employed simulation training [17], reflective journaling [8], and use of non-clinical tools (such as online decision aids) [18], and use of the humanities, such as art [19] and philosophy [20], [21], [22]. The iterative clinical reasoning process from diagnostic reasoning to management reasoning is prone to uncertainty. Interventions to manage uncertainty throughout this process include the use of clinical decision-making aids [23], use evidence-based medicine [24], decisional analysis [25] and implementation of case presentation using an advanced organizer tool that incorporates recognition and response to uncertainty [26]. Ultimately, medical students must develop the skill of communicating uncertainty to fellow clinicians, patients, and interdisciplinary care team members. In proposing a four-step model to communicate and teach uncertainty, Santhosh et al. recommended a framework for students to explicitly acknowledge the differences between diagnostic accuracy and diagnostic uncertainty [27], which also allows for more informed discussions of medical prognosis [28]. Communication with patients and families during shared decision-making conversations employs the techniques of acknowledging diagnostic uncertainty as well [29].

Another approach is to target interventions to counteract ways in which UME curricula train students for improving clinical confidence and certainty in their diagnoses. Pre-clerkship medical education promotes certainty through multiple choice testing and diagnostic reasoning curricula focused on development of intuitive thinking, namely illness scripts and pattern recognition [30, 31]. Educational curricula in the clinical environment reinforce these concepts via diagnosis-driven didactic sessions and limited physician-led discussions of management uncertainty [31]. To introduce uncertainty into these learning environments, interventions include use of examination questions with multiple correct choices [32] and clinical vignettes with multiple correct management scenarios [33]. Recently, Papanagnou et al. incorporated clinical uncertainty into their pre-clinical medical education case based and humanities curriculum. They evaluated how these interventions prepared clerkship students for uncertainty and found the most impact from storytelling, communication role play, simulation and clinical debriefs [34]. Importantly, they noted similar impact from peer-to-peer conversations and instructor vulnerability that occur when instructors share personal narratives that normalize and role model how to respond in times of clinical uncertainty.

Teaching the science of uncertainty at the GME level

Clinical uncertainty is prevalent at the GME level among residents and fellows alike. It arises frequently during clinical reasoning and when communicating with patients and families [2]. In addition to lacking universally standardized mechanisms for teaching the science of uncertainty to trainees of all levels, literature suggests that the hidden curriculum of medical training may discourage expression or acknowledgement of uncertainty [6]. Just as at the UME level, residents and fellows are often encouraged to give definitive diagnoses and propose definitive plans on rounds. Expressing uncertainty may be viewed as a sign of a poor or weak clinician.

Diagnostic uncertainty in critical care and emergency medicine settings is apparent when one encounters an undifferentiated critically ill patient. Moreover, diagnostic uncertainty is also encountered in the cognitive specialties, such as oncology, rheumatology and infectious diseases [35], [36], [37]. Recent qualitative studies in emergency medicine residents have shown that the majority of teaching of uncertainty at the GME level happens informally in the work environment, primarily through role modeling by attending physicians [38]. Most teaching of uncertainty across specialties occurs in this informal way and is contingent on the comfort of faculty with this topic.

Current practices for teaching the science of uncertainty in GME focus primarily on explicitly stimulating discussion of diagnostic reasoning processes with trainees [39]. Several authors have described the practice of asking open-ended questions (e.g., “How are you thinking about this patient’s anemia?”) during rounds to encourage residents and fellows to outline their thought process and ultimately identify areas of uncertainty [7, 39, 40]. The “diagnostic pause” is a diagnostic reasoning strategy that allows a clinician to explicitly and intentionally discuss uncertainty as a team. Both facilitated discussion and diagnostic pauses serve as cognitive forcing strategies that can help overcome cognitive error and bias [41].

A survey of emergency medicine residents showed that 51% had a strong desire for additional training in communicating diagnostic uncertainty [38]. Improving comfort with uncertainty going forward could be improved by adjusting the culture of graduate medical education to demonstrate greater acceptance of uncertainty as well as formalize mechanisms to identify, name, and discuss uncertainty with trainees, colleagues, and patients.

Incorporating the concept of uncertainty more explicitly into diagnostic reasoning education at the GME level facilitates a greater acceptance of uncertainty as a whole and increases standardization across residency and fellowship programs [6, 27]. Relying on attending physicians to introduce the concept of uncertainty via open ended questioning or facilitated discussion on rounds leads to variability of practice. Structured frameworks to explicitly discuss uncertainty can be helpful to consistently promote discussion of this topic. Such frameworks could be beneficial during patient handoffs as well, to clearly communicate areas of uncertainty in diagnosis or management when care is transferred between providers. Tools such as the ICU-PAUSE framework developed for discussing uncertainty at the transitions of care from the intensive care unit to acute care ward [42] and the SNAPPS framework [26], initially developed for outpatient precepting with medical students, encourage discussion of diagnostic reasoning and uncertainty. Creating a similar framework for the inpatient or critical care environment may serve this purpose [43].

Hayes et al. propose teaching inductive reasoning in the intensive care unit rather than deductive reasoning as a method to think more broadly about problems and avoid overreliance on memorization [40]. Another method they recommend is the use of conceptual diagrams to illustrate the pathophysiology of problems. When drawn, these diagrams illustrate areas of uncertainty that exist in the patient presentation and working diagnosis.

Other more structured curricula that focuses on communicating diagnostic uncertainty to patients have been developed and well received by residents in the emergency medicine settings [44]. Finally, just as trainees receive feedback after leading goals of care conversations, implementing feedback after discussion of uncertainty with patients may be a way to improve comfort with uncertainty itself as well as communication to patients and families.

Teaching the science of uncertainty at the CME level

Lastly, practicing clinicians, just like learners, also practice in an environment where they frequently grapple with uncertainty. Faculty have to consider how to teach these concepts to trainees and communicate uncertainty with patients. There is a paucity of literature regarding teaching concepts of diagnostic or management uncertainty to practicing physicians. Many institutions have local venues such as morbidity & mortality conferences or challenging case conferences for practicing clinicians to bring cases where there is significant diagnostic or management uncertainty to seek expert input from colleagues. One study of antibiotic prescribing among clinicians noted that diagnostic uncertainty was frequently associated with antibiotic prescriptions [45].

The Society to Improve Diagnosis in Medicine has multiple workshops and published curricula for practicing clinicians to focus on diagnostic error, including discussions about diagnostic uncertainty [46]. Similarly, the Society of Hospital Medicine has a continuing medical education (CME) module using learning cases to improve performance [47]. Engaging trainees in any of these venues or established curricula could improve the culture of more explicitly acknowledging and discussing uncertainty, and thus improve just-in-time teaching about this topic.

Common themes across the medical education continuum

When examining the teaching of the science of uncertainty across the spectrum of medical education continuum, common themes emerge (Figure 1):

Figure 1:

Teaching uncertainty across the continuum.

Conclusions

In conclusion, diagnostic uncertainty, management uncertainty, and prognostic uncertainty are ubiquitous in clinical practice, so training learners across the worlds of UME, GME, and CME is imperative. Although there are few published curricula documenting systematically how to best accomplish this task, in this article, we have outlined the limited evidence to date and delineated strategies to teach the concepts of uncertainty across the continuum of medical education. Educators can help learners at multiple levels recognize uncertainty, manage uncertainty, communicate uncertainty (to patients and families), normalize uncertainty, and explicitly discuss uncertainty to improve teaching on this topic in multiple modalities. Further research can help delineate what the most effective and desirable curricula to best address this complex topic, with the ultimate goal of improving diagnostic safety for patients.