Beyond Competency: Setting a Path to [Diagnostic] Expertise

 From the. 1/22/2021 newsletter

Perspective/Opinion

 

 

Beyond Competency: Setting a Path to [Diagnostic] Expertise

 

 

By Jayshil Patel, MD and Paul Bergl, MD

 

 

Drs. Patel and Bergl, who have a keen interest in “how doctors think,” discuss their curriculum to improve clinical reasoning skills …

 

Think about the last time you or someone you know went to a doctor with a symptom.  What did you or they seek?  Did you want an explanation for the symptom?  Did you want an efficient, yet thoughtful evaluation in arriving at a diagnosis?

 

Accurate and timely diagnosis is the foundation of medicine.  With it, management strategies have the best potential to positively modify disease and outcome.  Without accurate diagnoses, however, management is unguided and both potentially wasteful and ineffective. Thus, we advocate for undergraduate and graduate medical education to invest in setting learners on a path towards diagnostic expertise. 

 

Expertise does not stem from a superior natural capacity to analyze new information.  Rather, expertise is an adaptation, rooted in the ability to efficiently recognize patterns and compare it to what has been accrued in an individual’s extensive knowledge domain in their long-term memory.  Expertise requires deliberate practice, a concept coined by psychologist Anders Ericsson to describe the modus operandi of expert development.  Deliberate practice requires systematic and forced attention for refining performance.  Experts set a stretch goal, deconstruct its components, and hone narrow aspects of their performance until achieving mastery. 

 

 

Deconstructing the “diagnostic process”

 

Fortunately, the diagnostic process can be deconstructed, and its steps deliberately practiced, thereby allowing the practitioner to remain on a pathway to diagnostic expertise. Unlike chess, music, or individual sports, however, measuring expertise in medicine is challenging.  Yet if we assume that consistency, efficiency, tolerance for uncertainty and ambiguity, and adaptability are key features to any form of expertise, then enhancing diagnostic performance would be aided by a rich understanding of the diagnostic process and mastery of its components.  With systematic and forced attention towards elements of the diagnostic process, knowledge (within the limits of working memory) becomes amenable to processing and reorganization into more meaningful units called “chunks.”  

 

When new information arises, the working memory rearranges its components into a coherent cognitive representation by connecting and cross-referencing this emerging knowledge with established “chunks” of information already stored in long-term memory, ready for rapid retrieval into working memory.  Not only do expert diagnosticians possess extensive knowledge, but through deliberate practice, they store this knowledge in well-organized schemata in the rapidly accessible long-term memory.   

 

Consider the alternatives and consequences to deliberate practice.  Learners (or practicing clinicians) may “practice” medicine by logging thousands of hours seeing patients; however, without explicit knowledge of the diagnostic process or well-defined goals for the steps of that process, they are at risk of repeating habitual tasks and on the pathway towards arrested development.  Without coaching or reflection, unguided learners (or practitioners) may not recognize failures, and they may default to an intuitive mode of thinking, which when overutilized, is a key driver of diagnostic error. 

 

 

The mechanics of a diagnostic reasoning curriculum

 

Diagnostic reasoning curriculum ought to change learners’ attitudes of their growth potential, build knowledge around the language and science of reasoning, and enhance specific skills through deliberate practice and reflection.  Implementing deliberate practice across the training spectrum requires a fundamental shift in how we teach diagnostic reasoning in today’s complex clinical learning environment.  Learners may perceive diagnostic reasoning as an intimidating black box and expertise as unattainable.  For attitudinal change, we suggest explicitly defining and calling out the stretch goal, namely, to create expert diagnosticians.  

 

Changing language to include phrases like “expert development” and illuminating a deconstructed pathway to expertise may foster a growth mindset, one in which learners peer into the black box of diagnostic expertise and gain intimate access to its many inputs and outputs.  We suggest a vision for medicine training be to gain the foundational diagnostic knowledge and skills to independently care for patients while deliberately practicing them toward expertise.

 

To build knowledge, the stretch goal should be deconstructed into discreet, teachable components.  In our curriculum, we have deconstructed diagnostic reasoning into: 

 

[a] the semantics of the diagnostic process

[b] the science of thinking, learning, and decision-making

[c] mechanisms for reflecting and strategies which may enhance diagnostic accuracy

 

Before applying their knowledge, experts in non-medical fields like chess, tennis, and art learn and master a common language.  Likewise, before applying medical knowledge, students master the language where words like dorsum, ipsilateral, and morbilliform codify concepts we all understand.  Similarly, the steps and scientific concepts of diagnostic reasoning include well-defined terms like problem representation, illness script, diagnostic schema, dual process theory, and cognitive load.  We suggest educators and learners first master the language of clinical reasoning.  Without this language, we create competency scales that deem trainees ready for unsupervised practice if they merely “synthesize data to generate a prioritized differential diagnosis and problem list.”  Oblique references to diagnostic reasoning are confusing.  Furthermore, without a shared dialect, we cannot apply diagnostic reasoning knowledge nor expect learners to reflect or teachers to coach.

 

Next, teaching the science of thinking is crucial to optimize reflection, feedback, and clinical decision making.  In our curriculum, we differentiate normative from descriptive modes of decision-making by illustrating Bayesian principles and outlining System 1 and 2 thinking, respectively, using medical and non-medical examples.  Describing concepts like cognitive load and cognitive bias may enable trainees to recognize methods to optimize learning and limitations of human thinking, respectively.  Furthermore, the advantages and disadvantages to various types of reasoning can increase self-awareness and consciousness of the mode employed, empowering trainees to better calibrate their own thinking over time.  Faculty development in the science of thinking and learning may promote their ability to, for example, recognize a scenario where learners (or their teachers) may be experiencing high cognitive load, to then take the steps to minimize it by enhancing the clinical learning environment.

 

Explicitly granulating components of the diagnostic process provides targets for feedback and self-reflection.  For example, accessing and selecting illness scripts are steps in making a diagnosis.  Illness script components include epidemiology, pathophysiology, symptoms and signs, diagnostics, and response to treatment.  Our (unpublished) research shows almost all novice learners recall signs and symptoms of disease but often lack working epidemiological knowledge or pathophysiologic insults.

 

 

A clinical example

 

Consider a situation where a novice learner working in an intensive care unit identifies a patient with new thrombocytopenia.  The learner mentions the patient was receiving heparin and orders tests to evaluate for heparin-induced thrombocytopenia without mentioning its epidemiology or considering alternative etiologies.  Two days later, the patient was found to have disseminated intravascular coagulation.  During a feedback session, one form of feedback might be: “Read more about thrombocytopenia.” 

 

But, if the community of learners and educators speak a common language around diagnostic reasoning and understanding the science of thinking and learning, feedback may be transformed from a nebulous, “Read more about thrombocytopenia,” to  “I’d like you to work on the epidemiologic component of your illness script for heparin-induced thrombocytopenia and develop a pathophysiology-based diagnostic schema for thrombocytopenia.”  

 

 

Feedback tethered by the language and science of the diagnostic process has numerous benefits.  By delineating components of the diagnostic process, educators and learners can better assess diagnostic performance through targeted feedback, and in turn, deliberately practice towards, in this example, enriching an illness script.  From the perspective of an educator, awareness of what constitutes an illness script led to recognition of an incomplete script (lacking epidemiologic knowledge for heparin-induced thrombocytopenia).  By having knowledge of “how we think,” educators and learners can have a conversation to metacognate and identify cognitive bias leading to potential diagnostic error.  

 

In our example, the learner did not consider alternative etiologies for thrombocytopenia and anchored onto a diagnosis of heparin-induced thrombocytopenia.  As a result, educators construct remediation plans.  The learner was advised to develop a schema for thrombocytopenia (i.e., form a systematic approach to a clinical problem).  Consequently, longitudinal follow-up is established.  

 

On subsequent interactions, the educator can assess if the learner, indeed, developed a schema for thrombocytopenia. From the perspective a learner, such targeted feedback is constructive and actionable and serves as both a tool and a metric, in this case, to enrich the heparin-induced thrombocytopenia illness script and deliberately practice thrombocytopenia schema formation.  Importantly, the learner has a framework to self-reflect on how, why, and which cognitive bias(es) was invoked. 

 

 

Opportunities for incorporating clinical reasoning education into the clinical learning environment

 

To enhance reflection, skills, and reinforce effective habits for expertise, we suggest creatively infusing opportunities to deliberately practice components of the diagnostic process into our fast-paced, complex clinical environments that are fraught with actual or perceived barriers, like hand off medicine.  We advocate for dedicated undergraduate and graduate medical education didactic sessions that teach the language and science of diagnostic reasoning.  Morning report, morbidity and mortality conference, and clinical-pathologic correlation conferences could serve as ideal venues to deliberately practice diagnostic reasoning concepts and reap the benefits of crowdsourcing.  

 

As we have observed in our curriculum, infusing the language of clinical reasoning in one venue will invariably lead to utilization in other venues.  On the wards, when housestaff hand-off their patients, the face-to-face handoff period is an opportunity for a second opinion for the person giving the hand-off.  For the individual receiving it, it serves as an opportunity to practice diagnostic reasoning components, including refining problem representation as new data trickles in before rounds.  During rounds, we propose asking learners for why and how a diagnosis was ascertained, as opposed to just what. 

 

Democratizing rounds and asking all learners, not just the one presenting, promotes group discussion and creates a clinical learning environment where all learners can be empowered to think aloud.  For educators, asking learners to reason aloud promotes active reflection and generates opportunities for coaching and critical appraisal of their diagnostic reasoning.  Ideally, coaching would be longitudinal and intensive but, as demonstrated above, focused feedback need not be laborious.  Embedding problem representations followed by a reasoned diagnostic conclusion into electronic notes promotes script selection and real-time visualization of missing components, schema formation, and a tool for reflection since the diagnostic process is often evolutionary.  

 

 

The end game

 

The objectives for medical trainees are to “practice” good medicine and become lifelong learners, sentiments captured in a revered Oslerian axiom: “The art of the practice of medicine is to be learned only by experience; ‘tis not an inheritance; it cannot be revealed… Know that by practice alone you can become expert.”  Explicit in this epigraph is the need for experience.  Implicit, but especially relevant today, is the need for aspiring expert diagnosticians to deliberatelypractice the components of diagnostic reasoning.  Otherwise, carrying forward today’s practice habits creates a cadre of experienced but overconfident non-experts – a perfect recipe for stagnation, repeated errors, and adverse patient outcomes.

 

 

 

 

Jayshil Patel, MD is an Associate Professor of Medicine (Pulmonary, Critical Care and Sleep Medicine) at MCW. He is a member of the Curriculum Pillar in the Robert D. and Patricia E. Kern Institute for the Transformation of Medical Education.

 

Paul Bergl, MD is an Assistant Professor of Medicine (Pulmonary, Critical Care and Sleep Medicine) at MCW.