Diagnosis, the Missing Course
By Susan Carr | Senior Writer, ImproveDx
Mark L. Graber, MD, founder and now chief medical officer of the Society to Improve Diagnosis in Medicine (SIDM), has said, “There is no course on diagnosis in medical school.”
How can that be?
The ability to diagnose patients’ medical problems is central to being a physician. How is it possible that diagnosis is not taught in medical school? The answer in part is a clarification: diagnosis is a process learned in practice, more through on-the-job training than dedicated coursework in the classroom or even in simulation. Students begin to gain building blocks of diagnosis—eg, basic science knowledge and illness scripts—in medical school and then become increasingly proficient as interns and residents through clinical apprenticeship with experienced physicians. Classically, diagnosis is more learned than taught.
In recent correspondence, Graber noted that the classical approach is currently being revisited as part of a new focus on diagnosis education:
This new field emphasizes two major shortcomings of the current apprenticeship model: First, thanks to myriad contributions from the fields of cognitive and social psychology, we know much more about the clinical reasoning process and how it goes astray in cases of diagnostic error. Second, it is increasingly appreciated that successful diagnosis depends on many factors outside of clinical reasoning.
Clinical reasoning is a complex process that until recently has been taken largely for granted. A simple definition includes the cognitive process by which “clinicians observe, collect, and analyze information that ultimately leads to an action (i.e., diagnosis and therapy)”1(p111) and reflects the physician’s knowledge base and experience, in addition to process-based skills.
In Teaching Clinical Reasoning, Robert Trowbridge, MD, Joseph Rencic, MD, and Steven Durning, MD, recognize the influence of factors outside the physician’s control, including the clinical environment, interactions with the patient, and subconscious dynamics that affect the cognitive process.
Historically, clinical reasoning focused on the clinician’s brain (“the world inside the clinician’s head”) but has now expanded to include patient and environmental factors (the clinician’s head inside the world”)2(np),3
This concept, referred to as situativity, explores how interactions affect the clinical reasoning process. Collaboration is fundamental to leveraging an individual’s knowledge with that of others, including patients and colleagues, together with data and technology, for improved outcomes.3 Situativity would be one element of the missing course on diagnosis. Other essential elements would be content on health information technology, critical thinking, cognitive psychology, and cognitive bias.4-6
The National Academy of Medicine, in its landmark report, Improving Diagnosis in Health Care,7 also emphasizes the collaborative, social nature of diagnosis. The process must include effective communication and teamwork, wise use of consultants and decision-support resources, productive working relationships with laboratory and imaging services, and life-long learning from feedback on performance. Recognizing the need to include this aspect of diagnosis in training, Graber notes:
Diagnosis is a process that interdigitates with every other part of a healthcare system. If there is no specific training on this, students will be sub-optimally prepared to achieve high levels of diagnostic quality and safety in real-world practice (written communication, September 2019).
Competencies Form a Foundation
This new approach to teaching diagnosis is consistent with a recent trend toward using competencies to “define and codify essential professional abilities”8(np) in health care professions education. Competencies are performance-based outcomes of training, things that students must learn to do, as well as know.9
With support from the Josiah Macy Jr Foundation, SIDM convened an inter-professional group of experts to develop a consensus curriculum based on competencies for diagnosis. The competencies span three domains: individual, team-based and system-related and range, for example, from collecting clinical findings, to collaborating with patients and families, and identifying how the work environment affects human performance. They are relevant for all health care professionals and may be adapted for all specialties and disciplines.9 Viewed as a first step toward improving diagnosis education, the competencies can be used to fill gaps in programming and assess the results, and serve as content guidelines for that missing course on diagnosis.
It is one thing to identify gaps in current medical school training and another to actually implement new curricula. Having defined competencies to work with helps but is not sufficient for addressing the very real barriers programs face. Competition for program time and a paucity of interested and qualified faculty are the major challenges innovators encounter.
The June 2019 issue of Diagnosis is devoted to diagnosis education and includes reports from medical schools that have recently implemented new programs and curricula. These early efforts demonstrate both opportunities and challenges.
In 2013–2016 Albert Einstein College of Medicine/Montefiore Medical Center, offered PGY-2 residents in internal medicine an elective curriculum in clinical reasoning.10 Seven classes covered strategies and reflection, cognitive error, and methods for strengthening medical knowledge. A total of 47 students participated in three sessions. Two different surveys were used for assessment. In 1 survey, all 30 participating residents reported they had improved their skills across 15 domains of clinical reasoning. On the other, 25 residents’ scores on a multiple-choice test of knowledge (eg, cognitive error, Bayesian analysis) rose from 58% to 81%. The authors acknowledge the study has limitations: a lack of standardized tools for assessing the effects of this kind of curriculum and their inability to measure the effect of the training on the rate of diagnostic error in clinical practice.
Another article describes a program at the University of Pittsburgh School of Medicine, which introduced a clinical reasoning curriculum for 1 of 4 groups of third-year students in an internal medicine clerkship, thus delivering somewhat randomized and controlled results.11 The curriculum consisted of 6 interactive online modules and 1 case-based workshop. Students were evaluated after completing the curriculum with a 20-question quiz, comparative analysis of their hospital admission notes, and a survey about clinical reasoning skills. The results showed improved knowledge and skills in clinical reasoning among the intervention group (66 students) compared with students in the control group. While this study shows positive results and will be extended with a similar program for pre-clinical students, the authors acknowledge that “optimal methods for teaching clinical reasoning skills have not been established.”11(p169)
Pilot Programs to Teach Clinical Reasoning
In addition to a small number of medical schools that already have focused programs on specific aspects of clinical reasoning (eg, communication or Bayesian reasoning) and other early efforts to implement curricula, more than 20 professional schools of medicine or nursing are currently developing pilot programs to improve diagnosis.
With support from the Macy Foundation and SIDM, these schools aim to improve the diagnostic performance of their students and advance what is known about how best to design, implement, and evaluate these programs. The following list is a sample of pilot programs designed for medical schools:
- A series of online video lectures with quizzes and exercises about the use of clinical decision support.
- Clinical reasoning curriculum plus a workshop for interns that will cover dual process theory, problem representation, and illness scripts.
- Comprehensive curriculum of 4 didactic sessions and 10-15 case-based modules with individual practice through an app, podcast-based assignments, and interprofessional case discussion.
- An instructional tool to help students develop and justify differential diagnoses by comparing a patient’s results to prototypical and atypical disease manifestations.
- A guided reflection tool designed for use with inpatients who do not have a diagnosis after 24 hours in the hospital.
- A curriculum that promotes synergies between diagnostic accuracy and high-value care.
- A program that uses gamification, adult learning theory, and core clinical reasoning principles to reduce diagnostic error.
Clinical Reasoning and Resilience
The process of clinical reasoning in pursuit of a diagnosis is highly complex and full of risk. Trowbridge and Graber observe, “Given all these challenges, it is remarkable that physicians establish the correct diagnosis at the current level of accuracy…,”2(np) and point out that improvement is still possible and compelling. To avoid patient harm, as well as harm to clinicians and others, formal education and self-reflective practice should be part of all healthcare professional training from the beginning.
With education and training, physicians will be more likely to know and accept that everyone is vulnerable to making faulty judgments and committing errors. And they will have the ability to mitigate that risk. Armed with knowledge and strategies for clinical reasoning, the ability to reflect on their own thought processes, and exchange constructive feedback with colleagues, physicians will be in a position to improve the performance of their diagnostic teams. They will be a source of resilience in care delivery, able to perform at the highest level possible, recognize potential diagnostic errors, and work continuously to improve.
In this Issue:
Get ImproveDx in Your Inbox
Don't miss an article. Get the Society to Improve Diagnosis (SIDM)'s newsletter delivered to your inbox.