How do we Improve our Diagnostic Reasoning Skills? A Look at Problem Representation and Illness Scripts
Diagnostic reasoning is an art, and a very difficult art to master. We learn the prototypical features of disease from textbooks and lectures and are expected to apply this knowledge to the patients we see every day. Application of this knowledge is where the real challenge lies. How can we confidently make a diagnosis? This is not an easy question to answer.
Diagnostic reasoning concepts are hard. They’re hard to understand and even more difficult to apply. If you commit to the process (hey – you’re reading this blog!) you’re on the right track. It’s impossible to know everything in medicine. What we do need to know is how to manage clinical cases using a structured approach. How do we do this? We apply several diagnostic reasoning concepts to everyday cases. We develop meta-cognition to identify our clinical reasoning weaknesses. We debrief with colleagues. We learn, we refine, we make mistakes, and we grow.
If you’ve been following the blog so far, we’ve reviewed key elements of clinical diagnostic reasoning including system 1 and system 2 thinking (I suggest you read this post if you haven’t), pre-test probabilities, generating differential diagnoses using various frameworks, and several diagnostic accuracy concepts. Today, we dive into problem representation and illness scripts. As always, let’s demonstrate with a case.
Case: you’re seeing a 35-year-old Caucasian male with low back pain. How will you come to your diagnosis?
Just like a detective gathers information about a crime, the first step in the diagnostic reasoning pathway is acquiring data (1). Data acquisition is influenced by our knowledge, experience, and other contextual factors (B1). We gather data from the patient history, physical exam, and diagnostic test results (1). Early in the clinical reasoning process, we create a “problem representation.”
What is a problem representation? It is a brief (one-sentence) summary of the defining features of a case (1). Put simply, it’s our brain’s summary of the “problem” in front of us. A problem representation helps us generate a focused differential diagnosis. To develop a concise and relevant problem representation, we can answer the following:
Who is the patient? (demographics, risk factors)
What is the pattern of illness (acute, subacute, chronic), and tempo (stable, progressive, resolving, intermittent)?
What are the key signs and symptoms of the clinical syndrome? (2)
Back to the case: he tells you the back pain has been intermittent for the last six months and thinks it’s getting worse. It’s worst first thing in the morning and dissipates after about two hours. Movement seems to make things better. He works as a roofer so thinks this might be a contributing factor. He’s started massage therapy and takes ibuprofen a few times per week which helps to a degree. He feels more tired than usual. He denies fevers, IV drug use, recent trauma, urinary or fecal incontinence, or saddle anesthesia. He doesn’t have any past medical history. He takes no prescription medications. He denies any weight loss or drenching night sweats. He also tells you the back of his right heel hurts.
As you were reading through this paragraph, you may have been generating a mental picture of less likely, more likely, and must-not miss diagnoses of low back pain. You may have pulled information from past experience, theoretical knowledge, and key features of disease to steer you down a certain path. What would be your problem representation of this case?
Who is the patient? A young (35-year-old) Caucasian male
What is the pattern of illness? chronic, intermittent, and progressive
What are the key signs and symptoms of the clinical syndrome? Low back pain with morning stiffness that improves with exercise, malaise and Achilles heel pain.
If you were to present this case, the problem representation would be something like “I have a 30-year-old Caucasian male with intermittent low back pain associated with morning stiffness, malaise, and possibly enthesitis.” A succinct and accurate problem representation is key to getting to the right diagnosis. This is because it can trigger a specific illness script for a disease or group of disease.
BOTTOM LINE: A PROBLEM REPRESENTATION SUMMARIZES THE DEFINING FEATURES OF A CASE, WHICH CAN LEAD TO CREATING AN ACCURATE DIFFERENTIAL DIAGNOSIS.
What is an illness script? An illness script represents our knowledge about a disease in an organized summary. Recognition of illness scripts requires a combination of foundational knowledge about disease presentations and then seeing them in clinical practice (3). We store and recall knowledge as diseases, conditions, or syndromes, that we can connect to a problem representation (1). In other words, a problem representation triggers our memory, thus allowing us access to an illness script that we have learned about and seen in clinical practice (1). Some illness scripts represent groups of diseases (e.g. “inflammatory back pain” can represent ankylosing spondylitis, or psoriatic arthritis with sacroiliitis, and other rheumatological conditions); whereas others represent a specific disease (1).
Back to the case: A young male with chronic low back pain, morning stiffness, malaise and enthesitis (heel pain), triggers the illness script for inflammatory back pain. If you had no idea this was coming – don’t worry! For low back pain, let’s look at different etiologies and what the illness script for each might look like.
This doesn’t summarize all differential diagnoses (stay tuned for a future post on low back pain!)
Source: CORE back tool from Centre for Effective Practice
BOTTOM LINE: A PROBLEM REPRESENTATION CAN TRIGGER AN ILLNESS SCRIPT STORED IN MEMORY, WHICH CAN GUIDE US TO AN ACCURATE DIAGNOSIS.
Developing a problem representation and developing illness scripts links us back to theory of system 1 (intuitive) and system 2 (analytic) thinking. An experienced clinician may use intuitive thinking to automatically recognize key features of inflammatory back pain. A learner with less clinical exposure will use analytic reasoning to work through the case more deliberately by testing various hypotheses. However, when a clinical scenario deviates from the expected pattern or expected illness script, an experienced clinician may employ analytic thinking too.
To make things even more difficult, the defining features of diseases that we learn about in textbooks often don’t translate to the patient in front of us. For example with this case, if our patient wasn’t sure how long his back pain lasted in the morning, the illness script for inflammatory back pain may not be as obvious.
Patients don’t always give us the concise and clear history we hope for. We must be able to identify key features of the presentation, while discounting irrelevant information to form a problem representation that we can link to a recognizable illness script (4). This is why illness scripts for specific diseases are not fool proof – patients can have the same illness but will present in different (and often subtle) ways.
How can we improve our ability to work through this difficult process?
This comes back to the principle of exposure, exposure, and more exposure. The more patients we see, the more we learn. But seeing patients might not be quite enough. Individually, we can also reflect on cases. We can read around them and develop illness scripts to allow for quicker recall when a similar case presents itself in the future. Memory works best if we can anchor it to an experience. We can develop strategies to deliberately reflect upon our thinking processes and how we came to a diagnosis (referred to as meta-cognition) (5). If we’re teaching, we can prompt learners to think aloud to refine the problem representation and compare and contrast hypotheses of a case to challenge diagnostic reasoning skills (1).
BOTTOM LINE: SEE AS MANY PATIENTS AS YOU CAN, AND LEARN BY REFLECTING ON THE DIAGNOSTIC REASONING PROCESS.
Key Take Home Point:
To summarize: the frustratingly non-linear diagnostic reasoning pathway starts with knowledge acquisition, and melds into developing a problem representation, retrieving illness scripts, comparing and contrasting hypotheses, and forming the most likely diagnosis. All the while we are influenced by contextual factors, cognitive biases, foundational knowledge, and clinical experience. No one said this would be easy.
Readings:
Bowen JL. Educational strategies to promote clinical diagnostic reasoning. New England Journal of Medicine. 2006 Nov 23;355(21):2217-25. (https://www.cebm.net/wp-content/uploads/2013/06/judith-bowen.pdf)
Olenik J, Kohlwes J, Menesh RS, Connor DM. Problem Representation. [January 2024] Available from: https://www.sgim.org/File%20Library/JGIM/Web%20Only/Problem-Representation-Overview.pdf
Lubarsky S, Dory V, Audétat M-C, Custers E, Charlin B. Using script theory to cultivate illness script formation and clinical reasoning in health professions education. Canadian medical education journal. 2015;6(2):e61-e70.
Einstein DJ, Trowbridge RL, Rencic J. A problematic palsy: an exercise in clinical reasoning. Journal of general internal medicine. 2015 Jul;30:1029-33.
Chew KS, Durning SJ, van Merriënboer JJ. Teaching metacognition in clinical decision-making using a novel mnemonic checklist: an exploratory study. Singapore Med J. 2016 Dec;57(12):694-700. doi: 10.11622/smedj.2016015. Epub 2016 Jan 15. PMID: 26778635; PMCID: PMC5165179.