Teaching files, textbook examples--the case of the classic case
By Frank Davidoff, MD, FACP
The patient was a middle-aged man. Slightly bewildered by the presence of so many medical students--the 20 or so of us who had been summoned to his bedside from our many scattered teaching services--he listened quietly as our instructor pointed out the puffy, coarse features, the croaking voice, the slowed movements, the "hung" reflexes--virtually every clinical finding of a classic case of myxedema. In nearly 40 years of medicine since that day, a quarter of it in endocrinology, I've never forgotten him. And while I've seen many other patients with myxedema since then, I've never seen another patient so florid, so classic.
The demonstration of classic or textbook examples (paradigms, prototypical cases) is a hallowed tradition in clinical teaching, at least in teaching diagnosis. Classic cases are commonly collected into teaching files, a resource particularly beloved by people who work with medical images-radiologists, pathologists, endoscopists and the like-because, understandably, "picture perfect" images have so much intrinsic appeal. But, of course, classic cases also turn up in verbal form in grand rounds and teaching conferences and clinicopathological conferences; in textbooks (hence the expression "textbook case"); and in many other venues, including medical exam questions.
Memorable but rare
For all that they are memorable, a curious feature of classic cases is that they are also rare, which certainly proved true for my student encounter with "classic" myxedema. The basis for this important paradox is underscored by the traditional publications on diagnosis, those many papers that describe in meticulous detail the diagnostic findings of a particular disease in groups of patients. The tables in these papers almost always tell us that a few findings were present in 100% of the group, some findings were common (present in, say, 25%-75% of patients), and a long list of findings were unusual, each occurring in less than 5%-10% of the group.
It takes some doing to think about these data in reverse, that is, to translate the group information back into the clusters of findings that actually occur in individual patients. When you make that translation, however, it becomes obvious that it is the rare patient, statistically speaking, who can be expected to demonstrate every finding in the book. This is the "classic case," the patient whose findings most fully typify the disease (which, not coincidentally, is likely to be the patient with the most longstanding, hence perhaps also the most neglected or untreated, disease).
We have here, then, the makings of a pedagogical conundrum: the tradition of using case examples you may never see again to teach about diseases you are likely to see every day. For on the one hand, what could be more natural, more logical than to teach from classic examples? Or, stated the other way around, who would want to teach from the atypical case, the bizarre example, the exception? From the learner's point of view, the classic case is, after all, likely to be the most memorable case, and the creation of memorable mental structures, models or scripts is a crucial part of learning. But on the other hand, as we've seen, the "classic case" approach to clinical teaching is itself unrepresentative, even bizarre. Curiously, then, at some level, using extreme examples makes us uncomfortable. Perhaps the source of this discomfort is because at bottom it's reminiscent of the "best- in-show" mentality associated with dogs or cattle, where a champion is the exemplar of all the features that define the breed-and as clinicians and teachers we don't like to think we operate in a "best-of-breed" mode.
As a way of teaching diagnostic medicine, therefore, the use of full-blown, classic cases raises a number of interesting and important questions. Since its power seems to lie principally in being memorable, the use of classic cases may be an efficient way of "imprinting" beginners in the field with sets of diagnostic information. For a clinician with significant experience, however, the recognition of classic cases is generally a "no-brainer." For an expert clinician, encountering a classic case may be a pleasure, but the real challenge for expert clinicians lies rather in recognizing the subtle, the less-than-classic patient, the Sherlock Holmes-like unraveling of the baffling case. And both everyday practice and the published diagnostic literature described above teach us that the majority of patients with a given disease, in fact, present in less-than-classic, non-specific ways, most with only a few findings or even a single one. (There may be an even bigger reservoir of asymptomatic patients who would only be detected by some more sensitive means of testing, which is what screening is all about.)
Thus, we have now framed the reverse pedagogical conundrum: The patients you are likely to see every day (numerically most typical) are not likely at any moment in time to be classic cases of the disease (diagnostically most typical), or, stated more concisely, the typical patient is atypical. Two lessons emerge from this conundrum. First, the day-to-day experience of caring for ordinary patients provides the most important opportunity for diagnostic learning at an advanced level, that is, recognizing the subtle, the non-specific, the non-classic findings of disease.
Second, time is a critical dimension in case-based diagnostic teaching. For the simple reality is that many classic cases are not recognized initially, but rather are identified as such only after a good deal of diagnostic evidence is assembled. This is inevitably a process that unfolds over time, with many twists and turns, tantalizing leads and frustrating dead ends along its path. Yet in formal teaching exercises, the entire case history is traditionally presented all at once, giving learners a kind of bird's eye, after-the-fact version of the diagnostic process. Compiled case histories like these, presented after the fact, may be more classic, but they lack the clinical fidelity of time, hence oversimplify and distort the reality of learning and practicing clinical diagnosis.
It is precisely the ability to reconstruct this time dimension, at least partly, that adds teaching value to the particular species of clinical problem-solving exercises championed by Jerome Kassirer, MACP, and his colleagues (1,2). They accomplish this reconstruction through the simple stratagem of presenting "chunks" of information for discussion in sequence, as they emerged over time, rather than in the traditional cumulative snapshot.
Learning to weigh evidence
But there's more: Consider now the diagnostic dilemma of a 24-year-old woman who presents with the relatively rapid onset of a red and exquisitely tender big toe. Reasoning from cases like this, Custers and colleagues in Holland have argued that diagnostic information falls into several general categories, most importantly 1. enabling conditions (or "context" information, including age, gender, environmental and family history and the like); and 2. clinical consequences (which include signs and symptoms, clinical and laboratory findings) (3).
In the patient above, the presenting clinical consequence, podagra, is entirely typical (classic) for gout, but the enabling conditions (age, gender) are atypical (non-classic). Conversely, in the case of a 55-year-old hypertensive man who drinks a good deal of alcohol and presents with bilaterally red, tender wrists, the enabling conditions are classic for gout, but the clinical consequences are not. Interestingly, the Dutch investigators have shown that both experienced clinicians and medical students rely heavily on the typicality of enabling conditions (information that is important in determining prior probabilities) in arriving at diagnoses, but experienced clinicians place significantly more weight on typical enabling conditions than do students (3).
Thus, as much as we might wish otherwise, diagnostic information is not homogenous: A case can be partly classic, partly not. And, like it or not, learning advanced diagnostic skills involves not only learning how to manage subtle, non-specific, non-classic findings, but also how to combine and weigh dissonant information, the typical and the atypical, arising within a single patient. Classic cases occupy an important niche in the medical education environment, but it turns out after all to be a small niche. It's a pity: Classic cases are so clear, so recognizable, so memorable, so ... classic. You'd think that on their merits they'd count for more in clinical teaching than they do, more than the ordinary or the subtle. But, by the same token, where would the challenge be if everything were a classic?
Frank Davidoff is Editor of Annals of Internal Medicine.
1. Kassirer JP, Kopelman RI. "Learning Clinical Reasoning." Baltimore, MD; Williams and Wilkins, 1991.
2. Kassirer JP. Clinical problem-solving-a new feature in the Journal. N Engl J Med.1992; 326:60-1.
3. Custers EJFM, Boshuizen HPA, Schmidt HG. The influence of typicality of case descriptions on subjective disease probability estimates. Paper presented at the Annual Meeting of the American Educational Research Association, Atlanta, Ga., April 12-16, 1993.
Internist Archives Quick Links
New Leadership Webinars
The ACP Leadership Academy is offering FREE webinars covering the core tenets of leadership, leadership in hospital medicine, finance, and more.
Join ACP Today!
ACP membership connects you with like-minded colleagues and provides access to a variety of clinical resources, practice tools, and ways to earn MOC and CME.