Time Stamps

  • 01:18 Case introduction
  • 11:40 The Casablanca strategy
  • 23:17 The diagnosis revealed
  • 25:30 Logical fallacy: Denying the antecedent 
  • 33:22 Components of illness scripts
  • 41:37 Conclusion

Show Notes

  1. Clinicians sometimes respond to a clinical problem (whether a symptom, a sign, or an abnormal test result) by reflexively ordering a stereotyped battery of tests (e.g. iron studies, folate, and vitamin B12 in response to anemia), a behavior that Dr. Pat Croskerry has dubbed the Casablanca strategy, or “rounding up the usual suspects.” 
    • This contrasts with the hypothetico-deductive approach to evaluating a clinical problem, in which the clinician orders a limited number of tests to confirm or discredit specific diagnostic hypotheses, typically while following the principles of Bayesian reasoning.
    • We tend to employ the Casablanca strategy when we are novices; i.e. when we are early in our training, or when we are facing a clinical problem that is unfamiliar to us. These are contexts in which we lack the domain knowledge and pattern recognition necessary to constrain the problem to a manageable number of hypotheses.  
    • However, even experienced clinicians often use the Casablanca strategy in real-world practice. Although it cannot substitute for more sophisticated methods of diagnostic reasoning, this strategy enables the clinician to “buy time”,  occupying the patient — during which the clinician can reflect further on the case, while allowing the problem to evolve (and perhaps self-resolve).  
  2. A logical fallacy is an invalid form of argument in which the conclusion reached does not actually follow from the premises. 
    • Denying the antecedent (or the inverse fallacy) is a specific logical fallacy. This error is committed when we are given the premise “If P, then Q” and erroneously conclude “Therefore, if not P, then not Q.” 
    • For example, we often accept that if a septic patient defervesces in response to antibiotics, then the antibiotic coverage is adequate. However, it does not logically follow that if a septic patient fails to defervesce, the antibiotic coverage must have been inadequate.
    • The validity of an argument must be evaluated separately from the truth of its premises and its conclusions. For example, while it is undoubtedly true that “if you are a cardiologist, you went to medical school”, it is erroneous to conclude that, since I am not a cardiologist, I did not go to medical school.
  3. An illness script is a mental framework a clinician uses to organize their knowledge about a particular disease. An effective illness script comprises three elements:
    • The Enabling Conditions: features of the patient which explain why they develop the disease, such as demographics, exposures, comorbid conditions, and so on;
    • The Fault: the pathophysiologic processes that occur in the disease;
    • The Consequences: features of the resulting illness, such as the symptoms, signs, and findings on evaluation. 
    • Studies of diagnostic reasoning suggest that expert clinicians not only have a more expansive and more complex library of illness scripts, but that their scripts differ in emphasis as compared to those of novices. Specifically, the illness script of an expert typically encodes much more information about a disease’s enabling conditions, rather than focusing solely on its consequences.


CINDY: Hi everyone, Cindy Fang here. Welcome back to another episode of Hoofbeats, where we challenge you to solve diagnostically difficult, real-world cases alongside experienced clinicians. And as always I’m joined by co-host John Hwang.

JOHN: Hey everyone. By now you probably know Cindy and I like to try something a little different with each episode. Well, listeners have pointed out that our cases to date have essentially all been presented from the perspective of an inpatient physician. And we resolved to change that.

CINDY: For this episode, we are going to present you a case where the patient initially presented to an outpatient provider, and the diagnosis was made purely in the outpatient setting. Our discussant is a general internist who practices both inpatient and outpatient medicine, who is invited to solve the case wearing his clinic hat.

JOHN: And at the end of the episode, we’ll hear from the clinician who made the final diagnosis in real life, right in his office.

CINDY: Without further ado, let’s hear the case first from Dr. Shreya Trivedi.


A 65-year-old man visits his primary care physician’s office. He hadn’t seen his doctor in several years.

He says he’s lost his appetite and for the past two months, has been feeling vaguely fatigued.

While his exercise tolerance has been fair, overall he feels less energetic than usual. 

He thinks this is because he left his gym and no longer does aerobic exercise, leaving him more sedentary than before. 

Still, he is not troubled, because he is still able to walk without limitations and says in other respects he is in his usual health.

When asked, he specifically denies fevers, chills, night sweats, bleeding, or bruising. He is unsure whether he’s lost weight.

In terms of his past medical history:

He was diagnosed with major depressive disorder 20 years ago.  He sees a mental health provider regularly, and his symptoms have been well-controlled on buproprion for many years.

He has mild intermittent asthma, for which he uses an albuterol inhaler as needed.

He also has a heart murmur, which he has known about since he was a teenager.

He does not take other medications, nor OTCs or herbals.

He does not use tobacco, but drinks alcohol occasionally. He denies using illicit substances.

In terms of his family history, he remembers that his mother was diagnosed with breast and ovarian cancers, and that both his sister and mother have been diagnosed with mitral valve disease. An uncle on his mother’s side had leukemia.

On examination, he’s a middle-aged man with no particularly surprising findings. Specifically:

His vital signs are normal.

He appears well nourished.

There is a soft, late systolic murmur heard best at the apex that does not change appreciably with positioning or Valsalva.

The lungs are clear.

The abdomen and extremities are  unremarkable.

JOHN: So that’s the end of this patient’s first office visit. What do you think? What would you have done at this point? Take a moment to collect your thoughts, and we’ll hear how our expert discussant approached the case, after the break.

CINDY: Welcome back. This week, we sat down with Dr. Isaac Holmes, a general internist who splits his time between the inpatient unit and the medicine  clinic at Bellevue Hospital, who was also one of my favorite clinic preceptors back in my residency days.

HOLMES: So, you know, the original, the initial information is very vague, which is very true to clinic. And you know, my initial thought was there’s sort of two broad types of patients that come to clinic with complaints. I mean there’s the other separate set of people who are coming with chronic diseases that you’re actively managing or who just want a routine exam and some screening. But when people have complaints, I feel like they tend to fall into one of two categories. There’s people who are very hyper-aware and have a lot of very specific complaints. And then there’s people who are less aware of their own self, maybe, and have… just know they don’t feel well but are very, have a great deal of difficulty explaining why they don’t feel well, being able to verbalize exactly what feels bad to them or what feels different. And so this case, at least the initial part really fell into that second category of somebody who is coming to clinic saying, I feel bad, but I’m having a hard time really explaining why I feel bad.

HOLMES: My approach is really about trying to pin people down with questioning about how they feel bad. And my practice may be different from many because there are a lot of different languages spoken in my exam room and often that can be, it can be a language issue or an interpreter issue. And so asking the same question in a little bit of a different way can be very helpful.

HOLMES: I will say that I often ask people specific questions about how they’re engaging in their personal and professional lives when they don’t seem to be able to characterize exactly the symptoms that they’re feeling, as a way to get at how much functional impairment there is in their day-to-day life from these symptoms.

HOLMES: And maybe I’ll be more specific: Like, was he still working full time or did he quit his job, go down to half time? Was he still engaged? I don’t remember now whether he, he says he’s single, whether he has kids and whether he’s participatory and what’s going on with his kids…whether he knows what’s going on with his family or if he’d been more withdrawn from his family life and from his work.

CINDY: So we don’t have time to show you his whole initial reaction, but I was impressed by how he really spent A LOT of time talking about the specific questions he would ask the patient and what he would look for on physical exam. 

JOHN: Hearing him say this reminds me how I really don’t get the chance to see how other doctors interview patients anymore, now that I’m an attending. Anyone else feel this way? When I do get to observe an expert, it’s in morning report or noon conference, and I’m listening to how they interpret, organize, manipulate data. But how clinicians elicit that information from their patients in the first place, that has got to be an equally important step in the clinical reasoning process — and maybe even the most important, because everything else is downstream. So our discussant himself calls attention to how the very format of how this exercise can’t capture the essence of what he does in his office.

HOLMES: This reminds me much more about precepting a case than it does about seeing a patient. Because when I see a patient, I have the opportunity to observe the patient and their behavior directly. And that means a lot to me as an outpatient clinician. How did they walk to the room? How did they respond to their name? How do they look today? Look wasted? Cachectic? Do they look like, well engaged? Did they look like they’re carrying out their everyday tasks?

HOLMES: Spending a lot of time in clinic trying to answer diagnostic questions makes you very, uh, thoughtful about the way you examine patients, because that’s your… and the way that you, you take a history… those are your diagnostic tests.

CINDY: When we think about the diagnostic process, we often think about hypothesis generation, and obtaining data or tests to prove/disprove your hypothesis. One may be under the impression that clinicians always ask pointed questions with certain diagnoses in mind. But that’s not true all the time. Dr. Holmes is asking these questions to build a richer context first that will later on be the basis of his diagnostic process.

CINDY: He is not satisfied by the fact that he cannot feel, touch, smell… interrogate a live person  in front of him, and the missing information is valuable in helping him build an impression.

JOHN: In contrast, the tests he asks for at this point are pretty basic and sparse. He orders a CBC, BMP, a Hgb A1c, and a hepatic panel given his drinking history. And of course, he orders arguably the best test in an outpatient clinician’s diagnostic arsenal: Time.

HOLMES: I’d probably see him back for the two month exam first, try and get a sense of if his symptoms had changed, characterize if he was really having weight loss, do a repeat examination and see if anything had changed on his examination. Uh, and if his functional impairments had progressed at all.

SHREYA: Labs from the first clinic visit start to trickle back.

His CBC shows he is anemic, with a hemoglobin of 11.5 g/dL, an MCV of 87, and a hematocrit of 36%. He’s also thrombocytopenic to 87,000. His WBC count is normal, at 5.2.

His basic metabolic panel is completely normal.

His hepatic panel is also normal, although the albumin is very slightly low, at 3.4 g/dL.

His Hgb A1c is normal.

Two months after his initial visit, the patient returns for his follow-up.

HOLMES: So it’s interesting. Interesting labs. I think it really guides us in our care of this patient. The labs are markedly abnormal, and really specifically markedly abnormal.

HOLMES: For a 65 year old man with generalized fatigue, feeling tired, and weakness with this CBC, I say I’m alarmed about what could be going on with him. I think a little bit, you can be hard pressed to come up with a totally benign reason for someone to have this CBC. It’s a pretty abnormal CBC. The fact that it’s thrombocytopenia and anemia, and that it’s a normocytic anemia probably argues against things like chronic blood loss from his colon tumor…

HOLMES: And so it’s really arguing more towards either an absorptive process or probably more likely a bone marrow based process, and immediately makes me think about hematologic malignancies. The fact that he doesn’t have a big whopping white count certainly argues against acute leukemia or even chronic leukemia in an elderly patient like CLL, and points towards a lymphoma type process. I guess the other things are a chronic inflammatory disorders that might cause you to have thrombocytopenia and a low grade anemia with a normocytosis.

HOLMES: Other occult infections? They are definitely on that differential. Again, he’s been afebrile, he didn’t have a white count… You’re wondering about infections that maybe don’t drive like a big leukocytosis and febrile reaction. So atypical type infections, I mean I guess we should talk about…

CINDY: When I hear the complaint “fatigue”, I have a set of initial diagnostic tests that I typically fall back on. CBC, the thyroids, the basic metabolic, etc. When people hear anemia, most people react with the question, “Is it macro, normo, or microcytic?” while sending out the usual panel of iron studies, B12, folate, smear, retic count. I.e. the “anemia panel”.

JOHN: Same here. But that’s not what our discussant is doing here; he’s in full inductive reasoning mode. 

CINDY: Listening to Dr. Holmes’ thinking process, I realize how much I rely on my algorithms and routine panels in working up common complaints.

JOHN: It’s not just you, Cindy. This kind of stereotyped, “see abnormality, order panel” behavior is common enough among clinicians that it has a name in the literature: the “Casablanca strategy”. This term was coined by Pat Croskerry — Dr. Croskerry, as many of you probably know, a renowned expert in cognitive problem-solving in medicine.

Now, to be honest, I’ve never watched this movie, so I had to consult Wikipedia for a plot synopsis (please insert your joke about millennials here). Croskerry calls it the Casablanca strategy because apparently at the end of the movie the chief of police orders his men to “round up the usual suspects.”

CINDY: Right, so sending the anemia panel is one example. Another example would be reflexively ordering PTH, vitamin Ds, PTHrP to work up hypercalcemia. The Casablanca strategy is so frequently utilized, I was surprised to see that Dr. Holmes chose not to adopt it.

JOHN: Well, even though we all use this strategy from time to time, I think we tend to be reflexively disdainful of this approach. It feels unsophisticated, amateurish. Sherlock Holmes doesn’t order panels. He deduces, he makes amazing inferences based on small pieces of critical data. 

But Isaac Holmes doesn’t get to choose his cases, and he certainly doesn’t get to spend a whole book unraveling them. Isaac Holmes gets fifteen minutes in his clinic office  (tops, and that’s if he’s lucky, if he doesn’t have to unjam his printer, or call for a prior auth, or lose his progress note). Croskerry writes that the main benefit of the Casablanca strategy is “to buy time.” In the movie, when the chief of police orders his men to “round up the usual suspects,” it is not because it will help find the killer (he knows who did it). It’s merely to gain some time. Routine tests keep the patient busy and allow time to act as an intervention, to cull mild symptoms and diseases that self resolve. And so this strategy I think seems uniquely well-suited to the clinic, where limited cognitive bandwidth, and having separating signal from noise, are two perpetual problems for the clinician.

CINDY: While I agree it’s a useful strategy, it should not substitute for a formal workup tailored to the specific patient and the clinical situation. Say, for the hypercalcemic patient, ordering the usual panel is useful on the night of admission, but it should not replace a good history and physical to assess for this specific individual’s risk factors. How many bottles of vitamin D do you inhale daily? How many people have cancer in your family? How long have you had that massive lymph node on your neck? Those are still questions you should still ask.

JOHN: So instead of ordering panels, what did our discussant focus on? Well, he spent a lot of time asking clarifying questions: Have the symptoms progressed, has he had night sweats, has he lost weight? He asked whether there was lymphadenopathy on exam. And again, he ordered a very short, targeted set of tests: a reticulocyte count, a peripheral smear, and an abdominal ultrasound to look for splenomegaly.

SHREYA: The patient reports he is feeling about the same, as when he first saw his doctor two months ago. Generally fatigued, but not impaired functionally. 

He does wonder now though whether he’s lost some weight, though he’s not sure.

The peripheral smear does not show any abnormal WBC or RBC morphologies.

His reticulocyte count and index were borderline low.

His serum iron and total iron binding capacity measurements were both low, while his ferritin was 200

An abdominal ultrasound was ordered, and did show splenomegaly.

HOLMES: So this patient has splenomegaly, a low reticulocyte count, loss of appetite, maybe weight loss, maybe not, generalized weakness without night sweats… I think we’re making a story that is sounding uh, concerning for a lymphoma and I think my next step is going to be to look, I would get a CT scan to look for lymphadenopathy that we haven’t been able to feel on an exam as a way to possibly guide a diagnostic further diagnostic workup.

HOLMES: One thing we didn’t talk about that this would be a weird presentation of, but I guess is possible with an unexplained anemia, normocytic anemia, is myeloma — like a plasma cell dyscrasia — and whether there’s benefit in sending like a, a little bit of a myeloma workup with an immunofixation, well, protein electrophoresis and immunofixation in the urine, and then light chains.

HOLMES: I’m struggling to figure out exactly what might be causing his splenomegaly. Common explanations for splenomegaly-… So you have hepatic disease, right? Portal hypertension causing splenomegaly? Uh, you could have hemolytic disorders, uh, that are leading to splenomegaly. Hematologic malignancy is probably the big one that you think about and lymphomas especially. We haven’t seen it.

JOHN: Regular listeners of Hoofbeats should recognize this as the beginnings of our discussant’s diagnostic schema for splenomegaly… But nah, don’t worry. We just did a schema episode, so we’ll move on. Though if you do want to go down that rabbit hole I will make a quick plug for the Clinical Problem Solvers and their incredible collection of schema, including for splenomegaly — be sure to check out their podcast and app if you haven’t already.

And with that, it’s now time to move to this patient’s third and final visit to his physician’s office.

SHREYA: A month passes, and the patient returns for his third visit to his PCP’s office. Looking at the studies you ordered in advance of this visit:

A repeat CBC shows he is becoming increasingly anemic, now with a hemoglobin of 7.7 (down from 11.5 months ago), and an MCV of 80, down from 87 months ago. He is still thrombocytopenic as well, to 80,000 from 87,000.

A serum protein electrophoresis resulted as “trace detectable lambda chain.”

He also did undergo a CT of his chest, abdomen and pelvis, which confirmed splenomegaly but did not detect any lymphadenopathy or other abnormalities.

During his visit, he reports he is feeling increasingly fatigued. 

On exam he has lost 5 kilograms of weight since his first visit.

He is tachycardic to the low 100s; otherwise, his vitals are within norms.

HOLMES: So a trace detectable lambda light chain is not something that screams myeloma to me. I mean, when you’re thinking about myeloma, if you’re getting a light chain, it ought to be like a massive lambda or a massive kappa. To circle back to the real abnormality, right? 7.7, uh, wow. He’s gotten, he’s gotten quite ill. I’m really concerned there’s a number of processes right that could be causing this severe progressive anemia with thrombocytopenia. We were talking about lymphomas, but the other types of fibrotic, bone marrow processes, myelofibrosis… it doesn’t seem like aplastic anemia given he has a white count, he’s not pancytopenic, but you’re really wondering like, what’s going on with his bone marrow? And we have so much evidence pointing towards a hematologic disease with bone marrow dysfunction, in this case, that at this point I’m really just going to send this patient to hematology for a bone marrow biopsy, to evaluate for a bone marrow process, driving his insidious progressive symptoms.

HOLMES: If this person has a normal bone marrow biopsy, uh, then I think you’re really thinking about infectious processes that are causing a severe anemia of chronic disease.

HOLMES: So atypical type infections. I mean, I guess we should talk about possibly tuberculosis, intra-abdominal tuberculosis. Other pathologic similar pathologic processes like sarcoidosis, although he didn’t have a significant amount of lymphadenopathy in his chest. Uh, fungal infections, it seems unlikely. Arthropod borne diseases, usually you’re sicker with, uh, that can cause significant splenomegaly. He doesn’t seem to have a lot of potential exposure.

HOLMES: Certainly in our patient population I see a lot of, I do a lot of addiction medicine and I’m always wondering about either an occult hepatitis, which he doesn’t seem to have cause there’s LFTs are pretty normal, uh, or like a subacute endocarditis that people feel generally weak with and are tired but aren’t having a lot of localizing symptoms with a non virulent type bug.

SHREYA: Tests results slowly trickle back in the week after the patient’s third visit to his doctor’s office.

His HIV test is negative. 

Hepatitis B and C serologies are negative. 

His RPR is negative. 

Lyme titers were negative. 

But A C-reactive protein was elevated to 72. 

He is referred to a cardiologist; a transthoracic echocardiogram showed mitral valve prolapse associated with mitral valve regurgitation, but no vegetations, and no other structural heart disease. 

He is also referred to a hematologist for consideration of a bone marrow biopsy.

JOHN: It’s at this point in the case, in the real world, that the final diagnosis is made.

CINDY: With the additional information and the upcoming hematology evaluation, what do you think? What other questions, exams, or tests do you have in mind for this patient?

JOHN: Finalize your thoughts, and we’ll resolve this case, after the break.

JOHN: So our discussant has done what he can. Was he right? Enter Dr. Kenneth Hymes, the hematologist to whom this patient was referred, and the clinician who made the final diagnosis.

HYMES: So as I recall this patient, and it was a middle aged man who had a history of fevers, uh, fatigue, anemia — who hasn’t even had fevers, I believe, no fevers — just fatigue, anemia. And um, he was known to have a heart murmur.

HYMES: If you see someone with a low serum iron, a low iron binding capacity, and a ferritin that’s above a hundred… You’re feeling pretty comfortable the patient is not iron deficient and that everything you’re seeing is, is inflammation. From that point, you then have to think about what sort of inflammatory disorders can do it. We also have to think about malignant diseases… again, this patient was scanned from top to bottom, and nothing was found.

HYMES: Unusual presentations of infectious diseases? But I’m going back and why couldn’t the patient just have something simple like endocarditis? But endocarditis isn’t simple. Because it’s not just an infectious disease. It’s an immunologic disease. And as a result, you have to think when you see somebody who looks like they have an autoimmune disorder, but none of the serologies are classically positive.

HYMES: A lot of adults will have, you know, 60 year olds will have monoclonal antibodies, but you can also see it as part of an inflammatory process. So I think you, you, you have to sort of sift through that sort of data and put it aside and say, well, if it doesn’t really scream at me that the patient has a vasculitis, maybe there’s some other process, and it’s cheap and easy to get a blood culture.

And now I’m smiling, because as anyone who knows me will know, this is my absolute favorite diagnosis. Cindy, what did that blood culture show?

After the initial visit in the hematology clinic, Dr. Hymes directed the patient straight to the emergency room because the diagnostic certainty of endocarditis in his mind was that high. 

His blood cultures grew ¾ bottles positive for cardiobacterium hominis, a HACEK organism that’s part of the mouth flora.

Lo and behold, the repeat echocardiogram revealed a 1.5cm square vegetation on the mitral valve with a now severe MR and a completely perforated posterior leaflet. A thorough interrogation by the infectious disease consultant did not reveal any dental procedures or any other additional risk factors other than having the congenital valvular defect. 

CINDY: So I took care of this patient very briefly after he was diagnosed and referred to the inpatient side for further management. It’s exactly the type of case I love to talk about. The diagnosis is so bread and butter medicine that I should be able to pick up as an internist. It’s so obvious retrospectively, and yet I know I would not have been able to diagnose it, at least so effortlessly, the way Dr. Hymes did (who is a hematologist, btw). So I was thinking to myself, why is this case an obvious case of endocarditis to some but not to others?

JOHN: Personally Cindy, I wonder whether the reason SBE wasn’t entertained sooner was because the patient wasn’t febrile.

CINDY: I think that plays a big part of it. Growing up we learn endocarditis as a differential diagnosis for unexplained fever, but some people might just reject the diagnosis of endocarditis in an afebrile patient.

JOHN: I have heard residents talk like that, and honestly from time to time I catch myself saying things like that. Which is really troubling to me, and I’ll explain why — but it requires that we talk about propositional logic. Think back to high school, which was the first time most of us encountered propositional logic. Remember sentences like, “If P, then Q. Because not Q, therefore not P.”?

Let’s say I stated to you that if a patient spikes a fever, that patient has endocarditis. Because the patient in our case did not have a fever, therefore, this patient does not have endocarditis.” What’s wrong with this argument?

The most obvious thing wrong here is that the first statement, is medically false. Many things other than endocarditis cause fevers. In other words, the premise on which this argument is based is incorrect, and therefore, the conclusion is wrong.

But there’s a second thing wrong here, which is that the way in which this argument is constructed is logically unsound. If I said to you, “If you are a cardiologist, then you went to medical school.” Well, that’s a true premise. But if I then concluded “And because you are not a cardiologist, you didn’t go to medical school,” you folks out there in internal medicine, GI, rheum, ID, pulm/critcare, or maybe business school because you hated medicine, you would probably all be offended, or think I was crazy. The conclusion does not follow from the premise, even if the premise is 100% true. It is a non sequitur. In the language of propositional logic, we have committed a formal fallacy, the fallacy of the inverse, otherwise known as denying the antecedent. 

Logical fallacies are easy to detect when the content of the premise or the conclusion is this comically absurd. But it’s much easier to make this mistake when the content is more complex and cognitively demanding. Take the statement: If a patient being treated for pneumonia with ceftriaxone has an improving leukocytosis after 48 hours, then the antibiotic selection was appropriate. But let’s imagine we have a patient with pneumonia on ceftriaxone whose white cell count is not improving. It feels natural to conclude that our initial antibiotic selection was incorrect. Time to broaden to cefepime, maybe add vanc. Levaquin or amikacin? Here the premise sounds reasonable, the conclusion sounds reasonable, the management decision that results from all this sounds familiar and reasonable — but the logical structure of the argument underpinning all of this is flawed. 

CINDY: Logically we know it’s not true when we say it out loud, but our brain can be easily tricked into saying it, especially when we are in a hurry. The same person who rejects the diagnosis of endocarditis based on the fact that the patient is afebrile, you can ask that person to sit down and give you a lecture on Duke’s criteria, and very quickly that person would realize that fever is just one of the minor criteria, and an afebrile patient can easily be diagnosed with endocarditis by fitting all the other major or minor criteria.

JOHN: Yeah Cindy, and also, we don’t always get punished for committing logical fallacies. Fortunately, or maybe unfortunately, we get away with fuzzy or sloppy logic a lot in medicine, I think. More than we’d like to admit.

JOHN: We did ask Dr. Hymes himself what he thought enabled him to make the right diagnosis.

HYMES: When I went over the data, the referring doctor had thought of all of the critical things and had really excluded them. I mean that’s, as I said, that’s, you know… The last guy to look is the guy who gets the diagnosis.

JOHN: That was his first answer, and I think just him being humble. I feel compelled to point out: He’s the last guy to look because he’s the first guy to solve the case, not the other way around. Here’s his second answer, when we pressed him.

HYMES: I was pretty convinced that he had endocarditis because, because autoimmune diseases had, he didn’t fit any pattern there. He didn’t fit in a malignant disease. I didn’t think he was bleeding. And that’s where you have to start thinking about chronic infection as the cause. I mean, he was quite sick. This guy really looked moribund. He had lost a lot of weight. And this is what a chronically ill, chronically infected patient looks like. He looked like he had cancer. But we couldn’t find any cancer.

HYMES: In the old days, you know, in the seventies and eighties, you know, as a fellow, particular as a fellow at Bellevue, you saw a lot of bacterial endocarditis. And again, other patients come to me now with these sorts of unusual presentations. In other words, they’ve been through several referring physicians and they come to me sort of asking, why is this patient anemic? Or why are they losing weight, more so than why are they running fevers or why do they have embolic lesions?

CINDY: He seems to be used to having atypical presentation of endocarditis (and other miscellaneous infectious and autoimmune entities) referred to his office as hematologic mysteries, and therefore he had a high acuity to detect those as a result. The same way really good dentists are able to diagnose patients with atypical trigeminal neuralgia, which is commonly misdiagnosed or takes a while to diagnose.

JOHN: Right, you get the sense his illness script for endocarditis is fundamentally different from mine or yours, Cindy, not only because he’s seen much more of it, but because his cases are fundamentally different in nature. His attention is drawn to how sick and exhausted the man looked. I love reading really old case reports, from the 19th and early 20th century — really just for the hilarious jargon, like hysteria and dyspepsia and dropsy. But there’s this phrase, “toxemic exhaustion” that was used a lot in case reports of endocarditis from that era. Back when there was no treatment for this disease, you’d routinely see its late stages; doctors would watch as their patients become cachectic and listless, dissolve and waste away. And these case reports sound exactly the way Hymes remembers his patient. Makes you wonder whether they’d even bother presenting this patient on a 19th century podcast. William Osler would’ve probably diagnosed this patient at a glance.

JOHN: Speaking of illness scripts, Cindy, we’ve used that term without ever really defining or examining it on Hoofbeats, haven’t we? And hearing Dr. Hymes talk about he conceptualizes endocarditis, it makes you wonder, how can we build better illness scripts ourselves?

CINDY: I was doing a bit of reading to see if there’s a cognitive tool that could help me in this regard. This may not be the perfect answer to the question, but I think building a rounded illness script may be the way to go. The emphasis being on “rounded”.

CINDY: John, what’s your definition of an illness script?

JOHN: I could go on and on about illness scripts, but I think a commonly accepted (if somewhat incomplete) definition of an illness script is that it’s “an organized mental summary of a clinician’s knowledge of a disease”.

CINDY: And what usually goes in the script for you? Or let me ask the question another way. Listeners, if you don’t mind pausing and answering first before you continue. What’s your illness script for pulmonary embolism? And what would be your illness script for urinary tract infection? Now comparing these two scripts, are they structurally similar?

JOHN: I think you’re going to have to explain what you mean when you say “structurally similar”, Cindy.

CINDY: When the concept of illness script was first proposed in the 1980s, the clinical reasoning researchers thought illness scripts should have three major components. They are called the enabling conditions, the fault, and the consequences in the literature.

JOHN: To explain each of these individually: Enabling conditions are things about the patient, like age, sex, comorbidities, family history, habits, occupation, exposures — anything that influences whether the disease arises and how it manifests.

CINDY: The Fault would be Dr. Hwang’s favorite category — the pathophysiology of the actual disease entity.

JOHN: And Consequences are things about the disease: the symptoms that the fault produces, the tempo, severity, and other characteristics of those symptoms, the findings on exam, lab testing, imaging, etc.

CINDY: Later scholars added additional elements such as prognosis and management at the end, but traditionally it’s been proposed that illness scripts should at least contain these three basic components.

JOHN: Not surprisingly, when we start off as nonexperts about a particular disease, our illness script for that disease tends to be sparse. We simply lack knowledge. Our scripts are supposed to grow as we see more cases. But it’s not just that the illness scripts of experts contain more information (although they do) than those of nonexperts. Rather, they differ in what components they emphasize.

CINDY: It’s been noted that learners tend to focus solely on the consequences — the signs and symptoms of diseases — while expert clinicians have illness scripts that contain many more enabling conditions.

JOHN: Not only that, but when experts are given a bunch of patient-related information (medical history, medications, travel, etc), they are better at detecting potential connections between these seemingly-irrelevant data and potential diagnoses.

CINDY: Experienced clinicians are also noted to be better at acquiring clinical information that would be used as the enabling conditions, which in turn speeds up their diagnostic process and contributes to more accurate diagnosis by making the clinical  reasoning process more contextual. 

JOHN: For example, when interviewing a febrile patient in the middle of flu season, I might try to confirm my suspicion for flu by asking whether she has myalgias, or a sore throat. But a mark of expertise in this situation is not to try to confirm or reject a presumptive diagnosis of flu, but to probe for unsolicited information that might completely recontextualize my understanding of who the patient is. (“What do you do for work? Oh, you’re an abattoir worker? Hmm.)

CINDY: So when I think about a clinical problem, the question is not so much “What does this look like? ” but also “What is my patient at risk for?” especially when the signs and symptoms alone are vague, non-specific and not helpful.

JOHN: Right. I mean, case in point, listen again to how Dr. Hymes remembers the case, when we first sat down with him and asked him what he recalled. 

HYMES: So as I recall this patient, and it was a middle aged man who had a history of fevers, uh, fatigue, anemia — who hasn’t even had fevers, I believe, no fevers — just fatigue, anemia. And um, he was known to have a heart murmur. And I think that really is more of an edge than you get in many of these patients.

JOHN: It’s striking to me how much of a role the heart murmur played in his thinking process. I would’ve been tempted in his place to just write the murmur off as congenital, chronic, and therefore irrelevant. But Dr. Hymes was hyper-aware of that data point from the beginning, which made the diagnosis seem practically obvious to him, he said. During our interview Dr. Hymes actually shared some other diagnostically interesting cases he was involved in that ended up having non-hematologic diagnoses. And the common feature  in each case he made the diagnosis by focusing on the substrate, the patient. In this case it was a valvular defect, in another it was a recent surgery, and in another case it was an overlooked piece of travel history. By his account, he didn’t make these diagnoses by focusing on the symptoms or signs for which the patient had been referred to him — inflammatory anemia, barely detectable levels of light chain in the blood, and so on.

CINDY: Dr. Holmes our discussant, also asked us early on if this patient’s murmur got louder during the course of the workup, while he was asking many questions focused on the H&P. I intentionally did not provide him the answer and distracted him with other information, in fear of him solving the case in five minutes and leaving us nothing to talk about.

JOHN: Oh boy. Hope he’s not listening to this, Cindy.

CINDY: Certain conditions like HIV, transplant, hematologic/oncologic malignancies… these are consistently enabling conditions for a wide array of diseases. So going through residency I learned to be extra sensitive to those risk factors. Other risk factors are rarely as relevant, so I am not good enough to pay attention to them — say, a heart murmur. 

Right, and the two expert clinicians we interviewed for this episode seem to be relying on two distinct ways to recognize when something subtle and commonplace like a heart murmur becomes diagnostically significant. Dr. Holmes talked about how he relies on becoming deeply knowledgeable about his patient as a person, in a way that’s really unique to a primary care provider. And Dr. Hymes talked about how he became deeply knowledgeable about certain presentations of this disease because of extensive experience and the unique insight of being a consulting hematologist.

CINDY: And that’s exactly what I don’t have as a hospitalist. I am ashamed to say it’s very difficult for me to really learn a patient beyond a long list of past medical, past surgical, social history, medications especially first thing in the morning. I will have my routine intake conversation with the patient and their family, quickly sort into “relevant” vs. “irrelevant” buckets depending on why they are in the hospital in the first place, and maybe I’ll read the last three discharge summaries when I have time later in the day. But that makeshift knowledge about the patient is the only clinical context I have for the case, and that’s built it a hurry, and is very shallow in comparison.

CINDY: So…what do I do? In theory, if I build my illness script correctly from the start, that information should then activate the illness script for endocarditis through memory association for me. Is that how it works?

JOHN: Similar to what we talked about in the last episode with diagnostic schema, there’s debate in the clinical reasoning literature about whether these kinds of sophisticated illness scripts is what enables experts to be experts, or whether they emerge from the process of being an expert. In other words, some authors, particularly in the early literature, argue that experience and practice rather than education are what create physicians with conceptualizations of disease that allow them to instantly, effortlessly recognize some ordinarily small or nuanced detail as diagnostically powerful within a particular context. In this view, focusing on the script, trying to appropriate an expert’s script is futile — like stealing a trophy and expecting that’ll make you into a star athlete.  

Then there are others who have created modules and curricula meant to train students to develop illness scripts that mimic those of experts in both structure and richness. These studies are a great read and their results are thought provoking — we’ll include links in the show notes and you can decide what you think. I think for now these are best considered hypothesis-generating.

CINDY: So I just have to practice for 50 more years and wait for my white beard to grow then.

JOHN: Hey, I can’t even grow a beard.

CINDY: I’m still going to stay positive. I’ll still be actively building rounded illness scripts (ie to include the enabling conditions, the fault, and the consequences), and hopefully that’ll help my brain catch on.

JOHN: What happened to this patient after he was diagnosed, Cindy?

CINDY: He was transferred to the cardiothoracic service and underwent successful mitral valve repair as well as appropriate antibiotic treatment for subacute endocarditis.

JOHN: How about his anemia and thrombocytopenia? It’s great that we found an endocarditis and treated it, but was it the cause of all his problems? How do we know he does not also have a concurrent hematologic process going on?

CINDY: His anemia and thrombocytopenia completely resolved, and the CRP level trended down. On follow up visit a couple months later, he reports feeling fair and asymptomatic. 

JOHN: Alright. I am willing to provisionally accept that as a happy ending.

CINDY: For those of you who remember, we had a prior episode where a young woman presented with vasculitic rash who ended up having subacute endocarditis, another unusual yet sexy presentation for endocarditis. Now that we presented two endocarditis cases, does that mean we should retire the diagnosis from Hoofbeats?

JOHN: Never. Over my toxemically exhausted body.



Croskerry, P. (2005). Diagnostic failure: a cognitive and affective approach. AGENCY FOR HEALTHCARE RESEARCH AND QUALITY ROCKVILLE MD.

Topan, A., Carstina, D., Slavcovici, A., Rancea, R., Capalneanu, R., & Lupse, M. (2015). Assesment of the Duke criteria for the diagnosis of infective endocarditis after twenty-years. An analysis of 241 cases. Clujul Medical, 88(3), 321.

Turner, J., & Badireddy, M. (2019). Anemia. In StatPearls [Internet]. StatPearls Publishing.

Chapman, J., & Azevedo, A. M. (2019). Splenomegaly. In StatPearls [Internet]. StatPearls Publishing.

Sharara, S. L., Tayyar, R., Kanafani, Z. A., & Kanj, S. S. (2016). HACEK endocarditis: a review. Expert review of anti-infective therapy, 14(6), 539-545.

Custers, E. J., Boshuizen, H. P., & Schmidt, H. G. (1998). The role of illness scripts in the development of medical diagnostic expertise: Results from an interview study. Cognition and instruction, 16(4), 367-398.

Hobu, P. P. M., Schmidt, H. G., Boshuizen, H. P. A., & Patel, V. L. (1987). Contextual factors in the activation of first diagnostic hypotheses: expert‐novice differences. Medical Education, 21(6), 471-476.

Schmidt, H. G., & Rikers, R. M. (2007). How expertise develops in medicine: knowledge encapsulation and illness script formation. Medical education, 41(12), 1133-1139.

Keemink, Y., Custers, E. J., van Dijk, S., & ten Cate, O. (2018). Illness script development in pre-clinical education through case-based clinical reasoning training. International journal of medical education, 9, 35.

Tags: , , , ,