Time Stamps

  • Case (1:19)
  • Diagnostic weight (8:42)
  • Hypothesis driven reasoning (12:26)
  • Anomalous clinical data (19:30)
  • Final diagnosis (22:20)
  • Take away points (31:21)

Human Dx Case Link: https://www.humandx.org/o/co7yrer3dim2y59t5c0sbd370

Show Notes

  • When building your illness scripts, pay attention to clinical findings that are characteristic for the diagnosis, rather than merely consistent with the diagnosis.
    • Characteristic findings are hallmarks of a particular diagnosis. They are usually present, and when they are, they strongly support the diagnosis; when they are absent, they argue strongly against the diagnosis. I.e. their presence or absence significantly alters the likelihood of a diagnosis.
    • Expert diagnosticians rapidly recognize characteristic findings, based on the clinical context.
  • Hypothesis driven reasoning often involve both the fast and slow processor. Optimal utilization of the combination is crucial to avoid cognitive biases
  • An anomalous clinical data is often times the only clue to the incoherence of a hypothesis. Its presence calls for rethinking your diagnosis creatively, don’t ignore it!
  • Disease processes that involve multiple organ systems may have highly variable clinical presentation. Don’t be fooled by a rigid illness script.


CINDY Hi everyone, this is Cindy Fang. I want to welcome you back to another episode of Hoofbeats ( Hoofbeats ). In this segment we challenge you to solve diagnostically difficult, real-world cases alongside experienced clinicians. As always, I’m here with my partner, Dr. John Hwang.

JOHN Hey everybody. So if you remember, on our last episode we invited you submit your diagnoses to us via the online case platform HumanDx, so we could crowdsource a collective differential diagnosis.

And we purposely chose a case in which the final diagnosis was a questionable one, one we could practice subjecting to scrutiny. And to slightly subvert the concept of the case didactic, I think.

Well, we’re going to do things a bit differently this episode. For starters, we already submitted this case to the HumanDx community a while ago, incognito. So we already have a collective differential, which we can analyze later.

And, this time, we’ve got a case with a definite answer, which we’ll reveal at the end of this episode.

CINDY You’re 100% sure of the answer?

JOHN As close as can be reasonably expected. Nothing is 100% in medicine.

CINDY Very meta. Without further ado, let’s hear this case from our colleague, Dr. Marty Fried

MARTY A 43 year old man presenting with one week of diarrhea.

Two weeks prior to presentation, the patient started developing dyspnea on exertion. His exercise tolerance went from unlimited to 5-6 blocks due to shortness of breath, associated with 2-pillow orthopnea. One week later he developed profuse, watery, non-bloody diarrhea. He estimates he is having 5-6 bowel movements daily. There is no clear association with eating, and the diarrhea occurs as at night as well. He has also noted over this time period generalized fatigue, malaise, anorexia, and subjective weight loss, though he cannot quantify exactly how much. He believes he is urinating more frequently than usual, and that his urine appears darker and “frothy”. He also complains of occasional bilateral leg cramping.

He has no significant past medical or surgical history but does not follow regularly with doctors. He immigrated from China 10 years ago, lives alone in an apartment, and works part-time at a restaurant, although he has not been to work in the past week because of his illness. He is not sexually active. He does not take any prescription or over-the-counter medications, nor herbal supplements. He has a 10 pack-year smoking history, but quit five years ago. He used to drink five to ten shots of hard liquor daily while in his 30s, but stopped drinking entirely 3 years ago. He denies any history of drug use.

On arrival the patient is febrile to 103F, heart rate is 105, blood pressure 108/63, breathing 16 breaths per minute, saturating 93% on room air. He is a thin Asian male resting quietly in bed, alert and anxious-appearing. The sclerae are icteric. The jugular venous pulsation is grossly elevated, visible at the earlobe. There is a systolic murmur audible at the upper sternal border. The heart rate is irregularly irregular. There are crackles audible at both lung bases. There is mild tenderness to palpation in the right upper quadrant. There is no peripheral edema.

CBC: 4.9 (normal differential) | 13.6 (89) | 213

BMP: Na 140 | K 3.8 | Cl 110 | bicarb 20 | BUN 29 | Cr 0.5 | Glucose 89

LFTs: AST 39, ALT 30, AP 194, Tbili 4.6, Dbili 1.6

HIV: negative

Lipase: 42

Hgb A1c: 5.9

ANA, AMA, ANCA, anti-KLM serologies negative

CK: 78

Urinalysis: Trace protein, negative for LE/nitrites or blood; normal microscopy

Blood cx, urine cx results — all no growth

Stool studies:

– Lactoferrin normal

– C. difficile PCR normal

– Stool culture no growth

– Stool examination for O&P: negative

ECG shows atrial fibrillation with rapid ventricular response (can show on website)

CXR shows an enlarged cardiac silhouette and prominent pulmonary vasculature with small bilateral pleural effusion (can show on website)

RUQ ultrasound shows a normal-appearing liver with patent vasculature.

TTE: The left ventricle is mildly dilated, diffusely hypokinetic, with an LVEF of 30%. The right ventricle is also dilated and hypokinetic. Both atria are severely dilated. There is moderate tricuspid regurgitation. The pulmonary artery and inferior vena cava are severely dilated, with an estimated right atrial pressure of 15mmHg. No vegetations are seen.

( Short musical transition/cue ).

JOHN So that’s the extent of the information you’ll hear about this case. I just want to point out Cindy, this is the second case of a young man with diarrhea we’ve had on this show. I swear this isn’t on purpose, but everything seems to be about fever, confusion or diarrhea on Hoofbeats.

CINDY That’s part of the game we’re playing with our listeners, isn’t it? A common presentation does not equate to an easy case. Remember the phenomenon of “cognitive ease”?

JOHN When we’re in a familiar situation, we tend to feel good about ourselves, and rely more heavily on unconscious or intuitive thinking.

CINDY Exactly. So if you see “diarrhea” and relax, you might be walking into a trap.

JOHN So what do you think? Take a moment if you haven’t already to think through the case, and when we come back we’ll hear what our discussant had to say.

Music interlude

Act 2. Diagnostic weighting and hypothetico-deductive reasoning.

CINDY John and I sat down with Dr. Catherine Constable, a hospitalist at NYU. Here is her initial reaction to the case.

CATHERINE CONSTABLE  Okay. So I read the case and then I guess the thing that stuck with me or that jumped out at me… um, obviously he seems to be in heart failure. He’s also in afib. And so I started thinking about things that connect… You know, afib and heart failure, kind of an interesting chicken and the egg problem, in that, you know, it could be tachymyopathy, he could be in afib and then he, you know, a couple days later he comes in heart failure. But, as I said, you know, there’s multiple organ systems affected here. And so I found it easier to connect a cardiomyopathy causing afib, and then connect the cardiomyopathy to something else, you know, something causing a systemic syndrome.

So to me it seemed like the primary issue in terms of what’s going on with this guy’s heart is a cardiomyopathy. And then due to his dilated atria, he ends up in afib.

CATHERINE CONSTABLE So the next thing I moved on to his thinking about was his liver, I guess the reason why I was also focusing so much on heart failure because I felt that I could explain the LFT abnormalities in the setting of heart failure, hepatic congestion. In this case, the bili is elevated. His total bili was 4.6. Elevated bilirubin is actually the most common LFT abnormality with hepatic congestion, it’s found in 70% of cases, and other LFTs are only mildly elevated. AST and ALT are elevated in only about a third of patients. And the liver ultrasound was unremarkable. Not sure what to make of that, but he’s got a very dilated IVC, which is consistent with right heart failure causing hepatic congestion.

JOHN Now recall that starting with the data and building up into hypotheses is what we call forward, or data driven reasoning, which we talked about a bit in episode 1. So here you see our discussant quickly clustering the patient’s presenting signs and symptoms into congestive heart failure, and then further clustering that plus atrial fibrillation and LFT abnormalities into a putative diagnosis of primary cardiomyopathy.

CINDY So she’s started by assigning relative weights to features of the case. The heart failure is central in her thinking. She can not tell yet if the afib is primary or secondary. But she suspects the LFT abnormalities are an epiphenomenon of the heart failure.

JOHN This is, this is definitely a recurring theme when we analyze discussants’ thinking: Half of diagnosis is figuring out what’s signal and what’s noise.

CINDY Kind of, but I think it’s more accurate to say they’re “assigning diagnostic weight”. “Noise vs. signal” implies binary extremes: the data is either trash or it’s gold. Here she’s debating the relative significance of the findings, but she’s not throwing anything in the wastebasket yet. And it emphasizes that this weighting is supposed to be fluid. If you get stuck later on in the diagnostic process, you can tackle the case from a different angle by rethinking what you weighed earlier.

( Short  musical transition/cue ).

JOHN Expert clinicians are good at assigning diagnostic weight to findings. Put another way, they understand which findings are characteristic for a disease, versus which are merely characteristic of that disease. For example, we all learned in medical school that infectious mononucleosis presents with fever, exudative pharyngitis, splenomegaly, cervical LAD, and atypical lymphocytosis. But the thing is, if you see a 20 year old with a fever and sore throat, a finding of atypical lymphocytosis on CBC is very characteristic of infectious mono. It’s usually present in people who have mono; if it’s not there, it argues pretty strongly against mono. In contrast cervical LAD, if that’s present, that is consistent with IM, but it’s also seen in many other diseases that would be on your differential.

CINDY So Dr. Constable is treating the hepatic panel elevation as consistent in the picture of heart failure.

JOHN Right, which makes sense. Mild transaminase elevations are consistent with many diseases causing congestive heart failure. Anything that causes elevated right sided pressures. She doesn’t dedicate much thought to this. But I imagine that would be different if instead if the patient had marked transaminase elevations in the thousands. That pattern is characteristic of certain primary liver diseases. Or imagine if the transaminases were elevated, but the alkaline phosphatase was undetectably low, that would be very characteristic of a single particular diagnosis, right? Wilson’s disease. So experts know which findings tend to be diagnostically discriminatory in a given clinical scenario, situation. And anchor their thought process on those characteristic findings.

CINDY If you are talking about Wilsons this early in the episode, I hope that’s not the final answer, is it?

JOHN I’m not saying it’s not. We’ll just have to wait and see.

( Short  musical transition/cue ).

CATHERINE CONSTABLE  So, then I thought… Thinking a little bit more about things that cause cardiomyopathy… Then there was the part about his past alcohol use. So that, you know, it’s, it could, you could definitely tie in alcohol. It seems like he was a pretty heavy drinker in the past. Um, so you could tie in the alcohol use with dilated cardiomyopathy and afib, and then that could maybe tie in the liver. Has got some elevated lfts as well, but you know, quickly I kind of moved on from that because it doesn’t explain his, he’s got some sort of B symptoms going on, and he’s got fever when it comes in. Also alcoholic cardiomyopathy gets better with abstinence from alcohol and so it doesn’t really explain why three years after he stopped drinking, he would come in suddenly in heart failure

JOHN OK, so here Dr. Constable sees alcohol use and tries to tie it with the illness by coming up with the hypothesis of alcoholic cardiomyopathy, and then just as quickly abandons it after she finds that the history and timeline does not fit with this.

CINDY An example of a backward, hypothesis driven approach. She comes up with a hypothesis and looks for evidence to prove or disprove it. Like we talked about in episode one.

JOHN Right. And I do think this is an opportunity to expand on the hypothesis driven approach a little bit. Because if you think about it, it’s a two step process: hypothesis generation, and hypothesis testing.

CINDY I guess that makes sense. Step one: coming up with a hypothesis, step two: using clinical information to validating and eliminate the hypothesis. I have to say I don’t always go through step two all the time. Say someone comes in with acute rhinorrhea, sore throat, and a cough in the winter. I would think: most likely a viral URI, tell the patient to just go home and rest, no need to do further testing.

JOHN Maybe, but actually, I would argue you did go through step two: you ordered the test of time. Meaning you anticipate the outcome that should be expected with the hypothesis. If it is indeed a simple viral illness, you expect the symptoms to self-resolve. If the patient comes back after two weeks with persistent symptoms, then you would know it’s time to re-evaluate your hypothesis.

CINDY Alright. And if you think of the hypothesis-driven approach as a two step process, I do think that it more often than not combines both fast and slow thinking.

JOHN Okay, hmm, let me see what I think about that statement… I guess it makes sense that hypothesis testing step requires a good amount of slow thinking: I need to think pathophysiologically what the expected findings are that should be there if my hypothesis was true, and look for that information in the case. I need to think about what tests to order to “prove” my hypothesis.

CINDY As for the hypothesis generation step, it is a step that we heavily rely on fast thinking. Pattern recognition, representativeness… Sometimes a single cue can trigger a quick association of specific diagnosis. Say the patient has a rose garden in the backyard? Sporotrichosis comes in to mind.

JOHN I actually like that you picked that particular example, because to me it seems like we do not have a lot control over how we come up with a diagnosis or a hypothesis. Sometimes the thoughts come to me and get stuck in my head and again maybe this is because I have a soft spot for ID — but I cannot stop thinking about the sporotrichosis once I learn about the rose gardening, even though obviously there are plenty of gardeners out there who don’t have that.

CINDY Exactly. Fast thinking by definition is imprecise, involuntary, even intrusive at times.

CATHERINE CONSTABLE I definitely have to watch myself and my tendency to anchor a little bit, you know, I get excited about a diagnosis it’s the natural human tendency is to look for more supporting evidence of the thing that is your gut reaction.

JOHN So, Cindy, I want to be more Vulcan. How do I suppress that kind of fast obtrusive thinking when I’m generating hypotheses?

CINDY I don’t think we need to at all… remember fast thinking is an extremely powerful tool. Take pattern recognition: when used well by an expert clinician, like by Dr. Janjigian in episode one, it can get you to the correct diagnosis without a lot of time and effort. Plus, there is nothing wrong with coming up with a wrong hypothesis – as long as you can recognize it by correctly performing the hypothesis testing step. I do want to stress that telling people to simply slow down all the time is definitely not the solution in the world of diagnostic reasoning. Instead, we should look for cognitive techniques to counter our biases. Simply telling ppl “don’t anchor” is not a good way to end anchoring

JOHN Believe me, you don’t have to convince me of that. Do you remember the whole blue and black versus white and gold dress nonsense from a few years ago? I saw white and gold, and you can’t tell me to see blue and black. My brain doesn’t work that way. You cannot will the desired outcome through sheer mental effort

CINDY My husband Jonathon actually sees blue and gold – nothing wrong with that, just how his brain is wired. Unfortunately due to time constraints, techniques to counteract biases will have to be addressed later/in future episodes. We really should let Dr. Constable continue first.

( Short  musical transition/cue ).

CATHERINE CONSTABLE So, THEN, I moved onto the diarrhea. Which, I think… the diarrhea and the fever were the hardest things for me tie in with everything else. So, to me it sounded like a secretary diarrhea, um, because, you know, there’s negative lactoferrin, he’s having symptoms all night long, it’s not associated with food.  The three separate things that I’m investigating are… the heart failure, afib and the LFT abnormalities are all in one bucket to me. And then there’s the diarrhea and then there’s the fever. He’s a fairly young otherwise healthy person. without much past medical history, except for, you know, pretty heavy alcohol use, I don’t think it’s likely that he would come in with multiple separate problems. If he was 80 it might be different.

CINDY So here’s her problem representation: a unifying process that accounts for cardiomyopathy, a secretory diarrhea, and a fever. And not any fever, a fever of unknown origin.

CATHERINE CONSTABLE I guess the thing that jumps out, the most is when you read the case is that there’s multiple organ systems affected so you start to think about things that can affect… rheumatologic, or infectious things, so that your mind is already kind of thinking about certain categories. So that was kind of my initial impression when I first read it. I started thinking about TB because he’s from an endemic area.  It does occasionally caused a secretary diarrhea, has a predilection for the ileo-cecal area. Um, it’s not like a super common manifestation, but I felt it could be tied in with that.

So going back to the cardiac stuff and thinking about cardiac manifestations of TB, the two most common would be constrictive pericarditis and a myocarditis or myopericarditis. And the constrictive pericarditis, it could cause a similar right heart failure picture, the atrial enlargement, but it doesn’t really mention that there’s no really other echo findings consistent with this. You don’t see like respiratory variation or abnormal wall motion of the interventricular septum, and he’s clearly got a low EF.


I thought of just plain old amyloidosis, um, you know, linking the heart failure, proteinuria, diarrhea… Um, doesn’t explain the fever and I thought it was less likely. Could this guy have a malignancy and these are you know, you can have a paraneoplastic syndrome that presents a lot of different things. Um, so I would consider that, but we haven’t found anything that’s suggestive of cancer or metastatic cancer. So in some ways it could be, you know, thyroid with the heart failure and the diarrhea. I don’t think it’s endocrine, with the fever. I don’t think we’ve ruled out the rheum category.  But we don’t have any serologies suggested of that.

CINDY So Dr. Constable started going through systems and considering possible diagnosis within them to test out “where this patient fits”. She likes some proposal better than others, but acknowledges that none of of them explains all three core findings perfectly. The high fever of 103 seems to be a troublemaker.

JOHN I feel like I run into this problem from time to time. When we run into an “outlier” information in a case, or a “loose end” that we cannot explain away, what do we do?

CINDY The easiest way is to ignore it. The way qualitative researchers treat their data outlier. Come up with an excuse why its not real. “That fever must be a rectal temp done the ED. Its fake.”

JOHN Obviously not an approach that will sit well with most of us. We really ought to go out of our ways to ask “what does not fit” even when we think our proposed diagnoses are plausible. Anomalous information implies inadequacy of our hypothesis.

CINDY The opposite would be to keep generating hypothesis to force it to fit, like hammering a square peg into a round hole. This is what a computer program does when you input all the key findings and ask it to split out diagnoses – the answers at the bottom of the list is usually garbage. It’d just keep generating diagnoses even when they don’t fit anymore.

JOHN We would like to be smarter than computers, so knowing when to stop is important too.

CINDY A useful trick would be to approach the case differently, change directionality. If you have been taking the forward, data -driven approach and there is a loose end you just cannot account for, it may be time to go backwards and come up with an hypothesis with this data.

JOHN So actually Dr. Constable does this later in the interview, though we didn’t catch it on tape. When she reaches an impasse in explaining the fever, she inverts her problem representation. She makes it a patient coming in with a fever of unknown origin, and that activates a different diagnostic schema.

CINDY Another trick that may be useful would be to build a causal model that incorporate the current both the outlier and everything else. Is there an underlying process that could have caused both? Are is there a potential process that can link the outlier to the rest of the case in a cause and effect fashion?

JOHN And actually, fever is a great example of this. Acute febrile illness often triggers decompensation of an underlying rheumatologic or endocrine disease.

CINDY One last way to potentially look at it is to embrace Hickam’s dictum. Maybe it’s time to ask: are there two separate processes going on here?

JOHN True, true unrelated. Although our discussant implicitly rejects this because the patient is young and previously healthy; he’s begging for a unifying diagnosis.

CINDY To be honest, when I search through the literature, I didn’t find a lot of answers to this question. Please let us know if you have a great technique to manage loose ends or outliers when you’re dealing with a case.

JOHN And with that, I think it’s time to discuss what happened in this case. Make sure you’ve mentally locked in your diagnosis.

CINDY So John, you actually took care of this patient, right? Shall we?

JOHN Though to clarify, I actually met this gentleman on his second hospitalization, which occurred a number of months after his first. Okay, so the answer. So we presented this case to you with a lot of data available, but in retrospect that probably made it more challenging. Because the truth is the critical lab test was sent from the ED, on that initial hospitalization, off that first set of bloodwork. And that was a TSH, which was undetectably low. The thyroxine level was more than twice the upper limit of normal.

The team who took care of him that time, found no alternative cause to the fever despite extensive work up. And as soon as the patients was started on methimazole, the patient had dramatic clinical recovery in a few days. The fever resolved, the hepatic panel abnormality, the tachycardia, and the diarrhea disappeared.

CINDY Yea! A happy ending!

JOHN Right, except…

CATHERINE CONSTABLE What about the fever?

CINDY Yea Dr. Hwang, what about the fever? I mean, a low grade fever in thyrotoxicosis makes sense,and it would not surprise me if he presented in full blown thyroid storm and a fever of 103. But from the description we don’t get a sense that the patient was THAT sick, was he?

Something to explain all three of cardiomyopathy, diarrhea, and a culture negative fever…I have an idea: How about just a plain old viral illness? There’s a fever caused by a virus, he has a little bit of a viral enteritis, and its a really bad viral cardiomyopathy? The lack of leukocytosis doesn’t really fit, but I heard somewhere that sustained leukocytosis is not typical for a viral illness later in the course. The TFT could have been just a coincidence.

JOHN You know I distinctly remember promising our listeners a definite answer. Diagnostic uncertainty was last week’s episode, no? These are all good points. Though I’ll point out he’s coming in with an illness that’s lasting longer for two weeks. What convinces me, is… As I said, I took care of him the second time he was hospitalized. What happened after his first hospitalization… as I said, he did much better, TFTs normalized. At some point he was lost to followup. And he was admitted to my service several months with the exact same presentation: fever, abnormal LFTs, rapid afib, CHF… in the setting of having run out of his methimazole. So…

CINDY It’s like he intentionally stopped taking his medications to prove to you that you were right.

JOHN Yeah, he basically did his own n-of-1 study. Maybe he needed to prove it to himself. Some patients are like that. I want to go back to something you said earlier. You said the fever wouldn’t have surprised you so much if he presented in full blown thyroid storm. There is a scoring system for predicting thyroid storm in patients presenting with thyroxicosis, the Burch Wartofsky’s scale. It’s a scoring system that considers the severity of end organ damage and symptoms. Points are assigned for the following: fever, CNS effect, GI-hepatic dysfunction, tachycardia, CHF, the presence of afib, and the presence of precipitating events. A score of >45 highly suggests thyroid storm.

If you crunch the numbers, this patient actually had 75 points. Now of course this scale is a prognostic scale, it intentionally uses liberal criteria to avoid false rule outs. But I want to point out, if you accept that he has a very severe thyrotoxicosis, the high fever shouldn’t be that surprising.

CINDY I don’t typically use scoring systems, but even knowing the diagnosis, it really took a review of the B&W scale to remember that this patient was pretty sick. I guess I was fooled by my own rigid illness script – I think of thyroid storm as an ICU diagnosis. They come in crushing like the ones I admitted to the CCU in residency. But I ought to think of hyperthyroidism/thyrotoxicosis/thyroid storm as a continuum of the same disease instead of three distinct categories. Another reason I might be thinking of him as a not as sick patient is from my own bias. As hospitalized patients get sicker and sicker, I am getting more comfortable w/ peri-SDU pts sitting on the floor. Anyone who is not immediately crashing in front of me is not sick in my opinion… This is the contrast effect. And thats so scary to think about – I could have a pt w/ impending thyroid storm sitting on my service, yet I would not have realized it.

JOHN Well Cindy, if this makes you feel better, I said that we had submitted this case to the HumanDx community. So thyrotoxicosis was not high on the community differential. It ranked 10th, and that was behind some fairly esoteric diagnosis, like carcinoid syndrome and Chagas’. Folks ended up submitting some 35 distinct submitted entries.

I have been thinking about why it would be so challenging to recognize thyrotoxicosis in this case. Part of it might be because of the artificiality, of having so much data. But I also think because the process involves so many different organs, and because there is high variability in symptoms and signs…. At least to me it highlights how a single rigid illness script can be inadequate to cover a multisystem disease like thyrotoxicosis or another endocrinopathy.

CINDY When I learned it in med school, it’s all about the anxiety, heat intolerance, tremor. But not every patient with hyperthyroidism comes in with that. I actually found a 1988 case for example where a woman w/ afib and fever of 102 was hospitalized x 6 weeks for FUO, and ended up having hyperthyroidism. I mean, if a pt comes in w/ fever 102 and jaundice as the two prominent findings, I would not think of hyperthyroidism until after the 3rd ERCP…

JOHNRight, and I practice at Bellevue, and I’ve admitted patients who’ve had very little symptom overlap with the patient you just described. Especially elderly folks… I remember taking care of a couple of folks who presented with a clinical picture of rapidly progressive dementia, with apathy and psychosis, but certainly no tachycardia or fever or atrial fibrillation.

CINDY One more element that makes it harder is the forum we used to discuss the case. In real life I bet a lot of people will throw in the TFTs in the work up without much hesitation, even when its not high on the differential. we are “primed” to think in a fashion suitable to an academic conference, a CPC type environment where the end diagnosis must be something sexy, something where there are only 300 cases since 1980s.

CATHERINE CONSTABLE There’s absolutely something metacognitive. Yes. If we’re talking about it, you think it’s either a vasculitis or TB because we love to talk about that.

Conclusion. Teaching points and take-aways.

CINDY So shall we summarize what we’ve touched on working through this case and hearing from Dr. Constable?

JOHN Of course! First off, we talked about the difference between findings that are characteristic of a disease and findings that are merely consistent with a disease. And so expert clinicians build illness scripts on characteristic findings, which allows them to recognize them in a case, allowing them to swiftly narrow the differential diagnosis.

CINDY We talked about how hypothesis-driven reasoning often involves utilization of both fast and slow thinking. The combination can set you up for cognitive biases at the transition point.

JOHN We touched on what to do when faced with an outlier. Be creative. Look for a bigger answer, look for a second answer, or instead of treating it as the outlier, treat it as your anchor. But don’t ignore it and, just as importantly, don’t fixate on it.

CINDY We also got a reminder that certain disease entities that affect multiple organs, like hyperthyroidism, can have highly variable presentations. A rigid illness script only does you a disservice here.

JOHN And I think that should do it for this week. We want to thank Drs. Catherine Constable and Dr. Marty Fried for weighing in this episode. Special thanks to our audio editor for this episode Richard Chen and Harit Shah. As always an honorable mention to our fellow podcaster Dr. Steven Liu, who didn’t get the diagnosis this week — but he did just make genius level on the New York Times Spelling Bee… so congratulations Steve.