Daniel sent us this one — he's been digging through medical databases, as you do, and noticed something that a lot of laypeople notice when they start doing that. You search for evidence on a condition, and what comes back is this weird mix of big formal clinical trials and these much smaller, more human-feeling things — case reports, case series, sometimes just a single physician writing up one patient they treated. And the question is, what weight do these stories actually carry in evidence-based medicine? Do doctors have real incentives to write them, or is it pure scientific altruism? And have any of these humble observations ever actually sparked a drug discovery or a clinical trial?
The short answer to that last one is yes, absolutely — and we'll get to some wild examples — but what makes this such a good question is that it sits right on the fault line between how medicine feels and how medicine actually works. A case report feels human. It's a story with a beginning, a middle, and an end. Somebody came in with purple urine, the doctor was baffled, they figured it out, and now we all know something we didn't know before. A randomized controlled trial feels like a spreadsheet had a baby with a statistics textbook.
The spreadsheet baby is the gold standard, though.
It is, and for good reason. But that doesn't make the case report worthless. It just means we have to understand what each tool is for. So let's start by placing these things on the map. The evidence hierarchy in medicine, bottom to top, runs roughly like this. At the bottom you've got case reports — that's n equals one, a single patient. Then case series, which is anywhere from two to about thirty patients. Then observational studies, cohort studies, case-control studies. Then randomized controlled trials. Then at the top, systematic reviews and meta-analyses that pool multiple trials together.
Case reports are the most numerous publication type in medicine, right? Despite being at the bottom.
They account for about twelve percent of all PubMed-indexed publications, which is enormous when you think about how many different types of medical papers exist. The typical anatomy of a case report is pretty consistent. A physician notices something unusual — an unexpected presentation of a known disease, a surprising response to a treatment, a rare adverse event, or something that might be a brand new disease altogether. The threshold for publication is basically interestingness. Is this something my colleagues haven't seen before? Does it add to the collective clinical picture?
Which is a weird filter, when you think about it. Interestingness is not the same as importance.
And that's one of the central tensions here. A case report of a guy who sneezes whenever he eats dark chocolate might be fascinating, but it might not change medical practice. Meanwhile, a boring-looking case report about three patients with a weird infection might be the first signal of an emerging epidemic.
The filter is, did this make a doctor say "huh.
And that's not nothing. That "huh" has been the starting point for a lot of medical discovery. But before we get to the impact side, let's talk about why physicians write these things in the first place, because the incentive structure is genuinely strange and worth understanding.
All right, so walk me through it. Why does a doctor spend ten to twenty hours writing up a single patient for zero dollars?
Zero dollars is the key phrase here. Unlike clinical trials, which can involve pharmaceutical funding, site fees, consulting arrangements — case reports pay nothing. The physician is doing this entirely on their own time. So the incentives are what you might call reputational and educational currency. And the biggest one, especially for early-career physicians, is what I think of as the academic treadmill.
The academic treadmill.
Here's the situation. Medical students, residents, junior faculty — they all need to demonstrate scholarly activity. It's required for residency applications, for fellowship applications, for promotion and tenure committees. And a case report is, frankly, the easiest publication to get. You don't need a grant. You don't need a research team. You don't need institutional review board approval in most cases. You just need an interesting patient and the willingness to write it up. There was a study in Academic Medicine in twenty twenty-three that found sixty-eight percent of U.medical students have authored at least one case report by the time they graduate.
Sixty-eight percent. That's not a niche activity.
It's practically a rite of passage. And the incentives cascade upward. For residents applying to competitive fellowships, having publications on your CV matters. A case report counts. It's not weighted as heavily as a first-author original research paper, but it's not zero either. And for faculty at academic medical centers, the promotion committee is looking at your publication list. Case reports pad that list.
It's the academic equivalent of putting "proficient in Microsoft Office" on your resume. It doesn't impress anyone, but its absence would be noted.
That's a little harsh, but not entirely wrong. There's a darker side to this incentive, which is that it encourages what some critics call "me too" case reports. A twenty twenty-one analysis in JAMA Internal Medicine found that around forty percent of published case reports describe conditions that are already well-documented. They're not adding new knowledge. They're just adding lines to someone's CV.
The system is generating a lot of noise.
But here's the counterpoint — even redundant case reports can serve a purpose. If a medical student in Nebraska writes up a case of something that's well-known at Mayo Clinic, that report might still be educational for a physician in rural India who's encountering it for the first time. The redundancy is partly a function of how fragmented medical knowledge is globally.
Okay, so that's incentive number one — academic currency. What's number two?
The "first to describe" prize. This is a real, if somewhat intangible, incentive. The first physician to describe a novel syndrome or a novel drug reaction often gets their name attached to it. Stevens-Johnson syndrome. Hashimoto's thyroiditis. These are all named after the physicians who first wrote them up as case reports or case series. That's a form of immortality in medicine.
That's a pretty good incentive.
And it's not just historical. It still happens. If you're the first physician to recognize and describe a new adverse event from a widely used drug, your case report might end up cited thousands of times. Your name becomes shorthand for a clinical phenomenon. That matters to people.
Even if there's no check attached.
Even if there's no check attached. The currency is reputation, citation counts, and what you might call intellectual priority. It's the same impulse that drives scientists in any field — the desire to be the first person to say "I saw this, and nobody else has named it yet.
What about the journals themselves? Do they have an incentive to publish these?
This is incentive number three, and it's one that most laypeople don't think about. Case reports are cheap for journals to publish. There's no trial infrastructure to vet, no massive datasets to review, no statistical analysis to scrutinize. A case report is typically a few pages, a handful of figures, and some references. The editorial review is relatively straightforward. And yet case reports drive citation counts. A well-written case report of a rare condition might be cited for decades because it becomes the go-to reference for that presentation. Many high-impact journals operate dedicated case report sections. BMJ Case Reports is an entire journal devoted to this. The Lancet has a case report section. The New England Journal of Medicine has its famous case records.
Citations are the currency journals care about.
Impact factor is built on citations. Case reports punch above their weight in that regard. So the journals are happy to provide the venue.
What about negative data? That's a big problem in clinical trials — nobody publishes the failures.
Right, and this is incentive number four, and it's important. Clinical trials rarely publish negative results. If a drug doesn't work, the trial often just disappears into a file drawer. But case reports of failed treatments or adverse events can still get published. If a physician tries something unusual and it doesn't work, or if a patient has a bizarre adverse reaction, that's still publishable because the novelty threshold is met. This fills a critical gap in the evidence base. The entire post-marketing surveillance system for drugs is built on this principle.
The FDA's adverse event reporting system.
The FDA Adverse Event Reporting System received over two million case reports in twenty twenty-three alone. Those are all spontaneous reports from physicians, pharmacists, and patients. Each one is essentially a mini case report. And those reports are what trigger label changes, black box warnings, and sometimes drug withdrawals. So this isn't just academic busywork. This is the primary mechanism by which we detect rare adverse events that didn't show up in clinical trials.
Because a clinical trial might have three thousand patients. If an adverse event happens in one in ten thousand patients, you'll never see it in the trial.
You need post-marketing surveillance, which is just a fancy term for "doctors writing down weird stuff they saw." So to answer the first part of the prompt — yes, physicians have real incentives. Academic currency, naming rights, journal economics, and the negative data loophole. None of them are financial, but collectively they drive a massive volume of publication.
All right, so the system is built to generate these reports. The question is, do they ever actually change anything?
This is where it gets really interesting. Let me trace one specific lineage that I think is instructive. In nineteen eighty-seven, a physician published a case report describing a patient with porphyria cutanea tarda — that's a metabolic disorder that causes severe skin blistering — who improved dramatically when given hydroxychloroquine. This was an accidental observation. Hydroxychloroquine was being used for something else, and the physician noticed the skin lesions clearing up.
But it was intriguing enough that someone followed up with a case series — a handful of patients, same treatment, same result. Then a pilot trial. Then larger trials. And eventually, in twenty fourteen, the FDA formally approved hydroxychloroquine for porphyria cutanea tarda. That's a twenty-seven year journey from a single case report to an FDA-approved indication.
The case report was the spark.
The case report was the spark. And this is not a one-off. The nineteen ninety-six case series of three patients with Kaposi sarcoma who improved dramatically on HIV protease inhibitors — that case series directly inspired the clinical trials that led to protease inhibitors becoming standard of care for HIV-associated Kaposi sarcoma. That's all it took to change the treatment paradigm for a cancer.
The twenty twenty-two case report of a patient with treatment-resistant depression who responded to ketamine — that single case was one of the catalysts for the SPRAVATO clinical trial program, which led to the FDA approval of esketamine nasal spray for treatment-resistant depression. And one of the most famous examples — the London patient. In twenty nineteen, the New England Journal of Medicine published a case report of an HIV patient who experienced long-term remission after receiving a stem cell transplant from a donor with a rare genetic mutation. That single case reignited the entire field of HIV cure research.
These things can be catalytic. But I'm guessing that's the exception, not the rule.
It is very much the exception. There was a twenty twenty analysis in PLOS ONE that tracked five hundred case reports describing novel treatment effects over a ten-year period. Only one in twenty — five percent — was ever followed up by a formal study. The other ninety-five percent remain isolated anecdotes. They're published, they're read, they might inform clinical intuition, but they never get systematically investigated.
For every hydroxychloroquine story, there are nineteen dead ends.
And that's the fundamental tension. Case reports are high-sensitivity, low-specificity instruments. They're very good at detecting possible signals. They're very bad at telling you which signals are real.
The canary in the coal mine.
That's exactly the right metaphor. The canary dies, you know something might be wrong in the mine. But you don't know what, you don't know how bad, and you definitely shouldn't evacuate the entire mining operation based on one dead canary. You need to send someone down to check.
How do physicians actually weigh these things in practice? If only eight percent would change treatment based on a single case report, what are the other ninety-two percent doing with the information?
The survey data from twenty twenty-two in BMC Medical Research Methodology is really instructive here. Seventy-two percent of clinicians said they find case reports "somewhat useful" for clinical decision-making. But they're using them for hypothesis generation, not confirmation. A case report tells a physician "this is possible." It doesn't tell them "this is probable." And those are fundamentally different kinds of knowledge.
This is possible versus this is probable.
And physicians are trained, at least implicitly, to apply something called the Bradford Hill criteria when they're evaluating whether an observed association might be causal. These are nine criteria that an epidemiologist named Austin Bradford Hill laid out in nineteen sixty-five. Strength of the association, consistency across multiple observations, specificity, temporality — did the supposed cause come before the effect — biological gradient, plausibility, coherence with existing knowledge, experimental evidence, and analogy.
When a doctor reads a case report, they're running it through these filters without necessarily naming them.
If a case report describes a rare adverse event with a strong temporal association — the patient took the drug and within hours developed a bizarre reaction — and there's a plausible biological mechanism, that carries a lot more weight than a case report that says "I gave this patient vitamin C and their cancer went away, I don't know why." The plausibility filter is doing a lot of work.
That plausibility filter is built on everything else the physician knows.
It's built on their entire training and clinical experience. Which is why a case report that would be dismissed by a specialist might seem compelling to a generalist, or vice versa. The filter is only as good as the knowledge base it's drawing on.
You mentioned the three-signal rule earlier. What's that?
It's a practical heuristic. Physicians typically want to see at least three independent case reports or case series describing the same phenomenon before they'll consider it a credible signal worth investigating further. One report is a curiosity. Two is a coincidence. Three is a pattern. It's not a formal statistical rule, but it's deeply embedded in clinical culture.
Like adopting a feral cat.
I'm not sure I follow that one.
One shows up, it's a fluke. Two, you start wondering. Three, you've got a colony and you're naming them.
actually a pretty good analogy. And the three-signal rule is part of why the redundancy problem we mentioned earlier isn't entirely wasteful. If forty percent of case reports are redundant, some of that redundancy is actually providing the "second signal" or "third signal" that moves a finding from curiosity to pattern.
There's a kind of wisdom-of-crowds thing happening. No single case report is reliable, but the aggregate starts to mean something.
This is where the modern "n-of-one" movement is trying to formalize what case reports do informally. There are now adaptive clinical trial designs — the I-SPY 2 trial for breast cancer is a famous example — where individual patient responses in biomarker-defined subgroups can trigger expansion cohorts. If you see a strong signal in a small group of patients with a particular genetic profile, you open the trial to more patients with that profile. It's essentially a case series with a statistical framework bolted on.
The case report isn't being replaced. It's being upgraded.
It's being systematized. Bayesian statistics gives you a formal way to update your beliefs as each new patient is observed. That's what physicians are doing intuitively when they read case reports. The statisticians are just making the math explicit.
Let me ask you about the flip side. We've talked about case reports as hypothesis generators. What about the danger of false positives? How often does a case report send the field down a blind alley?
More often than anyone would like to admit. The replication crisis in case reports is real. That five percent follow-up rate I mentioned — that's the fraction that even get investigated. Of those that do get investigated, many don't hold up. The classic example is the wave of case reports in the nineteen nineties suggesting that various alternative medicine approaches cured cancer. Individual patients, compelling stories, no controls. Almost none of them replicated when subjected to proper trials.
The human impulse to tell a compelling story works against the scientific impulse to be rigorous.
This is where the "feels more human" quality that the prompt identified becomes a bug, not a feature. A case report is narratively satisfying. It has a protagonist — the patient — and a conflict — the disease — and a resolution — the treatment worked. Our brains are wired to find that persuasive. A randomized controlled trial that says "the treatment group had a 2.3 percent absolute risk reduction compared to placebo, p equals 0.04" is not narratively satisfying. It's a sentence only a statistician could love.
The trial is more likely to be true.
The trial is more likely to give you an estimate that's close to the truth. It controls for confounding variables. It accounts for the placebo effect. It uses blinding to prevent bias. A case report does none of those things. The patient who got better might have gotten better anyway. The physician might be seeing what they want to see. The placebo effect might be doing all the work.
When a layperson is reading case reports about their own condition, what should they actually do with that information?
This is the practical question, and I think the answer is fairly straightforward. Use the case report to generate questions for your physician, not to change your treatment. So instead of saying "I read about this treatment and I want to try it," say "I read about this treatment that seemed to work in a case report — has this been studied in larger trials? Is there a plausible reason it might work for my situation?" Those are the right questions.
The case report as conversation starter, not decision maker.
And a good physician will welcome those questions. They might say "actually, that observation did lead to a trial, and here's what we know now." Or they might say "that was an interesting case report, but it was never replicated, and here's why the biology doesn't really support it." Either way, you've had a more informed conversation.
What about for the physicians listening? The medical students and residents who are thinking about writing case reports?
But write good ones. There's a set of guidelines called the CARE guidelines — that's CAse REport, capital C-A-R-E — that provide a checklist for what a high-quality case report should include. Timeline, diagnostic assessment, therapeutic intervention, follow-up and outcomes, patient perspective. And prioritize writing about novel observations rather than the forty percent of cases that are already well-documented. If you're going to add to the literature, add something that isn't already there.
The -insight here, and I think this is the thing I want to land, is that the evidence hierarchy isn't a ladder you climb. It's a toolkit.
That's beautifully put. A case report answers the question "is this possible?" A randomized controlled trial answers the question "is this probable?" A -analysis answers the question "how probable, exactly, and under what conditions?" Those are different questions, and they all matter. The problem arises when people use the wrong tool for the question they're asking — when they treat a case report as if it answers the probability question, or when they dismiss a case report because it doesn't.
The distinction between possibility and probability. That's the core epistemic insight.
And it's one that even experienced physicians sometimes lose sight of. A compelling case report can feel so persuasive that you skip the step of asking whether the finding has been replicated. That's human nature. But the discipline of evidence-based medicine is precisely the discipline of not letting a good story override the need for good data.
Where does this leave us, the layperson reading a case report about their own condition at two in the morning?
The two AM Googler. I think there are a lot of those.
There's a whole economy built on them.
Here's what I would say. When you encounter a case report, recognize it for what it is. It's a physician saying "I saw something interesting, and I thought my colleagues should know." It's not a treatment recommendation. It's not evidence that something will work for you. It's a signal that something might be worth investigating further. Ask the three questions. Has this been replicated in a larger study? Is there a plausible biological mechanism? How strong is the temporal and logical connection between the intervention and the outcome? And then take those questions to someone who can help you interpret them.
If your physician dismisses the case report out of hand, that's also not ideal.
It's not. A physician who says "that's just an anecdote, ignore it" is missing the point as much as a patient who says "this anecdote proves the treatment works." The proper response is "that's interesting — let me look into whether there's been any follow-up research." Dismissal and credulity are both failures of evidence-based reasoning.
We've covered incentives, we've covered impact, we've covered how to read these things. Let me ask a forward-looking question. As AI tools get better at mining electronic health records for patterns, does the traditional case report become obsolete?
I don't think obsolete. I think it evolves. Right now, a case report depends on a physician noticing something unusual and deciding it's worth writing up. That's a very lossy process. Most unusual things are never noticed, and most noticed things are never written up. What AI could do — what it's already starting to do — is systematically scan electronic health records for anomalous patterns and flag them for human review. So instead of one physician writing up one interesting case, you might have an AI system that says "here are forty-seven patients across twelve hospitals who all showed this unusual lab pattern before developing this rare condition." That's a case series on steroids.
The physician's role shifts from "I noticed something weird" to "I'm verifying something the algorithm flagged.
Writing it up, interpreting it, putting it in clinical context. The human element doesn't go away. It just moves upstream. The AI does the pattern detection; the physician does the sense-making.
There's also the rise of patient-reported case reports. Platforms like PatientsLikeMe, Reddit's AskDocs, all these places where patients are describing their own experiences in ways that sometimes resemble case reports.
This is a fascinating and largely unregulated parallel literature. A patient on Reddit describing their experience with an off-label treatment is functionally writing a case report, just without the clinical rigor. The signal-to-noise ratio is even worse than in the formal literature, because there's no editorial filter, no peer review, no requirement for diagnostic accuracy. But the volume is enormous, and some of those reports are undoubtedly flagging real signals that the formal system is missing.
How should physicians engage with that?
I think with cautious curiosity. If a patient says "there's a whole community of people online reporting this side effect that isn't in the official prescribing information," a physician shouldn't dismiss that. They should say "that's interesting — let me look into it, and let me make sure your experience gets reported through the proper channels." FAERS accepts reports from patients. That's an underutilized pathway.
The patient becomes part of the surveillance system.
They already are. They just don't always know it. And I think one of the healthy developments in medicine over the past decade has been the growing recognition that patient-reported outcomes and patient-reported adverse events are legitimate data. They're not as clean as clinician-reported data. They're often confounded by expectation and placebo and a dozen other things. But they're real, and they matter.
All right, so to pull this together. Case reports are the bottom of the evidence hierarchy, but they're the most numerous publication type. Physicians write them for academic currency, naming rights, and because journals want them — not for money. About one in twenty sparks a follow-up study, but the ones that do can change medicine. And the key distinction for anyone reading them is possibility versus probability.
That's the episode. And I'd add one final thought. The next blockbuster drug might start as a footnote in a case report written by a sleepless resident at two in the morning. The system is messy. It generates a lot of noise. But it's the only system we have for detecting the unexpected. And in medicine, the unexpected is often where the breakthroughs hide.
The sleepless resident at two AM. There's something almost romantic about that.
The whole enterprise of medical discovery depends, at some level, on someone being curious enough about a single patient to write it up for strangers to read. That's a remarkable thing when you step back and look at it.
Now: Hilbert's daily fun fact.
Hilbert: The word "tardigrade" was coined in the late Victorian period by Italian biologist Lazzaro Spallanzani, who named them "il Tardigrado" — meaning "slow stepper" — after watching their distinctive lumbering gait under a microscope. The name has nothing to do with their famous indestructibility; it's purely about how they walk.
I feel seen.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If you found this useful, rate us five stars and tell a friend who's ever Googled their symptoms at three AM. We'll be back next week.
