Red Herrings in Clinical Cases: How to Spot Them
What is a red herring in a clinical case?
A red herring in a clinical case is a true detail that pulls you towards the wrong diagnosis because it isn't the actual cause. It is misdirection rather than a lie. A red flag signals imminent danger and should trigger action; a red herring points away.
Mamede and colleagues call these salient distracting features, findings strongly associated with a particular disease but unrelated to the problem at hand. They're true, they grab attention, and they don't explain why the patient is here today.
Smoked herring turns red and was dragged through woods to train young scent hounds to ignore other smells, which is the origin described in a tick's tale. In the same report, an 81-year-old with fever, myalgia and cough had a recent tick bite that suggested Lyme, but serology was negative and the fever likely came from Actinomyces pneumonia. The tick was the red herring.
That differs from an atypical presentation, where the disease itself looks odd, see when common diseases look odd.
Why one misleading detail is so powerful
A single salient but irrelevant detail can steer clinicians to the wrong diagnosis, especially when it appears early in a complex case. Experiments show it shifts attention toward a familiar alternative. Being warned about the trap often does not help, which is why red herrings in clinical cases are so sticky.
In one vignette study, 51 attending internal medicine doctors at a large teaching hospital each reviewed 10 clinical vignettes seeded with a single misleading detail. Wrong diagnoses occurred in 90% with the detail versus 30% without it, and warnings about misleading information did not help.
In a 2014 Erasmus study, 72 residents tackled 12 cases in three versions, without a distracting feature, with one at the beginning, or at the end. They defined salient distracting features as findings strongly linked to a particular disease but unrelated to the actual problem. In complex cases, an early distracting feature cut accuracy to 0.18 correct, about 18%, versus 0.43 without one, about 43%. Placed near the end, performance sat around 0.36 and did not affect decisions.
These are vignette experiments on paper cases, not ward outcomes.
The four kinds of red herring you will meet
- Old, unrelated history. One kidney stone about 15 years ago, passed, nothing since; unrelated to the leg. Long-standing reflux, controlled, that the case says isn’t the cause and shouldn’t be linked to the main complaint when asked. Both feel relevant, neither moves the present diagnosis.
- An incidental abnormal result. A mildly raised amylase is a classic trap in mesenteric ischaemia, it’s often misread as pancreatitis and used to stop the work-up. It fits a tidy story, then blocks the real one you need to chase.
- The patient’s own theory. “A bad stomach bug or something he ate.” “Trapped a nerve.” Hear it, don’t anchor on it. Calling a limb case “DVT” or “a trapped nerve” is a dangerous, limb-losing miss when the real process is worse.
- A label someone else gave earlier. Three admissions in 18 months with the same severe tummy pain, every scan and camera test normal, told it is stress or that he is after painkillers. Prior labels bias you to repeat the miss rather than reopen the differential.
The counter-move is to ask what the herring cannot explain, then go hunting for things the true diagnosis would need to be there, and for negatives that carry weight too.
Worked example: the raised amylase that is not pancreatitis
Meet Frank Sullivan, 72, a retired joiner who lives alone, an AI patient in the Diagnosica library. He developed sudden, severe central abdominal pain about six hours ago, constant and unrelenting, with HR 112, BP 106/64 and RR 24 on arrival.
You’re told the oddity that matters, pressing on his tummy hurts surprisingly little more, which signals pain out of proportion to the examination findings. That line is the case’s hallmark, and it should stop you anchoring on surface clues.
There are temptations all over this history. He vomited and passed one loose stool early, and he thinks it’s a stomach bug or something he ate. This setup is one of the classic overnight misses.
He’ll also mention long-standing reflux when asked. It’s a planted herring, included in the script to be parked, not to be threaded into the main story.
Then the lab result lands, a mildly raised amylase. That’s the classic trap here, and relabelling the case as pancreatitis is how the work-up stalls.
What breaks the spell is a simple pivot, ask about his heart rhythm and medicines. He has atrial fibrillation, and his blood thinner was stopped two weeks ago for a tooth extraction and never restarted.
Normal-looking early numbers don’t rescue you either. A normal lactate early, a normal CRP early and a normal abdominal X-ray do not exclude this diagnosis. The decisive test in this case is an urgent CT angiogram in the arterial phase.
Worked example: the normal scans that are not reassuring
Toby Marchetti is 30, an AI patient, on his fourth bout of severe tummy pain in 18 months. Every CT, ultrasound and both scopes have been normal, he was told it was stress or drug seeking, and he half-believes it is in his head.
Here the red herring is the reassurance itself. One CT to exclude a surgical abdomen is fair, but the repeated normal CTs are the diagnostic signal, and re-scanning adds radiation, not answers. We wrote about rescanning without a question.
You only get the real clues if you ask on purpose. His urine goes dark red-brown, port-wine colour after standing. His limbs feel weak, especially shoulders and thighs, with tingling.
He has been anxious, agitated and sleepless with one confused spell. There was crash dieting and a heavy drinking weekend before this attack. During attacks his heart races, and the blood pressure climbs, here HR 124 and BP 172/104.
The case lists IBS, functional neurological disorder and lead poisoning as distractors. Normal inflammatory markers alongside severe pain should push you toward a metabolic cause rather than another scan.
The decisive test sits in plain sight. A fresh, light-protected urine porphobilinogen, sent during the attack, is the trigger to the diagnosis, cheap, and it hinges on somebody thinking to send it. The diagnosis in Toby’s case is acute intermittent porphyria.
If you want to feel the pull of the herring, and then defuse it, take his history yourself.
A three-question check to defuse a red herring
In one vignette study, asking doctors to identify the leading detail then generate an alternative diagnosis cut errors by nearly 50%, while a generic warning made no difference, report Arzy and colleagues.
- Does this detail explain everything? A true driver links the main findings, not a single lab or an old footnote. A mildly raised amylase is a classic trap in mesenteric ischaemia, it must not be misread as pancreatitis and used to stop the work-up. Application: the amylase does not explain his rhythm history, and a remote kidney stone is unrelated to the leg.
- What is my leading clue, and is it the herring? Name the one feature actually steering your top pick, then test whether it is the planted hook. If your pick hangs only on the amylase bump, you have likely named the herring; if it hangs on atrial fibrillation with his blood thinner stopped two weeks ago, you have likely named the real lead.
- What would my diagnosis be without it? Delete the shiny detail and re-run the differential. Drop the amylase and mesenteric ischaemia stays in play; if you are on your third normal CT for abdominal pain, the normal scans are the signal and the next diagnostic step is urine PBG.
This sits on top of ordinary differential building, where you are testing each candidate diagnosis against the data and using a differential mnemonic to keep misses low.
Do exam and simulator cases plant red herrings on purpose?
Yes. Case writers often include irrelevant but true history to mirror real consultations and to test whether you are distracted by a salient detail. Real patients carry old diagnoses and pet labels. We cannot speak for any particular exam board's writing process, but this pattern is common in teaching cases and simulators.
On Diagnosica, many of our cases carry a deliberately planted, labelled red herring such as childhood asthma, hay fever, an old kidney stone, occasional heartburn, or a smoker's cough. The AI patient says it plainly when you ask, does not link it to the complaint, and leaves the sorting to you.
Using red herrings in clinical cases reflects how wards work, because real patients volunteer side stories, and it trains the habit of filtering. Every case is written and signed off by a doctor.
You can ask for and receive examination findings as stated facts, but you cannot examine the patient yourself.
Practise spotting them
The quick case takes two minutes, the patient says one line, you get five clues one at a time, and you guess after each, which is good practice for noticing when an early clue is pulling you the wrong way. Try the clue-by-clue format in these clue-by-clue diagnosis games.
Full cases let you take the history by voice or text, order investigations, commit to a diagnosis and a plan, then see your scorecard. Every case is written and signed off by a doctor. It doesn't give advice about real patients, and it's not a diagnostic system. Treat every output as educational.
Drill: before you order anything, write down your leading clue, then run your three questions. If it holds, proceed. If not, park it and look for a better anchor.
Open the app and start a free case.
Educational use only. Not medical advice. AI-generated; verify clinically against primary sources.
Clinical review pending.
