Why Doctors Miss Pulmonary Embolism

Every acute specialty has a diagnosis that functions as its conscience. For emergency medicine, a strong candidate is pulmonary embolism — the clot that travels to the lungs' circulation. It appears with grim regularity in coroner's findings, in case reviews, in the literature on diagnostic error, and in the 3am thoughts of clinicians replaying a shift. Not because doctors don't know about it. Quite the opposite: PE is drilled into every clinician from medical school onwards, precisely because it is common enough to matter, lethal enough to fear, and slippery enough that knowing about it is not enough.

That's what makes it worth an essay. The ways PE gets missed are not exotic lapses; they are ordinary reasoning mechanisms producing a catastrophic result. Walk through them and you have a tour of how diagnostic error actually happens — anywhere.

A disease with no face

The first mechanism is the disease itself. PE has no obligatory presentation. The textbook offers pleuritic chest pain, breathlessness, and a swollen leg; real cases offer any of those, or one, or none. PE presents as sudden breathlessness with a clear chest. As a faint at the bus stop, recovered by arrival. As unexplained tachycardia. As 'I just can't catch my breath properly since last week', said calmly, while walking. As mild pleuritic discomfort in someone young and fit. Occasionally as collapse and cardiac arrest, where the diagnosis announces itself too late to argue with.

The protean quality means PE almost always wears the costume of something more common. Breathlessness reads as infection or heart failure or anxiety; pleuritic pain reads as musculoskeletal or viral; the post-faint patient reads as dehydrated. In each costume, the more common explanation is genuinely more probable — which is exactly the trap. The clinician who diagnoses the common thing is usually right. PE lives in the word 'usually'.

The false reassurances

The second mechanism is a set of findings that feel exclusionary and aren't. They deserve naming individually, because each one has its own casualty list.

Normal oxygen saturations. The intuition says a clot in the lungs must drop the oxygen. Often it does. Frequently — especially in younger patients with good reserve, and in smaller or more peripheral emboli — it doesn't, or not yet, or only on exertion, which a patient sitting on a trolley is not doing. Normal saturations at rest lower the probability somewhat. They exclude nothing. The number's authority — objective, digital, reassuringly green — vastly exceeds its evidential weight in this disease.

A normal heart rate, a normal ECG, a normal chest X-ray. Each has the same shape: abnormal findings support the diagnosis, but normal findings do little to retire it. The chest X-ray in PE is classically normal or near-normal — its main role is excluding the mimics. The reassurance these tests generate is real psychologically and weak evidentially, and the gap between those two is where the misses grow.

An available alternative explanation. The most dangerous false reassurance isn't a test at all — it's the presence of another diagnosis that fits. The patient with known anxiety whose breathlessness is anxiety, the post-operative patient whose tachycardia is pain, the patient with a chest infection and a clot. 'Satisfaction of search' is the formal name: once the mind finds one explanation, it stops looking for the second. PE, which loves to coexist and to follow exactly the situations that produce alternative explanations — surgery, immobility, illness — exploits this mercilessly.

Youth and fitness. The young, healthy patient triggers a prior — 'people like this don't have serious disease' — that is statistically defensible and case-by-case treacherous, particularly with risk factors that hide in plain sight: the long flight, the new medication, the family history nobody asked about, the recently immobilised limb.

The cognitive mechanics

Underneath the false reassurances run the standard engines of diagnostic error, and PE is where they're easiest to watch.

Anchoring: the first frame — 'sounds like a chest infection' — set at triage or referral, bending every subsequent finding towards itself. Confirmation weighting: the crackles that support the infection story counted; the heart rate that doesn't, explained away as fever. Premature closure: the label applied, the reasoning concluded, the discharge letter already half-composed in the mind. Base-rate seduction: 'it's usually not PE' deployed as if usually were a safety margin rather than a description. None of these is a character flaw. They are the operating characteristics of fast pattern-matching — the same machinery that makes experienced clinicians efficient — running on a disease selected, almost adversarially, to exploit it.

This is why the specialty's answer to PE is not 'think harder'. It's structure. Validated probability scores to make the risk estimate explicit rather than felt; rule-out criteria for the genuinely low-risk; the d-dimer blood test deployed after probability assessment rather than as a reflex; imaging gated by the arithmetic rather than by anxiety. The structure exists for a precise reason: it is a bias-correction device. The score doesn't know about the anchor, doesn't share the satisfaction of search, and doesn't find the patient's youth charming. Used honestly — and 'honestly' means including the gestalt item, documenting the estimate, accepting the answer — it catches exactly the cases intuition was about to wave through.

And yet structure has its own failure mode, which honesty requires naming: the over-testing spiral. The d-dimer is sensitive and unspecific — sent indiscriminately, on the mildly breathless and the barely symptomatic, it returns positives that compel scans that find incidentals that beget more tests, while irradiating many to find few. A department that scans everyone hasn't solved the PE problem; it has traded missed clots for a different ledger of harms. The discipline lives, permanently, on the ridge between the two error types — which is the actual lesson of the disease.

What the misses teach

Looking across the mechanisms, three durable lessons emerge, none confined to PE.

Reassurance has a structure, and it can be counterfeit. The findings that feel most exclusionary — the green number, the normal trace, the available alternative — must earn their exclusion power evidentially, not psychologically. The question 'what has actually ruled this out?' has a specific answer or it doesn't.

The second diagnosis is a discipline. The most reliable protection against satisfaction of search is the deliberate, slightly artificial habit: this explanation fits — what else could also be true, and have I looked? It costs thirty seconds. Its absence costs more.

And probability tools work only as honest inputs to judgment, not as armour after it. Scores gamed to justify the decision already made — the gestalt item quietly zeroed, the borderline history rounded down — reproduce the original bias with better paperwork. The structure protects exactly to the degree it's allowed to disagree with you.

What this means

PE keeps its place in the specialty's conscience because it concentrates, in one disease, everything that makes diagnosis hard: a presentation with no face, tests whose normality means little, a habitat full of alternative explanations, and a victim profile that disarms suspicion. The misses it produces are not failures of knowledge — the knowledge is universal — but failures of mechanism, which is precisely why they recur and why no individual brilliance retires them. The honest response is the one the specialty has been building for decades: explicit probabilities, structures licensed to overrule intuition, deliberate searches for the second diagnosis, and a standing respect for how little a normal number can mean. The clot doesn't care how experienced you are. The structure is how experience gets to count anyway.