The clinicians who miss the aortic valve are the fast ones.
That’s not what anyone expects to hear. We assume a missed diagnosis means somebody didn’t know enough. Mark Graber and his colleagues looked at 100 diagnostic errors made by internists and published the results in Archives of Internal Medicine in 2005. Inadequate knowledge was rarely the problem. Cognitive factors contributed in 74% of the cases, and the single most common one was premature closure: the clinician reached a diagnosis and stopped considering the alternatives. They knew the answer. They just picked it too early.
Take the swollen ankles. A 78-year-old comes in with bilateral leg edema. She’s gained six pounds. Her legs are heavy by evening. Every clinician who’s practiced more than a year sees the answer before she finishes the sentence: furosemide, compression stockings, cut the salt, follow up in a month. Most of the time that answer is right, which is what makes it dangerous. The pattern fires fast because it’s been rewarded a thousand times, and nothing about the pattern asks the next question: what’s pushing fluid into her legs in the first place?
Sometimes it’s a tight aortic valve. In adults over 75, aortic stenosis is common. A 2013 meta-analysis in JACC by Osnabrugge and colleagues pooled the population studies and put the prevalence of any aortic stenosis at 12.4% in that age group, and severe stenosis at 3.4%. One in 30. The valve narrows over years. The left ventricle thickens to push against it, the same way an arm gets bigger lifting weights. For a long time the heart wins. Then it can’t, and pressure backs up through the left atrium into the lungs and the right heart, and the fluid finally lands in the ankles. The edema is the last link in a chain that started at the valve.
If all you do is add diuretic, you’ve treated the puddle on the floor and left the faucet running. It’s what I tell APPs starting in cardiology. And the faucet, in this case, is a surgical problem. Severe symptomatic aortic stenosis has no medical therapy. No pill slows it. The diuretic makes the ankles better for a month while a patient with a stage D valve sits at home when she should be in front of a valve team.
So why does an experienced clinician write the furosemide script anyway? Because the shortcut and the expertise are the same machinery. Amos Tversky and Daniel Kahneman described anchoring in Science in 1974: whatever you land on first pulls every later judgment toward it. Pat Croskerry brought that work into medicine in a 2003 Academic Medicine paper and named the clinical versions, anchoring and premature closure among them. His point was that pattern recognition is how experienced clinicians think at all, and the speed that makes them good is the same speed that makes them close early. The 2015 National Academies report on diagnostic error concluded that most people will experience at least one diagnostic error in their lifetime. Those errors come from clinicians who know medicine, which is to say the rest of us.
I’ve done it. A woman with a prior bypass came into the emergency room with chest pain, and I anchored on a non-cardiac cause, because she’d already been fixed and her cardiologist at the office named it pericarditis. Her troponin climbed. The EKG changes were there but nondiagnostic. Her LAD graft was occluded. I knew the physiology of graft failure well enough. Knowing it didn’t help, because I’d already decided.
Another one, from the other direction. I got consulted for heart failure on a man with shortness of breath. I walked in and he was sitting with his legs crossed And reading a book. Comfortable people aren’t generally in overt heart failure. His story was six to eight months of gradual exertional dyspnea, and he slept flat with no trouble. Then I looked at the med list and found amiodarone, started the year before. Amiodarone lung toxicity. The referring provider had matched dyspnea to heart failure and stopped. The pattern was reasonable. It was just the wrong one, and nothing in the workflow made him check.
The standard advice has the same gap. Every review on cognitive bias says the same things: keep a broad differential, look for disconfirming evidence, consider the alternatives. All true. All useless at the end of a full clinic day with three patients waiting, because “consider the alternatives” has no trigger. It doesn’t tell you when. Croskerry’s answer was what he called a cognitive forcing strategy: a fixed question, tied to a fixed moment, that you ask whether or not you feel like you need to.
Mine is attached to the order. Before the diuretic goes in, before the antibiotic, before the PPI, I ask what’s upstream of this. Not “is there anything else,” which the brain answers “no” to in half a second. What is causing the thing I’m about to treat? For edema, that means I’ve listened for the murmur, asked about angina, exertional dyspnea, and syncope, checked the neck veins, and felt the carotid upstroke before I’ve written anything. If there’s a systolic murmur nobody has chased down, she gets an echo before she gets a water pill. And if the differential is wide, I run it out loud with the patient, in plain language. Saying “this could be the valve, the kidneys, the liver, or the veins” takes ten seconds and makes it much harder to close on the first one.
A protocol wouldn’t get you there, since a protocol for edema would still say furosemide. What I’m describing is a habit, and habits get built the same way pattern recognition does: by repetition, ideally on cases where the wrong answer costs nothing. That’s why the clinical reasoning coach in Voice2VoiceSim walks students through the differential and won’t hand them the diagnosis, so the “what’s upstream” question gets asked a few hundred times before a real patient is on the other end of it.
Croskerry named the trap in 2003. Graber counted it in 2005. Twenty years later, the diuretic still goes in first.