Daniel Kahneman showed that expert judgment is systematically biased. In obstetrics, the most dangerous version is this: a woman who just had a cesarean comes in with symptoms, and her doctor explains them away. Anchoring. Premature closure. Normalcy bias. They are not psychology abstractions. They are mechanisms of maternal death. This week on ObGyn Intelligence.
How cognitive bias turns warning signs into background noise, and what to do about it.
She is two days postoperative. Cesarean delivery, uncomplicated by the operative note. She calls the triage line reporting shortness of breath and chest tightness. The nurse asks about her pain medications. The resident documents atelectasis, encourages deep breathing, and tells her to follow up with her doctor in a week.
She dies 18 hours later of pulmonary embolism.
This is not a hypothetical. Variants of this scenario appear throughout maternal mortality reviews on both sides of the Atlantic. The diagnosis was not difficult in retrospect. It was missed because the clinician had already decided what was wrong before fully evaluating the patient.
How Judgment Works, According to Kahneman
Daniel Kahneman spent a career documenting what the human mind does when it believes it is reasoning but is actually shortcutting. His framework, developed with Amos Tversky and extended in Thinking, Fast and Slow, distinguishes two modes of cognition. System 1 is fast, automatic, pattern-based. System 2 is slow, deliberate, effortful. The problem is that System 1 is always running, and it is often wrong in ways System 2 never gets the chance to correct. (1)
Medicine trains clinicians to trust pattern recognition. That training is not wrong. A seasoned obstetrician who sees late decelerations and immediately calls for an emergency cesarean is using System 1 appropriately. The trouble begins when System 1 generates a diagnosis before the data warrant it, and System 2 never asks whether the pattern is actually fitting the patient in front of us.
Kahneman called this substitution: we replace the question we should be asking with an easier one. What dangerous condition is this patient showing signs of? becomes What do postoperative symptoms usually represent? The second question is easier to answer. It is also the wrong question.
What follows is an extensive insight into 4 biases we must be aware of:
Anchoring bias — fixing on the first piece of information (she just had a cesarean) and insufficiently updating when new symptoms arrive.
Premature closure — reaching a satisfying explanation and stopping the diagnostic search too early.
Normalization of deviance — abnormal findings gradually accepted as expected, drifting collectively across shifts and teams.
Normalcy bias — assuming that because things have been stable, they will remain stable, causing clinicians to underweight new deteriorating data.
ObGyn Intelligence: Safety analysis, the evidence critique, and the verdict are below -- for subscribers who want the full picture.
