Medical Gaslighting Is a Systems Problem, and You Can Train For It

Diagnostic equity and medical gaslighting training for medical schools, residency programmes and health systems — what actually causes patient disbelief, and how to build curriculum that changes clinical behaviour.

Thirty-three doctors. Seven years. Not one of them was cruel to me.

I want to establish that before anything else, because it is the part that gets lost, and because it is the finding that makes this trainable. The doctors who failed to diagnose me were not bad people having a bad attitude. Several of them were excellent clinicians who were kind to me in the room. The disbelief was not a personality defect distributed among thirty-three individuals. It was produced by the structure of the encounter, reliably, in people who did not intend it.

That is either a depressing conclusion or an operationally useful one, depending on what you do next. If medical gaslighting were about unkind doctors, the remedy would be recruitment. It is not about unkind doctors. It is a systems problem with identifiable mechanisms, which means it is a curriculum problem, which means you can do something about it in a training programme.

The mechanisms, specifically

Absence of evidence read as evidence of absence

Normal bloods, clear imaging, unremarkable examination. The clinically correct conclusion is that these particular tests found nothing. The conclusion that gets recorded is that there is nothing to find. That slide happens in the gap between the investigation and the note, and once it is in the note it propagates — because the next clinician reads the note, not the patient.

The prior-note cascade

By my fifth specialist, my file contained the word anxiety. Every subsequent clinician read it before meeting me. Anxiety is a reasonable differential on first presentation and a self-confirming diagnosis by the fifth, because now the patient is anxious — they have been undiagnosed for three years and are being sent away again, and anxiety is the appropriate response to that situation. The record had become the evidence for itself.

The tiring patient

I arrived, eventually, with a folder. I had read the literature. I had a list. Every clinician reading this knows exactly how that presentation lands, and knows the internal response is not really about the medicine. A patient who arrives with a folder is a patient who has already been disbelieved, and the folder is a rational response to the disbelief — but it reads as a warning sign, and the encounter is compromised before the first question.

Diagnostic momentum under time pressure

Twelve minutes. A patient with multi-system symptoms that do not fit a pattern. Opening a complex differential means a referral you cannot complete inside the appointment, and possibly an admission of not knowing. Closing it means the patient leaves and the list moves. The incentive gradient runs one direction, all day, for everyone.

Whose reports get discounted

The evidence here is uncomfortable and it is robust: women, racialised patients, patients with psychiatric histories, patients with obesity, and patients with existing chronic diagnoses have their symptom reports weighted lower. That is diagnostic inequity, it is measurable, and it operates in clinicians who hold no conscious prejudice whatsoever.

None of those five mechanisms requires anybody to be unkind. All five are teachable.

Why most training on this fails

It is framed as empathy. Empathy training in this domain mostly produces clinicians who are warmer while making the same diagnostic error, and who are now harder to challenge because their manner is good. Warmth is not the missing input. Warmth is what most of my thirty-three had.

It is delivered as a moral appeal, which puts a room of clinicians in the position of being accused. The predictable result is defensiveness, and defensiveness is the end of learning. Say "medical gaslighting" in a lecture theatre of registrars and watch the arms fold. I open differently, and I open by saying that nobody in my story was cruel — which is true, and which is also the only way the room stays with me.

It is disconnected from the record. The damage in my case was done substantially through documentation, and almost no curriculum addresses how you write a note about diagnostic uncertainty. That is a concrete, examinable clinical skill and it is nearly untaught.

And it has no measurement attached, so it cannot be improved and cannot be defended in a curriculum committee that has to choose between this and something with an assessment framework.

What the curriculum has to contain

Uncertainty documentation, taught as a technique. How to write a note that preserves the difference between "no cause found" and "no cause" — and that leaves the next clinician the room to look again. This is the highest-leverage hour in the whole programme and it is pure craft, not sentiment.

Reading a file for propagation. Give trainees a real, anonymised, five-clinician record and have them find the point where a working hypothesis became a fact. They find it. They are good at it. Then they recognise the pattern in their own notes, which is the transfer you want.

The folder patient, rehearsed. Actual practice with a well-informed, exhausted, slightly combative patient, with the debrief focused on the clinician's internal response rather than their outward manner. Naming that response out loud in a safe room is what stops it from operating unexamined.

The base rates. Time-to-diagnosis data for the conditions your trainees will most often miss, and the differential weighting evidence. Clinicians move on data. Give them the numbers and the moral argument becomes unnecessary.

A patient in the room who is also an analyst. Testimony alone is moving and does not transfer. I am in the room to show the mechanism from the inside — here is the sentence that closed the encounter, here is what the note did to the next appointment, here is what one different question would have cost you and saved me. That is a case presentation, not a story.

Measuring it, so it survives the curriculum committee

Ask the trainees, before and after, to write a note on a presented case of unexplained symptoms, and score the notes for how they handle uncertainty. That is a real, blindable, repeatable measure of the actual target behaviour.

Track referral and re-investigation rates for unexplained-symptom presentations in the cohort, where your data permits.

And ask your patients the one question that matters most and is almost never asked: were you believed. Not whether staff were courteous. Whether they were believed. It is the single most predictive item you can add to a patient questionnaire and it costs you one line.

The argument for doing this now

Diagnostic error is among the most expensive categories of harm in any health system, and delayed diagnosis of multi-system conditions is a large and poorly-measured slice of it. My own seven years cost somebody an enormous amount of money — dozens of specialist appointments, repeated imaging, two admissions, years of inappropriate treatment — and every pound of it was spent not finding something that was findable.

The clinical case and the financial case point the same direction here, which is rarer than it should be. And the training works on the mechanism rather than on the character of your staff, which is also why your staff will accept it.

I design curriculum and run this training for medical schools, residency programmes and health systems, with a measurement framework attached. See the engagement →