You get a diagnosis you want checked, so you book someone else. Different doctor, different practice, fresh eyes. That is the whole point.
Then the second doctor opens your file and reads what the first one wrote.
Which means the fresh eyes started from someone else’s conclusion. Not their observation, their conclusion, already written down and already framed. Whatever the first clinician decided mattered is now the first thing the second one sees.
Why this is hard to avoid
It is not that second opinions are worthless. They catch real errors constantly.
But a doctor reading a referral note is in a different position than one meeting you cold. The note tells them what to look for, which is helpful, and also tells them what the last person concluded, which is anchoring. Both arrive in the same paragraph and there is no way to take one without the other.
This gets worse the thinner the note is. A detailed record gives the second doctor raw material: what you reported, what was examined, what was ruled out. A two-word summary gives them only the conclusion, with nothing underneath it to reassess.
“Anxiety, stable” hands the second clinician a verdict. A note describing what you actually said, what was checked and what was excluded hands them evidence. Only one of those supports an independent look.
How often records fail to support their own conclusions
There is measurement on this, and it is not reassuring.
The US Office of Inspector General audits whether medical records support the diagnoses recorded in them. Across diagnoses it considers high risk, roughly 70 percent were not supported by the associated records, with some categories above 90 percent.
In one audit of 97 records carrying an acute stroke diagnosis, not one was supported as an acute stroke by the documentation. In 68 of those cases the patient had genuinely had a stroke, recorded in their history, with the current entry describing it as active.
Read that as a second opinion problem rather than a billing one. If a diagnosis sits in your file without the evidence that produced it, every clinician who reads it afterwards inherits a conclusion they cannot evaluate.
Getting a genuinely useful second look
- Ask for your actual records, not the referral letter. Test results, imaging reports, clinical notes. The referral letter is a summary of a conclusion; the records are what it was based on.
- Describe your symptoms from scratch rather than repeating the first diagnosis. “I was told it is X” starts the second doctor at the same place the first one finished.
- Ask what else it could be. A direct question that invites a differential rather than a confirmation.
- If the file contains a conclusion with no supporting detail, say so. “The note says stable, but I was not asked about that” is useful information.
Where this is heading
The underlying fix is making sure the evidence gets recorded alongside the conclusion at the time of the visit, rather than hoping it can be reconstructed later. That is what AI risk adjustment coding systems are increasingly built to support: prompting for the specific findings that justify a diagnosis while the clinician is still with the patient, so the record carries reasoning and not just a label.
A second opinion is only as independent as the record allows. Right now, that record is usually a conclusion with the working removed.
