Modern Medicine (Radiology)

#
1	This is a cat.	This is a hemangioma.
2	This is most likely a cat.	This is most likely a hemangioma
3	This is consistent with a cat.	This is consistent with a hemangioma.
4	This is most likely a cat, but get a follow-up picture to make sure it wasn't a baby tiger all along.	This is most likely a hemangioma. Recommend follow-up to document stability.
5	This is most likely consistent with a cat.	This is most likely consistent with a hemangioma.
6	This is likely a cat, but can't exclude a tiger hiding behind it way in the distance.	This is likely a hemangioma, but can't exclude malignancy, sarcoid, etc.
7	This is likely a cat. Why don't you take a look for yourself and stop bothering me?	This is likely a hemangioma. Recommend clinical correlation.
8	This is likely a cat, but get a saliva sample and send it in for genetic analysis. Better yet, kill the cat and dissect it.	This is likely a hemangioma. Recommend biopsy. Open biopsy may be required.

One of the biggest crap apps out there is the modern radiology report. In the real world (with the cat), anything other than statement #1 will get you laughed at. In radiology, as it's practiced in The United States at least, statement #1 is rare. Instead, we teach our residents and fellows, by our own weak examples, to be as non-declarative as possible.

Statements #2 and #3 are as declarative as most radiologists get. "I said most likely. What more do you want from me?!"

Statement #4 just passes the figurative buck to the next radiologist and the literal buck from the patient or 3rd party payer to the radiologist's practice. A great example of Arnold Kling's concept of Premium Medicine: Added cost with no (or negative) benefit.

Statement #5 (a real-world example!) manages to stuff 2 different hedge words into one sentence!

Statement #6 is the reason Bayes rolls in his grave every time a radiologist signs a report.

Statement #7 is basically saying, "Thanks for the money suckers! This report was useless." We have access to so much patient data these days that it baffles me to see this in reports. Of course, this doesn't apply to cases where we're reading in isolation and when the only history we get from referrings is "pain," or some random ICD code. This negligent absence of data in a requisition borders on (is?) malpractice. I've seen it in referrals my family members get from their doctors and it aggravates me to no end.

Statement #8, I don't even... For a cat/hemangioma?

Look, sometimes we have to hedge. Sometimes we are no better than Plato's cave captives, squinting at shadows with no idea of what's behind us. We know that two or more widely disparate entities can have identical imaging features. But when you know something can only be one thing, just say so. Save the patient some anxiety. And, save the rest of us some money by reducing unnecessary imaging.