Radiologists’ Cant

January 22, 2016

Radiologists and referring clinicians should learn the language of voice recognition.

I’ve written a couple of pieces about voice recognition software in the past. I imagine anybody who’s had the slightest brush with such technology has a thought or three on the subject.

Liked or loathed, it’s not going anywhere. I can’t imagine any set of economic circumstances where human transcriptionists would reenter the picture. That said, the voice rec industry doesn’t seem in much of a hurry to improve its accuracy on account of us users. The only variable that would seem to be left under our control is how we react to it.

Once upon a time, I considered my dictated radiology reports the closest I would come to producing works of art. The most minor of typos would vex me, and some days the software seemed bound and determined to screw up every single thing I said. I couldn’t wrap my mind around how some of my colleagues signed off their stuff with nary a proofreading glance, not a care in the world.

It’s amazing how a few years and a gazillion dictated cases can change your perspective. I’m not saying I’m one of those carefree folks mentioned above, but I have found myself focused less on my own dictating habits and the foibles of the VR software, and more on the outsiders who occasionally reach into my insular little radiological world to request an addendum regarding a transcription error.[[{"type":"media","view_mode":"media_crop","fid":"45174","attributes":{"alt":"voice recognition","class":"media-image media-image-right","id":"media_crop_8978591727525","media_crop_h":"0","media_crop_image_style":"-1","media_crop_instance":"5128","media_crop_rotate":"0","media_crop_scale_h":"0","media_crop_scale_w":"0","media_crop_w":"0","media_crop_x":"0","media_crop_y":"0","style":"height: 190px; width: 200px; border-width: 0px; border-style: solid; margin: 1px; float: right;","title":"©Anya Ku/Shutterstock.com","typeof":"foaf:Image"}}]]

Granted, some of these things need fixing. If, for instance, you dictated in both the body and impression of your report that there was “No pulmonary embolism,” but the VR software quietly omitted the “no” from the body, a clinician noticing the discrepancy would understandably want it clarified.

However, most of the inquiries I get are far more trivial, to the point that I find myself wondering how someone reading the typo could possibly not understand what the intended wording was. The erroneous word rhymes with the intended one, for instance, or even a homophone…or the context leaves no uncertainty as to what word belonged there.

Human brains are clever bundles of mush, after all, and most of us have seen more than a few examples, on Facebook if nowhere else, of how we practically subconsciously fill in the blanks to derive meaning. Sentences missing all their vowels, for instance, or with rearranged errors in every single word…yet a moment’s attention is all it takes for comprehension to occur.

Indeed, there are more than a couple of historical examples of intended linguistic alteration, for game playing, exclusivity, or downright sneakiness. One that leaps to mind is (Cockney) rhyming slang. Far better described by the Wiki-gods than I, it uses plain old English words and phrases to turn straightforward statements into what seems like complete gobbledygook to someone unfamiliar. Yet it causes not a moment of confusion when spoken to one familiar with the slang…rather, it might add a sense of comradery that one is dealing with others “in the know.”

I suspect, in the fullness of time, as more and more radiologists weary of making verbally-pristine reports (which are, of course, reimbursed an average of $0.00 more than error-ridden ones), and more clinicians tire of trying to track down rads for explanations/addenda, we’ll gradually evolve our own sort of understanding as to which erroneous words and phrases tend to get transcribed for others. A physician reading a report will thus have no confusion at all when encountering a turn of phrase that would seem meaningless to an outsider.

Not quite a modern equivalent of the “Thieves’ cant,” from days of yore…but if it occasionally trips up snooping regulators and ambulance chasers combing over our reports in the hope of making trouble for us, it’ll be a bit of icing on our cake.