Brady spells out how to write bad radiology reports

2021 01 27 17 02 7498 Brady Adrian 400

Writing good radiology reports is a vital skill, but how best can you get that message across? Showing what bad ones look like -- and explaining to trainees in an ironic way how not to write reports -- is a fun and effective way to do it, according to Dr. Adrian Brady.

Dr. Adrian Brady, first vice president of the European Society of Radiology.Dr. Adrian Brady, first vice president of the European Society of Radiology.

Brady, a consultant radiologist from Mercy University Hospital in Cork, Ireland, sampled the best of the worst in a whimsical masterclass on report-writing that was part of a webinar streamed by the British Institute of Radiology (BIR) on 28 April.

The talk comprised a crash course in vague, verbose, sloppy, defensive, and imprecise radiological prose found in reports that confuse the referrer, don't state what the findings mean, fail to commit an opinion, and offer nothing useful for patient management.

Anyone can "consistently write lousy reports" if they stick to the following three founding principles -- the very opposite of what should actually be done, Brady emphasized:

  • Avoid exposing yourself to risk (i.e., the risk of somebody finding out you're wrong or suing you for it).
  • Do minimal additional work and divert as much work as possible to others.
  • Shift decision-making and responsibility for patient care onto others.

How NOT to do it

Brady gave eight tips on writing bad reports, which in reality, must be avoided at all costs:

  1. Ignore the question asked. "If the physician asks you a very specific question, don't attempt to answer it, give a generic recitation of the details of what you see in the study and move on," said Brady, adding that referral notes and requests could be odd -- such as "deodorant can in rectum" -- and answered very directly, as in "you are correct."
  2. Don't try to make a report readable. "If they're unreadable, referrers will probably not bother and therefore you won't be exposed to any risk," he noted. Hence, opaque MRI reports of "such nonsense that you lose the will to live" halfway through and hopeless monographs tediously starting every sentence with "there is." In cases like these, "the radiologist has very successfully achieved the aim of writing a bad report and preventing anybody ever reading it to find something with which to blame him/her."
  3. Don't correct typos, grammatical and sense errors, etc. As literary giant Ernest Hemingway once said, "The first draft of anything is shit," Brady quipped. "Not correcting typographical errors will ensure a report is even less likely to be read and ensure the radiologist is safe." Or it can just be "mind-boggling," like a report on a toe that read "background mild metatarsus penis various."
  4. Use abbreviations, acronyms, and unfamiliar terms. "Using these can confuse the referrer no end and make sure they won't blame you for anything." Sometimes referrers do this with clinical information. One chest x-ray request stated, "81-year-old male with acute hypoxia after IVF." "I hope that meant intravenous fluids, but I'm not altogether sure!" Brady said.
  5. Embrace ambiguity. "Don't say what you mean or don't give it any meaning in your report." Baldly stating a kidney measurement without clarifying detail will leave a nonspecialist with no clue whether it is huge or tiny, or if its size is in any way significant.
  6. Sit on the fence. "A skill that every radiologist learns early and adopts to a varying extent as their guiding principle," he said. "It's been said that the hedge is the tree of our speciality and that a radiology logo should depict a weasel eating a waffle under a hedge." But "avoiding diagnostic commitment," sometimes wrapping it in elaborate language, can take years to perfect, Brady said. He listed words and phrases for hedging/ fence-sitting, including "no evidence of," "apparent," and words to avoid saying normal (so-called "normal imposters"), including "unremarkable," relatively," "within normal limits." He singled out "clinical correlation recommended" as the "granddaddy" of all pointless phrases in shifting responsibility onto the clinician. To which an insulted referrer might counter, "recommend examining the images carefully."
  7. Report inversion (or, "assuming we're paid by the word"). Reports with long conclusions that effectively repeat the findings are hard to read.
  8. Attack is the best form of defense, so criticize the referrer. "Blame them for the fact we haven't been given the appropriate information," he said.

On a more serious note, Brady, warned of the problems with communication failure in general. He said it was the fourth most common reason for radiologists in the U.S. being sued and 60% of these cases were due to a failure to highlight an urgent or unexpected abnormal finding and to emphasize it appropriately in reports.

"If our reports are incoherent, rambling, and verbose -- and if it's impossible for the referrer to clearly understand what is most important in them -- then we have failed to communicate, and we have written bad reports," he commented.

Pitfalls of voice recognition

Brady highlighted the pitfalls of using voice recognition (VR) -- an algorithm that predicts the likelihood of any given word on the basis of the two previous words.

"It's pretty accurate: about 98%. But that means two in every 100 words of a report will be incorrect in the best hands. That can be a lot of very significant words in a long, complex report," he said.

VR has a "significantly higher" error rate than old-fashioned manual transcription, Brady continued, citing research that showed how frequently unrecognized mistakes were signed off.

"Certainly, editing our own reports takes a bit of time -- it generally increases the time by about 20%-30% over the time required to interpret the study and to initially dictate it," he said. "But we are failing to communicate adequately, and we may be in danger of producing nonsense or incorrect reports if we don't do this."

Make your reports simple and clear, include what matters, do not include the irrelevant, he concluded.

Editor's note: A second article on the BIR webinar, featuring the other speakers' talks, will appear in May.

Page 1 of 1244
Next Page