The Radiology Report

“’Normal’ is one of the most powerful words a radiologist can use”: Curtis P. Langlotz MD PhD, Professor of Radiology, Stanford University

After I used “clinically correlate” thrice in a row in my report, the attending radiologist asked, “How would you feel if the referring clinician said on the requisition for the study “correlate with images”? When you ask them to clinically correlate, you’re reminding them to do their job.”

I had been a radiology resident for six months – too soon to master radiology but not too soon to master radiology’s bad habits. I had acquired several habits, tics to be precise. These tics included saying “seminal vesicles are unremarkable,” which I stated remorselessly on the CT of the abdomen in males, even if the clinical question was portal vein thrombosis, sending, I suspect, several young men to existential despair. But the tic that really got under my attending’s skin was “cannot exclude.”The attending was Curtis P. Langlotz, the author of The Radiology Report, a book about writing effective radiology reports.

Ubiquitous in clinical care, and sometimes parody, radiology reports are enigmatic. What’s most striking about radiology reports is their variability. Reports vary in length, tone, precision and frequency of disclaimers. Reports vary in strength of recommendations for further imaging.

One radiologist may say “small pancreatic cyst, recommend MRI to exclude neoplasm.”Another, aware that the patient may cross St. Peter’s gate sooner rather than later, may bury the findings in the bowels of the report, hoping the clinician will spots its irrelevancy. Yet another, eager to be non-judgmental,might say “small pancreatic cyst, likely benign, but MRI may be considered if clinically indicated,” which, Langlotz notes, is vacuous because with pancreatic cysts there’s nothing clinically the clinician can anchor that recommendation on.

Radiologists, conscripted to ail uncertainty in diagnostic medicine,have responded by introducing their own uncertainty.We remind physicians that “CT does not exclude ligamentous injury” on CT of the cervical spine which is negative for fracture. The pedagogic value of this disclaimer is lost by the second time it is read.

Why do we mention metaphysical truisms such as “sub segmental pulmonary embolism is not entirely excluded with absolute certainty?” This is part honesty and part disingenuity. Uncertainty is a fact of life. But radiologists know that clinicians know that no pathology can be excluded with absolute certainty. Stating this truism throws the ball back in their court, legally. The radiology report is a legal document.

The Radiology Report has been written in the way Langlotz hopes his colleagues write their reports. The book is short, readable, on point, and, importantly, the author takes a stand. He doesn’t hedge. Langlotz advocates standardized reporting, unapologetically. One may argue with the stand, and he knows that I often do(I argue with everything). But here is the point. Because Langlotz takes a stand, the conversation moves. The stand anchors the narrative.

Medical decision making is similar – you have to take a stand. Whileit is important to be right, it is better to be wrong than vague. Because when you’re wrong about a diagnosis, at least the clinicians know which diagnoses are wrong so that they can move to the next. By being vague, no one knows what should not be suspected. As I remind radiology residents before their call, “make a decision.”

Langlotz’s prescription for useless radiology reports is simple. Hedge less, be consistent and say “normal.” Like a date too polite to decline the second date, radiologists make all sorts of excuses for “normal.” Such as “liver is unremarkable.” Or “no CT evidence of abscess,” leaving one wondering whether they should wait for the non-CT evidence of abscess. Or, “no definite pulmonary embolism is seen” so that we’re all clear – radiologists don’t hear definite pulmonary emboli, let alone indefinite emboli.

Perhaps the most mysterious word in radiologist’s lexicon is “prominent” such as “pulmonary arteries are somewhat prominent.” Not even radiologists know what “prominent” really means. Is it that the pulmonary arteries are seen? Famous? Loud? Is Donald Trump “prominent”? Or is the structure big and we’re too gentlemanly to say so? Or is it another word that reflects our discomfort that normal and abnormal overlap?

Many radiology reports are baroque, clunky, and literal interpretations of the idiom “picture says a thousand words.”This is not just habit, but culture. I’ve seen many equate pith with laziness and verbiage with work ethic. Thus, “the liver demonstrates a normal size and contour, intact vasculature, and is devoid of a focal mass” has replaced “the liver is normal.” Bones are “osseous structures” rather than “bones.”“Osseous structures reveal mild degenerative disease in the lumbar spine” leaving oneshellacked –what a revelation. Why was it concealed?

If pith is underused, pith is also under appreciated. I once read a CT in a middle-aged man with abdominal pain. I said “no abnormalities in the solid viscera, hollow organs, bones or vessels.” I thought all bases were covered.Minutes later the referring clinician called. “You said the solid viscera were normal, but you didn’t say anything about the pancreas. We’re suspecting pancreatitis.” I amended the report: “I have looked at the pancreas and seminal vesicles and they are unremarkable.”

Radiologists see far more on cross sectional imaging, such as CT, than projectional 2 D imaging, such as chest radiographs. Thus, radiologists have become the chroniclers of age, as well as the finders of pathology. Everyone with a college degree, and nearly everyone without one, has “mild degenerative disease of the lumbar spine.” Show me a nonagenarian who doesn’t have “perivascular hypodensities suggestive of small vessel ischemia” and I’ll show you an incorrectly labelled CT head. And then there is the “somewhat atrophic uterus” – a finding which reassures the reader that the radiologist has a functioning visual cortex.

The greatest source of variability in radiology reports is agreeing about which of the epidemic of clinically insignificant findings should be mentioned in the report. I veer towards relevance. I’m happy to make a judgment call. Some veer towards completeness.

Langlotz doesn’t ask radiologists to rid disclaimers altogether. He identifies many situations where honesty about the imprecision is apt. For example, when a fracture is dubious, not slam dunk, on a radiograph,it is reasonable asking that findings be correlated with tenderness at the site. Context maketh a radiology report.

The Radiology Report covers more than smart reporting. There’s a nifty section about probabilities. And a hilarious chapter about speech recognition errors, my favorite – “right internal jugular coitus” when the radiologist said “right internal jugular cordis,” giving the erroneous impression that radiologists are watching Fifty Shades of Grey whilst dictating.

Langlotz has done an E.B. White for Radiology. I advise radiology residents to read The Radiology Report in their first year and after overnight call.  Non-radiologists should read this book, too, and hold reports to the standards he describes.

I credit Langlotz for my visceral dislike of “cannot exclude.”I also no longer say “seminal vesicles are unremarkable.” If you have nothing nice to say about someone’s seminal vesicles, say nothing at all.

Saurabh Jha is a radiologist based in Philadelphia and a contributing editor for THCB.

Categories: Uncategorized

Tagged as:

13 replies »

  1. The problem with calling something normal on an imaging study is that the reader doesn’t always know what the limits of the imaging modality are. For example, to call a gallbladder “normal” on a CT scan would be misleading, since many gallstones are not visible on a CT. Likewise, small chip fractures can be missed on MRI. Nondisplaced fractures, especially stress fractures, can be invisible on xrays. Solid components of a soft tissue lesion, critical to the correct diagnosis, may not be see on an ultrasound. All of these facts are well known to radiologists, but clinicians have variable levels of awareness of these issues. I imagine some of the awkward or abstruse reporting styles we see have been shaped by our collective experience as a profession of getting into trouble for implying more certainty and diagnostic power than we really have. While brevity and pith are good aspirational goals, we need to be aware that we have a critical role in clarifying what the limits of our images are, and sometimes this takes more verbiage than we’d like.

  2. I agree, Dr. Hadler, that radiologists should be clinical colleagues who share risk. Furthermore, I don’t think a report need be rendered for every single ICU film, every single CT of the head, etc. Perhaps the greatest value of a radiologist is at multi-disciplinary conferences where the imaging expert integrates information.

    I do not see this happening in the foreseeable future for reasons that go beyond the incentive structure.

  3. Thanks for reading. Variability is a very interesting issue. I have a different take on variability which I might articulate in these pages sometime. Suffice to say that if variability is unscientific, uniformity is no more scientific (I recall Aristotle’s fallacy of golden mean)

  4. Doctor Hadler, why do we need radiologists to “read/diagnose” pictures? I would much rather have my specialty specialist read the picture and be better able to relate it to his experience of his speciality and my condition.

    The specialist will have the benefit of seeing first hand if the picture is at all relevant to past diagnosis and patient symptoms. I’ve had hip pictures, one where I could read the picture myself, especially when it confirmed my pain, the other was not so clear and the discomfort not so acute or constant, so I relied on my hip surgeon’s knowledge to say, “Yes, I’ve seen that before, many times”.

  5. A fine post that brought a fine smile when read on a bad day.

    I no longer will hold that young radiologist with disdain when he reads a CT or MRI of the head and tells me the seminal vesicles are normal. 🙂

  6. Dear Dr Jha,

    This is an interesting post. I read a study a few years ago about variability in radiology diagnoses. I do not recall the source, but it seemed to be respectable. Someone had radiologist read images that had positive findings. They gave the same images to the same radiologists later and they changed their diagnoses about 30% of the time.

    To my way of thinking, that can only happen in one of three ways. 1)the radiologists are not accountable to anyone who ever checks the quality of their work. If the patient gets a second opinion at some place like the Mayo clinic, the errors are seen clearly. 2) the radiologists are really not studying the images very well. Or 3) there are generally no standards for reading images.

    I visited Virginia Mason once and saw first hand the good job they are doing in creating measurable standards for reading certain images. Perhaps that’s uncommon.

    Am I wrong?


    Tom Emerick

  7. Loved your post.

    At the bottom, at the root of all images, we don’t have digital. We have wave functions that can be both zero or one. Everything is not only more vague than digital, it is less certain than analog. Since Schrodinger, we have been in a mess. Even mathematicians argued for a hundred years about what a derivative really is.

    Keep all the beautiful nuance and uncertainty and eqivocation. Standardization only fools the reader. “I’m more certain than I really feel.” If the images yell, yell. If they whisper, you whisper too.

  8. I don’t think non-clinicians can appreciate how well stated and relevant your post is to modern practice (modern meaning reliance on imaging). Between the satirical lines, you hit the bullseye. Well done. My only criticism? The cost of the book. Take a look at the Amazon link.

    I read Dr. Hadler’s comment as well, with keen interest. I have to add, the dysfunction goes both ways. Radiologists need to use language more carefully, but bedside clinicians need to apply probabilistic thinking and stop ordering tests unnecessarily. The quagmires we often find ourselves in stem from our ordering mistakes and making.


  9. Dear Dr Jha,
    For this clinician and clinical educator, the fact that radiologists are semantically challenged is the least of the failings of your specialty. The most serious is that you have allowed yourselves to be considered the purveyors of an assortment of “diagnostic tests”. True, some of your tests require expensive equipment, extensive operator training, and experience in interpretation. True, as well, your specialty commandeers an important segment of our medical “free market” (https://thehealthcareblog.com/blog/2016/01/30/there-is-nothing-free-about-the-health-care-market/ ). But the price you are paying is to be considered “providers” of the reports you bemoan, providers who usually are physically (even off-site, sometimes very far off-site) and emotionally distant from the bedside. And the price practitioners and patients pay is to be held hostage to those reports.
    For most of my career, radiologists were consultants. For most of my career, a stop in the basement where the imaging troglodytes are often housed was part of the day. I would knock on the door of the “reading room” and ask if the oracle was in? Tests are ordered solely to test clinical hypotheses generated by the patient’s presentation. If I’m ordering a serum calcium, I need to be reassured that the laboratory meets quality standards but I don’t need to discuss the result with the “provider.” When I’m ordering an imaging study, I am requesting a consultation.
    Many an American medical center has become a sprawling enterprise with providers housed separately, often at a distance. The clinician who called you to ask about “pancreatitis” was trying to circumvent the barriers to care that have been erected by well-intended experts in “systems”, experts who have no idea what it means to care. Assigning practitioners to boxes in an organizational chart will grow the organization but don’t expect patients to be better off as a result.
    Nortin M Hadler MD MACP MACR FACOEM