Elements of Radiological Style
last modified 6-Jul-2020
This is my personal style guide for reports, intended primarily for radiology residents and fellows, from the perspective of a body radiologist. Sometimes lost during a busy radiology rotation is the essential notion that the report is the primary “product” of a diagnostic radiologist. The report often constitutes your entire contribution to a patient’s care. In many settings, the report is also the only way the referring clinician knows you. Your diagnostic expertise is of little value if you cannot communicate effectively.
Style is a matter of opinion, and not all radiologists (and more importantly, referring clinicians) will agree about what is correct or important; witness the divide between structured and narrative reporting. However, clarity is always the goal. An important part of radiology training is the opportunity to observe and critique different attending reporting styles, so that you can make informed decisions about how best to report radiological findings.
Spelling, grammar, and diction matter.
Little did I imagine that being an academic radiologist required teaching English composition. But obvious errors in spelling, grammar, and diction convey sloppiness and undermine confidence in the accuracy of the radiology report. Unfortunately, imperfect speech recognition systems render reports even more vulnerable to comedic or confusing errors. These errors are at best embarrassing, and at worst harmful to the patient. Get into the habit of reviewing your report with a critical eye before signing off; don’t rely on your attending as an editorial safety net, especially in the latter half of your residency. Keep in mind that it is typically more difficult to detect errors in your own reports than in the reports of someone else.
Discard excessive technique verbiage in diagnostic studies (excluding fluoroscopic and interventional reports).
Some details about the technique in a diagnostic study are essential for understanding or required for billing, such as the fact that intravenous contrast was given. The technique section is also appropriate for patient radiation doses, adverse reactions to contrast, and significant limitations or deviations from standard technique. However, other routine details are often unnecessary and merely clutter the report (e.g., a CT scanogram was obtained, reformats were generated, images were reviewed on PACS). Long and detailed technique sections copied from report templates (common in body MRI) are also often inaccurate, as they are tedious to verify, and minor technique variations are common. Anyone who really needs to know all the technical details of the study will probably require the DICOM image set anyway.
Beware of canned “normal findings text” in report templates.
Canned text in report templates may be a useful checklist for imaging findings. But you must be careful to edit or remove canned text when it conflicts with findings in the study. This requires meticulous editing habits. I prefer not to have canned normal findings text in the templates I use most often.
In the findings section, prefer description to single-word summaries.
A useful mental model is to consider a hypothetical radiologist who reads your report, and subsequently reviews the study; if your report is well-written, the radiologist should not be surprised by any imaging findings. However, sometimes I see reports with one-word descriptors in the findings, such as “cholelithiasis,” with no other qualifiers; this appears to be more common in structured reports. A single word usually encompasses a wide range of severity, raising the likelihood of “surprise.” To avoid this, simply describe (e.g., “single punctate gallstone” or “numerous small stones distend the gallbladder”). Brief one- or two-word summaries may be acceptable in the impression section.
When in doubt, provide image numbers.
For small or subtle findings, or in cases where a finding may be confused with others (e.g., a growing nodule among other nodules), take a moment to provide an image number; future readers will thank you. Make sure also to specify the series number if there is more than one main series. If you’re wondering whether a finding deserves an image number, err on the side of providing it.
For a mass which has changed in size, give the measurement on both the current and prior study.
Clinicians often want to know how much a mass has changed in size. Even if the prior report is readily available, the measurement in that report may not be obvious, especially if the prior report was dictated in a different style by someone else. Furthermore, the mass on the prior study may need to be re-measured to provide an accurate comparison. Save the clinician a headache by providing measurements on both the current and prior study. Remember that in some cases of slowly growing masses, it is important to compare to more than one prior study.
Do not dictate measurements at sub-millimeter precision.
This is all about significant digits; no radiology modality has reliable sub-millimeter precision. I don’t care if the PACS or ultrasound machine reports a distance of “7.6 mm” – round it to 8 mm in your report. For infiltrative or ill-defined structures, millimeter precision may be inappropriate; consider rounding to the nearest centimeter. Note that “2.0 cm” implies a higher level of precision than “2 cm.”
When giving comparison measurements, or several measurements in the same sentence, use the same units.
Clinicians often read reports quickly, and may overlook the change in units if you say that a nodule has changed in size from 3 mm to 1 cm. It’s better to say it has increased in size from 3 mm to 10 mm, or 0.3 cm to 1.0 cm.
Some radiologists will disagree, but to me the word multiple just means “more than one,” which is already conveyed by the plural noun (e.g., “nodules”). It’s often better to specify how many things you’re seeing (up to a point; precise numbers over 10 are unlikely to be helpful). Alternatively, there are other adjectives that better convey the number of items, e.g., “few,” “several,” “numerous,” “innumerable.”
Avoid “grossly,” “no obvious,” “no definite.”
These words suggest that you did not examine the study carefully; it doesn’t take a radiologist to detect “gross” or “obvious” findings. I might tolerate these terms when the study quality is so bad that the reader should know how woefully limited your interpretation is. In these cases, you should clearly specify the technical limitations of the study.
Only use eponyms and abbreviations if they are well-known and improve clarity.
Obscure eponyms can be fun to talk about, but they invite confusion in reports. Even if we discuss Amyand hernias during readout, dictate that the appendix is in the inguinal canal. However, a well-known eponym such as Crohn’s disease is preferable to the more cryptic alternative (regional enteritis). Another helpful eponym is the “Cloquet node” at the junction of the deep inguinal chain and external iliac chain. This is sometimes inadequately reported as either an inguinal node or an external iliac node, prompting confusion about the proper surgical approach to the node. An informal survey of our surgical oncologists verified familiarity with this eponym (though I would still provide an image number for those who are unfamiliar).
“Anasarca” implies severe and generalized subcutaneous edema.
I prefer not to use this term at all. I see it overused to describe any degree of subcutaneous edema, even very mild edema.
Do not bury recommendations in the findings section.
It is common knowledge that many clinicians do not have the time or interest to read the findings section of the report, and will only read the impression. Any prescriptive comments (e.g., about follow-up) should therefore be in the impression.
When describing the volume of non-loculated collections, think in terms of possible interventions.
If a pleural effusion or ascites is “large” in volume, it typically should be large enough that percutaneous therapeutic drainage is straightforward and justifiable, often not requiring imaging guidance. “Small” effusions may be appropriate for diagnostic drainage, though feasibility will vary. Too often I have seen reports describing a “large” volume of ascites where the actual volume is low enough to limit the efficacy of therapeutic drainage, prompting awkward discussions with clinicians about what we consider “large.”
Rank impression statements by importance, starting with the most important.
Not only is the impression possibly the only part of the report read by the clinician, but clinicians are less likely to attach importance to items listed toward the end of a long impression section. This is especially true if preceding items in the impression describe stable findings or matters of lesser clinical significance.
Do not assume familiarity with radiological classification systems and guidelines.
Whether to mention a radiological classification or guideline requires some judgment and familiarity with your referring clinicians. Guidelines that are well-known among radiologists (e.g., Fleischner Society guidelines for pulmonary nodules) may not be familiar to your readers. If you do not provide information beyond the name of an unfamiliar guideline or classification, prepare for questions from clinicians.
Be very careful using the word “progression” in oncology report impressions.
Oncology patients are often on clinical trial protocols where “progression” is strictly defined, for instance by RECIST, based on differences from baseline measurements of a certain date. If you do not know these criteria or the study baseline measurements for a particular patient, then describe the disease in objective terms, e.g., whether masses are increased in size. Otherwise, the word “progression” in the report impression may conflict with determinations on RECIST tracking forms, requiring phone calls and addenda for reconciliation.
Check attending edits of your reports.
Attendings sometimes go to great lengths to revise trainee reports. These edits may be highly instructive, but attendings may not have the time to review them in detail with you. Reviewing these edits is part of your training. At LAC+USC, the dictation system will highlight these edits. At the private sites (Keck, Norris, and HC2), you can use the ReportDiff system to review these edits efficiently.