Structured reporting increases ‘normal’—but ‘unremarkable’ persists

Implementing structured reporting templates has helped an academic radiology department increase its collective use of the clear and definitive term normal. However, use of the fuzzier term unremarkable also went up, suggesting more training may be in order.

Stephany Ross, MD, Ross Filice, MD, and colleagues at MedStar Georgetown University Hospital, which operates in and around the nation’s capital, report on their project in a study published online June 17 in the Journal of the American College of Radiology.

Upon searching their database for all CT scans of the abdomen and pelvis before and after the template implementation in April 2015, the team came back with 878 pre-template reports and 875 templated reports with which to compare them.

A total of 13 attending radiologists and 27 trainees, ranging from first-year residents to abdominal imaging fellows, had filed reports meeting the study’s criteria for report volumes in both periods.

The team analyzed the before-and-after reports primarily to see if the template implementation had helped normal progress over unremarkable—a departmental aim partly propelled by JACR’s ongoing “Speaking of Language” article series. This has called for, among other quality improvements, language in radiology reports that’s sufficiently unambiguous to serve patients as well as referring physicians.

They also compared the implementation’s effect on word counts and on numbers of changes made as reports moved from preliminary to final status.

The researchers found no significant changes on these latter two measures between the pre- and post-template periods.

As for the specific language they were looking for, the per-report frequency of the word normal increased from an average of 5.29 before implementation to 8.92 after, while unremarkable increased from 0.11 to 0.22.

“The increase in normal in post-template reports reflects a trend toward more definitive reporting, which may increase clarity and satisfaction among referring physicians,” the authors conclude. “The increased use of the less definitive term unremarkable identifies an area in need of further training and improvement in our department.”

The authors acknowledge their single-site design as a limitation, as was their focus on normal and unremarkable. 

“It is possible that these terms are listed more frequently in templated reports simply because the tabulated style of templated reports leads to a decrease in the use of compound sentences,” they write, explaining that a radiologist might have used normal to characterize multiple findings with one usage while the template would use it, separately and repeatedly, for each finding.

The authors also note that the word unremarkable may not necessarily represent less definitive language for all radiologists, referring clinicians and patients. “There is sparse literature on the use of unremarkable in reporting,” they point out, “and published pieces are expert opinion, rather than evidence-based research findings.”

Ross et al. suggest the relationship between report language and referrer satisfaction is an area in need of survey-based research.

“Another avenue to further explore is teaching the use of clear language in radiology reporting,” they add. “Residents do not get a great deal of formal training in generating definitive reports, and this is an area that could benefit from further education.”