‘Epidemic’ of free-text entry holding back radiology’s efforts to reduce inappropriate imaging
An “epidemic” of free-text entry may be hindering radiology’s ability to reduce unnecessary imaging, according to an analysis published in this month’s issue of JACR.
Recent legislation has required physicians to consult practice guidelines when ordering advanced imaging exams, often using an electronic aid. Carefully structured orders allow such clinical decision support systems to make their determinations. But they run into a roadblock when confronting “free text” directives that avoid using a template, noted experts with the University of Pennsylvania.
To better understand this issue, they analyzed all outpatient advanced imaging orders at their large, multi-site health system over a two-year period. Researchers found that 42% of procedure orders used free text. However, a staggering 97% of those could have been mapped to an existing structured indication. About 59% of orders went unscored by clinical decision support.
“In other words, in 6 out of 10 advanced imaging orders, the CDS system could not map the examination indication to [appropriate use criteria] and thus no determination was made about the appropriateness of these image orders,” first author Jessica Fried, MD, a former Penn resident who is now with Michigan Medicine’s Department of Radiology, and co-authors wrote March 1. “This represents a missed educational opportunity to provide best-practice feedback to ordering clinicians.”
For the analysis, Fried et al. studied outpatient advanced imaging orders placed across Penn’s enterprise following the implementation of a commercial clinical decision support system in 2017. Providers were prompted to select either a structured indication or free-text. And researchers manually reviewed the two imaging exams with the highest rate of freestyle entry: enhanced computed tomography of the abdomen/pelvis and unenhanced CT of the head.
Among nearly 40,000 advanced imaging procedures that met the study’s criteria over two years, 59% went unscored for appropriateness by the system. Free-text indications were found in 71% of such unscored scans and 42% of the entire body of exams. The manual review of the two most common free-text culprits revealed that 3,132 or 97% could be mapped to a single existing structure indication.
Addressing this issue will require a multi-pronged response. But experts believe this work is crucial as CMS increases pressure on imaging appropriateness through the Protecting Access to Medicare Act.
“Reduction of free-text indication entry will likely require interventions that both address the psychology underlying provider free-text entry, such as ingrained habit, and facilitate provider selection of an existing structured indication, potentially through artificial intelligence solutions,” Fried and co-authors concluded. “These findings can help inform the successful implementation and optimal function of imaging CDS in practices across the United States.”
You can read the entire study — “Lessons from the Free-Text Epidemic: Opportunities to Optimize Deployment of Imaging Clinical Decision Support” — in March’s Journal of the American College of Radiology here.