Discrepancy Rates and Errors a Bigger Problem with Body MRI Than Realized

There is more disagreement between radiologists and more errors identified on secondary interpretations of body MRI scans than previous studies have reported, leading some experts to call for more sub-specialty interpretations.

Nearly 70 percent of body MRI interpretations have at least one discrepancy, according to researchers at the University of Vermont and the University of Southern California Medical Center. And, since most of these errors are cognitive – a misidentification of a finding – lead study author Danielle E. Kostrubiak, M.D., a diagnostic radiology resident at the University of Vermont Medical Center, said there should be a push for having sub-specialty trained providers read these studies.

“The data suggest that sub-specialty interpretations should be encouraged at tertiary care centers,” she said, “and institutions should provide adequate resources for these interpretations to occur.”

The team published their results in the American Journal of Roentgenology. It is the first study to focus on secondary interpretations of body MRI evaluated by type of likely error, they said.

Related Content: Diagnostic Errors: Lessons Learned and Mitigation Strategies​

Interpretation errors are common in radiology, particularly with MRI scans, and those mistakes frequently lead to delayed or incorrect treatment plans. According to existing research, discrepancy rates can range from between 2 percent and 6 percent for resident reports over-read by attending providers to up to 56 percent for secondary interpretations. Most of these problems occur with pelvic and abdominal imaging, they said.

To determine the actual discrepancy rate, as well as why these mistakes happen, Kostrubiak’s team retrospectively reviewed 357 secondary body MRI reports that were captured between January 2015 and December 2018. They also analyzed the initial reports and divided the discrepancies by the most likely underlying reason.

56-year-old woman with benign hemangioma. Lesion was originally reported as indeterminate enhancing mass, and outside report recommended biopsy. Classic features of benign hemangioma are shown. Error was attributed to faulty reasoning. A, Axial MR image obtained 5 minutes after contrast agent administration shows peripheral nodular discontinuous enhancement. B, Axial MR image obtained 10 minutes after contrast agent administration shows centripetal progression of enhancement (arrow). C, Axial fast imaging employing steady-state acquisition (FIESTA) MR image shows lesion is homogeneously hyperintense compared with liver parenchyma. Courtesy: American Journal of Roentgenology

Based on their evaluation, they identified at least one discrepancy in 246 cases (68.9 percent), and they pinpointed a second discrepancy in 54 of those cases. Cognitive errors accounted for most discrepancies – 68.8 percent of errors – perceptual errors accounted for 59.3 percent of secondary discrepancies.

“This difference [between cognitive and perceptual errors] may be explained by the fact that abdominal and pelvic MRI is frequently not the first imaging modality used and is often performed to characterize an abnormality detected on CT or ultrasound,” they said. “Therefore, the abnormality has already been perceived by the time MRI is performed.”

When they drilled down in the underlying reasons behind these discrepancies, they found that faulty reasoning – a misclassification of the abnormality – was responsible for 34.3 percent of all instances, as well as 37.8 percent of primary discrepancies. In addition, satisfaction of search occurred with 37 percent of second discrepancies, and 15 percent of overall discrepancies. In these instances, the team said, an MRI scan has likely been ordered to answer a specific question, and once that answer has been determined, the radiologist likely did not examine the rest of the scan for any other abnormalities.

There could be several reasons why these discrepancy rates are higher than what has been previously reported, the team said. Existing research already shows that body imaging frequently has highest error rates, and double-reading by sub-specialists also increases the discrepancy rate, they added. In addition, general radiologists may be unaware of MRI’s high sensitivity can be used to determine specific diagnoses.

It is possible that similar studies will be limited by the innate subjectivity of error classification, but the team said related research should be easier to conduct in the future as medical practices expand their use of electronic medical records.

“The next step,” they said, “would be to explore how these discrepancies may impact patient outcomes and overall cost to the system associated with these radiologic errors.”

For more coverage based on industry expert insights and research, subscribe to the Diagnostic Imaging e-Newsletter here.