‘Significant variability’ found in radiologists’ BI-RADS category 3 usage rates
Radiologists are inconsistent with their usage of the American College of Radiology’s Breast Imaging Reporting and Data System (BI-RADS) category 3, according to new research published in Academic Radiology. The authors also noted that increased usage of BI-RADS 3 did not correlate with the specialist’s experience level or an improved cancer detection rate (CDR), though other correlations were present.
“Although flexibility in the BI-RADS category 3 is necessary for practical purposes, there is justifiable concern regarding appropriate utilization of this assessment category,” wrote co-author Emily B. Ambinder, MD, of the Russell H. Morgan Department of Radiology and Radiological Science at Johns Hopkins Medical Institutions in Baltimore, Maryland, and colleagues.
Ambinder et al. reviewed all diagnostic mammograms performed at their institution from July 1, 2013, to August 8, 2017. Overall, this included interpretations by 13 breast imagers and more than 24,000 examinations. The team found “significant variability” in BI-RADS 3 rates between the facility’s radiologists, ranging from 8 percent to more than 19 percent. In addition, BI-RADS 3 rates were found to negatively correlate with BI-RADS 1 or BI-RADS 2 rates. A positive correlation was noted between BI-RADS 3 rates and recall rates. There was no association, however, between BI-RADS 3 rates and experience levels or CDR.
“We suggest that for radiologists with high BI-RADS 3 rates, a shift of some diagnostic cases from BI-RADS 3 to BI-RADS 1 or 2 assessments may be appropriate,” the authors wrote. “We also hope that our findings begin a discussion of a national benchmark for BI-RADS category 3 rate in order to further standardize usage.”
Information on recent research on BI-RADS category 5 assessments, also published by Academic Radiology, can be read here.
Michael has more than 18 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.