E4672. Evaluating Long-Term Effects of Changes in Density Assessment Criteria on Density Reporting Patterns of Radiologists in Clinical Practice
  1. Abid Irshad; Medical University of South Carolina
  2. Brielle Paolini; Medical University of South Carolina
  3. Ian Smith; Medical University of South Carolina
  4. Rebecca Leddy; Medical University of South Carolina
  5. Madelene Lewis; Medical University of South Carolina
  6. Abbie Cluver; Medical University of South Carolina
  7. Leonel Vasquez; Medical University of South Carolina
This study aims to assess the density reporting patterns of radiologists several years after the changes were made in density assessment criteria in BI-RADS 5th edition by removing the percent-density value system as a criterion for density assessment.

Materials and Methods:
After IRB exemption, a 4-year single institution mammography dataset of density reporting patterns was obtained from January 1, 2010–December 31, 2013, before the BI-RADS changes in density assessment criteria, which was published in early 2014 (BI-RADS 5th edition). To evaluate long-term effects of this change, another 4-year dataset was obtained from January 1, 2019-December 31, 2022, 5 years after the new guidelines, hoping for the changes to be fully incorporated in routine clinical practice. The datasets were obtained through the Penrad reporting system for a group of 6–8 radiologists, and data were compared for any significant changes in density reporting patterns after the new guidelines. The four density categories were evaluated – almost entirely fatty (AEF), scattered areas of fibroglandular density (SFGD), heterogeneously dense (HD), and extremely dense (ED). Because of its practical impact, special attention was paid to the binary classification of nondense (sum of first two categories) versus dense breasts (sum of last two categories) because these categories are reported to patients for potential supplemental screening opportunities in patients with dense breast. Comparison of nondense versus dense breast category was specifically made to evaluate any changes that could have led to a change in the needs for supplemental screening after new guidelines. Chi-square statistics was used for comparative analysis, and the p-value was considered significant at p < 0.05.

A total of 40,942 mammograms were included in the 4-year dataset before the new density assessment guidelines. Distribution of AEF, SFGD, HD, and ED categories was 6070 (14.8%), 19,167 (46.8%), 13,228 (32.3%), and 2475 (6.0%) mammograms, respectively. The 4-year data after the new guidelines included 91,554 mammograms, with AEF, SFD, HD, and ED categories having 6840 (7.5%), 48,829 (53.3%), 31,103 (34%), and 4612 (5.0%) mammograms, respectively. Prechange data showed 25,237 (61.6%) mammograms categorized as nondense and 15,703 (38.4%) as dense. Postchange data showed 55,671 (60.8%) mammograms categorized as nondense, and 35,715 (39.2%) categorized as dense. Comparing the reporting patterns of radiologists before and after the guideline changes showed no difference in the reporting pattern for the dense breasts (x²= 2.261, p = 0.132), or nondense breasts (x²= 1.998, p = 0.157). However, a significant decrease was incidentally noted in categorizing predominantly fatty breasts after the new guidelines (p < 0.001).

Radiologists' reporting patterns for categorizing dense or nondense breasts did not change several years after removing the percent-value density assessment criterion from the guidelines, which in turn may not have significantly impacted the overall need for supplemental screening in our population.