Misclassification of Breast Imaging Reporting and Data System (BI-RADS) Mammographic Density and Implications for Breast Density Reporting Legislation

USA states have begun legislating mammographic breast density reporting to women, requiring that women undergoing screening mammography who have dense breast tissue (Breast Imaging Reporting and Data System [BI-RADS] density c or d) receive written notification of their breast density; however, the impact that misclassification of breast density will have on this reporting remains unclear. The aim of this study was to assess reproducibility of the four-category BI-RADS density measure and examine its relationship with a continuous measure of percent density. We enrolled 19 radiologists, experienced in breast imaging, from a single integrated health care system. Radiologists interpreted 341 screening mammograms at two points in time 6 months apart. We assessed intra- and interobserver agreement in radiologists'; interpretations of BI-RADS density and explored whether agreement depended upon radiologist characteristics. We examined the relationship between BI-RADS density and percent density in a subset of 282 examinations. Intraradiologist agreement was moderate to substantial, with kappa varying across radiologists from 0.50 to 0.81 (mean = 0.69, 95% CI [0.63, 0.73]). Intraradiologist agreement was higher for radiologists with ≥10 years experience interpreting mammograms (difference in mean kappa = 0.10, 95% CI [0.01, 0.24]). Interradiologist agreement varied widely across radiologist pairs from slight to substantial, with kappa ranging from 0.02 to 0.72 (mean = 0.46, 95% CI [0.36, 0.55]). Of 145 examinations interpreted as "nondense" (BI-RADS density a or b) by the majority of radiologists, 82.8% were interpreted as "dense" (BI-RADS density c or d) by at least one radiologist. Of 187 examinations interpreted as "dense" by the majority of radiologists, 47.1% were interpreted as "nondense" by at least one radiologist. While the examinations of almost half of the women in our study were interpreted clinically as having BI-RADS density c or d, only about 10% of examinations had percent density >50%. Our results suggest that breast density reporting based on a single BI-RADS density interpretation may be misleading due to high interradiologist variability and a lack of correspondence between BI-RADS density and percent density.