A large observational study examining the variation in breast density assessment among radiologists in clinical practice has found a wide variation—from 6.3% to 84.5%—in the percentage of mammograms rated as showing dense breasts, which persisted after adjusting for patient characteristics. The finding has important implications for debates about mandatory reporting of breast density information, clinical management of patients who are told they have dense breasts, and investigators using radiologists’ subjective measures of breast density in cancer research. The study by Brian Sprague, PhD, of the University of Vermont, and colleagues was published in Annals of Internal Medicine.
About half of the states in the United States have legislation mandating radiology facilities to disclose mammographic breast density information to women and often includes language-recommending discussion of supplemental screening options for women with dense breasts.
The researchers conducted an observational study using prospectively collected data from 30 radiology facilities within the 3 breast cancer screening research centers of the population-based Research Optimizing Screening Through Personalized Regimens (PROSPR) consortium. The screening mammograms were conducted between 2011 and 2013 among women between the ages of 40 and 89. The final study population for analysis consisted of 216,783 screening mammograms from 145,123 women. The mammograms were interpreted by 83 radiologists.
Patient characteristics, including age, race, body mass index (BMI), and history of breast cancer at the time of the examination were obtained via a radiology clinic patient questionnaire. Other details of the mammography examination were also obtained from the radiology facilities, including date of the examination, identification number of the interpreting radiologist, and descriptor of mammographic breast density. The information on breast density was clinically recorded using the Breast Imaging Reporting and Data System (BI-RADS), which provides four possible categories for breast density: almost entirely fat, scattered fibroglandular densities, heterogeneously dense, or extremely dense. Heterogeneously dense and extremely dense categories were considered “dense” for analyses.
Overall, 36.9% of mammograms were rated as showing dense breasts. Across radiologists, this percentage ranged from 6.3% to 84.5% (median, 38.7% [interquartile range, 28.9%–50.9%]), with multivariable adjustment for patient characteristics having little effect (interquartile range, 29.9%–50.8%). Examination of patient subgroups revealed that variation in density assessment among radiologists was pervasive in all but the most extreme patient age and BMI combinations. Among women with consecutive mammograms interpreted by different radiologists, 17.2% (5,909 of 34,271) had discordant assessments of dense vs nondense status.
A limitation of the study is that quantitative measures of mammographic breast density were not available for comparison.
The widespread enactment of breast density notification laws presents physicians with the difficult task of discussing the potential benefits and harms of supplemental breast cancer screening in the absence of consensus guidelines. “Overall, our findings suggest that a woman’s likelihood of being told she has dense breasts varies substantially on the basis of which radiologist interprets her mammogram,” wrote the study authors.
“As breast density is increasingly used in screening decision-making, the development of further professional standards, potentially including increased training or use of automated density quantification tools, may lead to more effective clinical care,” concluded the researchers.
Funding for this study was provided by the National Cancer Institute. ■