Experience with DBT increases radiologists’ recall, cancer detection rates for digital mammography alone

As digital breast tomosynthesis (DBT) continues to gain popularity throughout the United States, researchers are learning more and more about its impact on radiologists and their facilities. According to a new study published in Radiology, for example, experience with DBT can increase a radiologist’s recall rate, cancer detection rate (CDR) and positive predictive values (PPV) for digital mammography.

The authors performed a retrospective study of the performance of six radiologists before and after they each completed a minimum of eight hours of DBT training in 2012. Data taken from before the training (2009-2011) and after (2012-2014) were compared.

Overall, the radiologists interpreted more than 108,000 digital mammograms from 2009 to 2014, more than 50,000 before the DBT training and more than 58,000 after. The average recall rate increased from 6.8 percent before the DBT training and 7.9 percent after. The CDR, meanwhile, increase from 2.5 per 1,000 exams before the DBT training to 3.5 per 1,000 exams after. Positive screening result of biopsy recommendation (PPV2) and positive screening result of biopsies performed (PPV3) also “increased significantly.”

So why did these specialists’ recall rate for digital mammography alone change after they learned more about DBT? The authors wrote it might be they simply can’t ignore the fact that they’ve seen firsthand the power of DBT in action.

“We suspect that the increased recall rate for digital mammography alone after experience with DBT could be a result of decreased radiologist confidence when interpreting digital mammograms alone,” wrote lead author Ana P. Lourenco, MD, department of diagnostic imaging at Alpert Medical School of Brown University in Providence, R.I., and colleagues. “The experience of seeing cancers at DBT when the concomitant digital mammographic findings are negative (eg, architectural distortions seen at DBT only) or subtle may have caused radiologists to change their threshold for recalling patients when interpreting digital mammograms only.”

Lourenco et al. added that their study did have several limitations. For instance, it was more challenging to control certain factors that can impact recall rate due to the study’s retrospective design. Also, the study covered six specialists at the same institution and they all had the same subspecialty. “It may be difficult to generalize these results to a general radiology practice,” the authors wrote.