Availability of priors, tomosynthesis independently influence recall rates

The availability of prior examinations and digital breast tomosynthesis (DBT) are independent primary factors in reducing recall recommendations during mammographic interpretations, according to a study published online Dec. 6 in Academic Radiology.

With the recent advent of DBT, many have high hopes for the modality as it’s expected to reduce recall rates while simultaneously increasing cancer detection rates. However, many studies that have substantiated these beliefs with data have failed to include the availability of prior examinations during interpretation.

“It is well known that the availability of prior examinations during the interpretation of screening mammograms reduces recall rates substantially by enhancing the ability of the observer to assess change over time, if any, while at the same time discard depicted abnormalities that clearly represent a variety of benign (and stable) findings,” said the study’s lead author, Christiane M. Hakim, MD, of the Magee-Womens Hospital of the University of Pittsburgh Medical Center, and colleagues.

Prior exams and DBT are the two primary contributing components to the reduction in recall rates in women who had undergone mammography examinations previously, and the authors designed a two-reading mode observer study to offer a preliminary assessment of the interaction between these two components.

During the study, eight radiologists independently interpreted 36 mammography exams twice, each of which had current and prior full-field digital mammography images (FFDM) and DBT. The data set included two verified cancer cases, 22 benign cases recalled in clinical practice, and 12 negative/benign cases not recalled.

After analysis of recall rates and possible interactions between prior exams and DBT, results indicated that the average recall rates in noncancer cases were significantly reduced with the addition of DBT and priors. The addition of DBT to FFDM reduced recall rates from 0.67 to 0.54 and from 0.54 to 0.27 when DBT was available before and after priors. Conversely, recall reductions went from 0.64 to 0.54 and 0.42 to 0.33 when priors were available before and after DBT.

“Our study is a pilot in preparation for a pivotal study that will include a larger sample size and a more representative set of cases,” the authors concluded. “This pilot study provides us with estimates of the effects of availability of DBT and prior information as well as between-reader variability of recall rates and the possible expected interaction between the two types of additional information.”

A recent study published in the December issue of Radiology revealed different findings in regards to DBT and recall rates. Click here to read more.