Supplemental training packs potential to improve performance in reading screening mammograms

Radiologists can encounter challenges when reading screening mammograms, but recent research reveals that supplemental training has the potential to improve their performance on this front.

In a study published online June 25 in the Journal of the American College of Radiology, Tanya Geertse, of the Dutch Expert Centre for Screening in the Netherlands and colleagues observed that “the prevalence of disease is low in breast cancer screening.” Because radiologists can detect only a small number of cancers in a large number of images,” they wrote, “reading mammograms in a screening setting ... requires different skills compared with a clinical setting.”

The researchers sought to examine not only whether such training has a positive impact on radiologists’ ability to interpret breast cancer screening studies but also how improvements in this ability are measured. Using PubMed on August 3, 2017, they initially identified 2,199 studies centering on breast cancer screening outcomes. From these, they amassed a final group of 18 studies for review, selecting only those that focused on improving radiologists’ performance in reading breast cancer screening mammograms through supplemental training and contained a report of at least one outcome measure.

Studies whose population included non-radiologists were excluded from the final group, and the quality of studies under consideration was assessed using the Medical Education Research Quality Instrument.

Of the 18 studies, 17 showed improvement in at least one outcome measure for at least one training activity. Two measurement approaches—radiologists’ performance on test sets and radiologists’ performance during actual breast cancer screening—were spotlighted in the studies. Sensitivity and specificity were the most commonly reported outcomes for performance on test sets, with sensitivity improving in all studies in which it was an outcome and specificity improving in 75% of studies in which it was an outcome.

Meanwhile, recall rate was the most reported outcome for performance during actual screening. Recall rate improved in 83% of studies in which it was an outcome.

“On the basis of this review, we conclude that supplemental training appears to improve radiologists’ performance,” the researchers wrote. While the “considerable heterogeneity in interventions used and effect measures reported” made formulating accurate conclusions about the best types of supplemental training an impossibility, they noted, the review increased their understanding of the importance of supplemental training in enhancing the performance of breast screening radiologists and underscores the importance of well-designed studies to assess such performance. 

Geertse et al also cited a need for more high-quality research that explores performance in screening practice and seeks to identify methods of isolating the effect of training on screening performance. Many of the studies identified, the authors stated, were merely of “moderate” quality as a result of small sample sizes and the absence of control groups.