Academic affiliation affects radiologists' diagnostic mammo performance

Recent research published in the American Journal of Roentgenology studied radiologist characteristics that may impact false-positive rates and sensitivity in diagnostic mammograms performed either for additional evaluation of a recent mammogram or to evaluate a breast problem.

No characteristics associated with false-positive rates could be identified, but the authors did find that academic affiliation and 10-19 years of interpretive experience can lead to greater sensitivity in at least one of the two types of diagnostic mammography that were studied.

Sara L. Jackson, MD, division of general internal medicine at the University of Washington, and colleagues studied survey data from 244 radiologists from 1998 to 2008. The data covered over 274,000 diagnostic mammograms.  

Looking over the data, Jackson and colleagues discovered that the overall accuracy of diagnostic mammograms was higher than what had been reported by previous studies. Diagnostic mammograms performed for evaluation of a recent abnormal screening mammogram had a false-positive rate of just under 12% and sensitivity of over 90%. Diagnostic mammograms performed for evaluation of a breast problem had a false-positive rate of over 7% and a sensitivity of just under 84%.

Does this mean previous studies were wrong, or that the examination quality has improved? Not necessarily, according to the authors. It may just mean that the radiologists who took this survey were especially effective.

“It is possible that overall accuracy of diagnostic mammography interpretation has improved over time or that the additional registries included in our study have higher performance,” the authors wrote.

Digging deeper into the data, however, the authors suggested one reason that the improved accuracy could be completely legitimate.

“Arguing for improvement over time, we did find that radiologists with 10–19 years of experience interpreting mammography had higher sensitivity for additional evaluations of a recent mammogram than those who had been interpreting for 20 years or longer,” the authors wrote. “Ten to 19 years represents the period since 1992 when the Mammography Quality Standards Act was passed, and it is possible that with increased standardization of breast imaging training, accuracy has improved.”

That higher sensitivity for radiologists with 10-19 years of interpretive experience is clear when looking over the data. Diagnostic mammograms they performed for additional evaluation following a recent screening mammogram had a sensitivity of 91.9%, compared to 88.1% by those with less experience and 89.6% by those with more experience.

The authors did not find 10-19 years of experience to matter as significantly with diagnostic mammograms performed for evaluation of a breast problem.

One radiologist characteristic that improves sensitivity for both types of diagnostic mammography is academic affiliation.

Diagnostic mammograms performed by an academic radiologists for evaluation of a recent mammogram had a sensitivity of 93.5%, compared to 89.4% by those with no academic affiliation. Diagnostic mammograms performed by an academic radiologists for evaluation of a breast problem had a sensitivity of 85.9%, compared to 83% by those with no academic affiliation.

The authors noted that this is likely related to the fellowship training that academic radiologists receive. In addition, academics receive more referrals, including more complex cases, so they gain that valuable experience.

And while it would be unfair to expect a significant increase in how many diagnostic mammograms get performed by academic radiologists, one change the authors suggest is improving the training for nonacademic radiologists.

“These results suggest that a curricular review of training in academic settings for core content might be standardized and made available to all radiologists,” the authors wrote.