New research suggests that the correct interpretation of mammography results varies between facilities. Moreover, there are characteristics that predict which facilities are likely to provide more accurate readings.

New research suggests that the correct interpretation of mammography results varies between facilities. Moreover, there are characteristics that predict which facilities are likely to provide more accurate readings.

The findings from a number of studies have indicated that the accuracy varies greatly among radiologists. "What's unique about this study is that we looked at variation in mammography interpretations across facilities where mammography is performed," lead author Dr. Stephen Taplin, from the National Institutes of Health in Bethesda, Maryland, told Reuters Health.

In the study, reported in the Journal of the National Cancer Institute, significant variation in specificity, the number of women testing negative for cancer; positive predictive value, the number with positive results who were actually positive; and the likelihood of cancer among women referred for biopsy; was noted between facilities.

"The most surprising finding," Taplin said, "was that there were characteristics that did explain some of the differences" in the facilities. "This is good news because it means that facilities could begin to consider adopting characteristics associated with better performance."

Factors associated with increased interpretive accuracy included offering only screening mammograms, having a breast imaging specialist (at least 50 per cent of time spent doing breast imaging) read the mammograms; not performing double readings; and conducting audit reviews at least twice a year.

The study was conducted between 1996 and 2002 and surveyed 53 mammography facilities, 44 of which provided data suitable for analysis. A total of 484,463 screening mammograms were performed on 237,669 women, including 2686 diagnosed with breast cancer during follow-up.

Across the facilities, the average specificity, positive predictive value, and likelihood of cancer among biopsy-referred women were 90.2, 4.1 and 38.8 per cent, respectively.

"The take-home message for practitioners would be to ask whether a radiologist is devoting a substantial amount of time to reading mammograms," Taplin said. "The presence of a breast imaging specialist was associated with better performance."

TrackBack URI for this entry

Comments (0)

Subscribe to this comment's feed