A research group led by Dr. Stephen Taplin of the National Cancer Institute has published a paper in the June 18th issue of the Journal of the National Cancer Institute that assesses whether certain factors differing between mammography facilities are associated with better interpretive accuracy on screening mammograms.
The bottom line is that there are several things that were signficant. Sensitivity, the ability to detect breast cancers, was quite high and varied little. However, specificity and positive predictive value varied significantly between sites. Thus, it seems that the degree to which sites may be "overcalling" their reads (presumably to avoid missing something) varies.
The following mammography facility traits were associated with better interpretive accuracy for screening mammography:
- Performing only screening mammography at the facility (as opposed to screening plus diagnostic mammography)
- Having a breast imaging specialist interpreting the mammograms
- Not performing double reading of mammograms (more false positives when read by more than one radiologist)
- Conducted audit reviews 2 or more times per year (where individual radiologist-level performance data is shared with the radiologist)
The authors note that it was surprising that double reading did not improve overall test accuracy and that this contrasts with the results of randomized clinical trials. It may be that double reading methods differ or have been implemented differently in typical clinical practice environments outside of clinical trials. This deserves a closer look in the future. Given the relatively weak effect of this predictor of performance, the authors suggested that it should not be utilized to differentiate amongst facility performance until further detailed research is done to sort out the controversy.
It is important to keep in mind that the above factors affected specificity and positive-predictive value primarily. The authors point out that whether sensitivity or specificity is most important in picking a mammogram facility is dependent on a value judgment. That is, some women may be most concerned about sensitivity at all costs (i.e., a higher false positive rate leading to a biopsy does not matter). Other women may wish to consider factors like those noted above that affect specificity and positive predictive value (i.e., they would like to avoid unnecessary biopsies while keeping sensitivity high).
The results of this retrospective study should be confirmed elsewhere before they are utilized for decision-making. Nevertheless, you may want to have a dialogue with your physician about mammography facility choice.
Comments