The accuracy of mammographic interpretation can vary widely, but the source of the variability has not been explained. To investigate the relationship between radiologists' characteristics and actual performance, William E. Barlow, Ph.D., of Cancer Research and Biostatistics in Seattle, and colleagues surveyed 124 radiologists and tracked cancer outcomes from the more than 460,000 screening mammograms they interpreted between 1996 and 2001.

Their findings were published in the Journal of the National Cancer Institute.

Greater volume of mammograms interpreted and more years of experience were not associated with greater accuracy. However, greater volume was associated with higher sensitivity (more true positive results in women who had breast cancer) and lower specificity (more false positive results in women who did not have breast cancer) whereas more experience was associated with lower sensitivity and higher specificity. The authors conclude that increasing volume requirements for radiologists is unlikely to improve the interpretation of mammograms.

"Although radiologists differ in performance, accuracy does not appear to be simply attributable to years of experience or number of mammograms interpreted," the authors write.

"Direct feedback of performance characteristics couple with training may be more helpful than experience without feedback. The most instructive exercise may be to have an open discussion of misjudged mammograms, but concern about malpractice claims may prevent this opportunity from occurring."

SOURCE:
Journal of the National Cancer Institute, December 15, 2004