Mammograms consist of shadowy outlines of fat and soft tissue in varying shades of gray. Interpreting them requires skill and experience, and, as for every type of imaging test, different radiologists may interpret the same mammograms differently. Many factors influence the accuracy of individual radiologists in recognizing clinically important abnormalities during screening mammography, but many other factors influence the consistency of mammographic interpretation (reviewed by Beam, Elmore, Sickles, and colleagues 7,15,35,41,105).
The accuracy of radiologists in interpreting mammograms depends on many factors, including case variation, practice variation, training and experience, and the type of screening program in which they practice. Box 3-4 summarizes those factors that have been reported in the peer-reviewed literature to be correlated with the interpretation of screening mammograms. It is important to keep in mind that estimates of performance are different for screening and diagnostic mammography. In screening, the central decision is whether or not to conduct additional workup (i.e., the callback decision). The goal of screening mammography is not to provide a definitive diagnosis or to recommend biopsy without further consideration. A true positive in screening occurs whenever a woman with breast cancer is given a recommendation for additional workups, whereas a true positive in diagnosis would be whenever breast cancer is detected.
Individual characteristics such as breast density or history of breast cancer are known to increase the likelihood of both false-positive and false-negative results (Table 3-3).8,24,27 In addition, ambiguous mammograms such as those revealing possible microcalcifications—which are often difficult to interpret—increase the likelihood of disagreement among radiologists. A mammography practice that serves younger women is likely to have an overall lower sensitivity rating than an otherwise identical practice that serves older women. This is reflected in the observation by Beam and his colleagues that case-related differences accounted for more variation than individual differences among radiologists.10 For example, mammographic sensitivity increases with a woman’s age (Figure 3-4).
Also, approximately twice as many breast cancers are detected at first screens as compared to subsequent screens. This is because a cancer detected at a subsequent screen generally would have developed to the point where it can be seen on a mammogram only since the previous screen, whereas a cancer detected at a woman’s first screen could have been present for years. The result is not only a higher rate of cancers detected at first