Response: Re: Effectiveness of Computer-Aided Detection in Community Mammography Practice
Joshua J. Fenton, Patricia A. Carney, Joann G. Elmore, W. E. Barlow · JNCI Journal of the National Cancer Institute · 2011
Nishikawa et al. and Levman overlook the primary findings of our study of computer-aided detection (CAD) and highlight a secondary finding: The increase in ductal carcinoma in situ (DCIS) detection in unadjusted analyses. However, in fully adjusted analyses, the association between CAD and increased detection of DCIS was not statistically significant. Nor was CAD associated with significantly increased (or decreased) detection of invasive breast cancer or with the stage or size of invasive breast cancers. Nishikawa et al. write that “increased detection of DCIS with CAD should not be considered overdiagnosis, but rather earlier detection: an important goal of screening mammography.” Women, clinicians, and policymakers must grapple with a more complex reality that the long-term prognosis of DCIS detected by mammography is not fully understood and that some DCIS detection likely constitutes overdiagnosis ( 1 ). Any screening endeavor has the potential for benefit as well as harm. Simply asserting that early detection of DCIS is beneficial is holding the bar too low for CAD, particularly in light of its negative impact on specificity.