This story summarizes the results of a study of 409,000 mammograms done with computer-aided detection (CAD) & without CAD among women aged 67-89. The headline and most of the content of the story point out that CAD may not provide overall value. Use of CAD did not improve the rate of detection of invasive breast cancer. CAD use did lead to more false positives as well as an increase in further diagnostic imaging & biopsies. The potential for over-diagnosis & over-treatment also increased because CAD detected more ductal carcinoma in situ (DCIS).
We think the story should have added a line that this study does not answer questions about use of CAD in younger women.
CAD is used with 75% of mammography readings. In addition to increasing the cost of a mammogram, its use leads to increased costs due to false positives and over-diagnosis of DCIS. This over-diagnosis causes many older women to have unnecessary tests and treatment.
The story points out that Medicare pays an additional $11 for mammography readings that use CAD. It would also be useful for readers to know that women can’t choose whether to have their mammograms read using CAD or not using CAD.
The story pointed out that CAD use did increase the likelihood that a woman’s cancer would be diagnosed at an earlier stage. Readers should also have been told that the study found the increased association between using CAD & an earlier stage of cancer diagnosis may be stronger among women older than 75.
The endorsement of the use of CAD & minimization of its problems by Dr. Daniel Kopans, should have been balanced by comments from another expert not connected with the study.
Beginning with the headline “Are Pricey Computer-Aided Mammograms Worth It?” and continuing throughout the piece, the potential problems with CAD are noted. The article includes several quotes from Dr. Joshua Fenton, lead author of the study, on three key problems with CAD. Dr. Fenton observed that “the (overall) rate of invasive breast cancer diagnosis was no different with or without CAD.” In addition, “CAD was also associated with increased diagnostic testing among women who turned out not to have breast cancer.” These women had additional mammography, or ultrasound, or biopsies. Finally, because CAD led to increased diagnosis of DCIS, Dr. Fenton observed that “If you are an older woman, (there is) the risk of over-treatment of noninvasive lesions.”
The story should have included limitations of the study that were mentioned in the study itself. The study authors noted that there was “a short follow-up for cancer stage, potential unmeasured confounding, and uncertain generalizability to younger women.” There is a big gap about what the implications are for younger women where there is considerable controversy about use of mammograms in the first place.
There is no disease mongering.
The story includes comments by Dr. Daniel Kopans, who was not connected with the study. Because Dr. Kopans’ comments are extensive & mainly support the use of CAD, observations by a second independent source with a different opinion would have provided needed balance for the reader. Comments by lead study author Dr. Fenton, point out some problems with CAD but analysis by another independent source would be helpful for readers.
Kopans clearly offers a counterpoint to Fenton, but some of Kopans’ statements, such as the following, should not have gone unchallenged and it would have been easy to find other experts who believe this is not the case:
“While he calls finding more DCIS with computer-aided detection also ”likely a good thing,” he acknowledged the debate about the possibility that a very early cancer in an elderly woman may not cause a problem in her lifetime. Although the treatment of DCIS is debated among experts, Kopans said, “In my mind, it is always worth it to find additional cancers.”
By explaining the limitations & benefits of adding CAD to standard mammography readings, the article informs readers about both methods.
Early in the article, the study author is quoted observing that “It’s an add-on and now is used in probably three-quarters of U.S. mammograms.”
Women would have to ask if CAD is being used to read their mammograms.
The piece points out that use of CAD increased from 3.6% in 2001 to 60.5% in 2006 & is at approximately 75% in 2013. The study examined mammography data obtained from 2001-2006.
By including quotes from the lead author & from an independent source, the article clearly has not relied solely on a press release.