Zero Added Value With Computer-Aided Digital Mammography

Megan Brooks

September 28, 2015

Computer-aided detection (CAD) does not improve accuracy of digital screening mammography, a large study has found.

Even more concerning, radiologists tended to miss more cancers when using CAD, lead author Constance D. Lehman, MD, PhD, codirector of the Avon Comprehensive Breast Evaluation Center at Massachusetts General Hospital, Boston, told Medscape Medical News.

It is possible that radiologists reading with CAD are overly dependent on the computer and ignore suspicious lesions if they are not flagged by CAD, she said.

The study was published online September 28 in JAMA Internal Medicine. It joins two earlier studies that suggested that CAD had little, if any, impact on breast cancer detection. The studies raised concerns that CAD may increase recall and biopsy rates.

"The naysayers of these two studies had some valid criticisms," Dr Lehman noted. One study included an older Medicare population, and the other paired CAD with outdated film mammography. "So I thought maybe we will find that there is some benefit, so it was a bit surprising that there was none at all," Dr Lehman said.

The researchers compared the accuracy of digital screening mammography interpreted either with CAD (n = 495,818) or without CAD (n = 129,807) among 323,973 women. A total of 271 radiologists from 66 centers in the National Cancer Institute–funded Breast Cancer Surveillance Consortium interpreted the mammograms. On the basis of tumor registry data, a total of 3159 breast cancers were identified within 1 year of screening.

CAD failed to boost sensitivity or specificity of digital screening mammography, the researchers report.

Measure CAD (95% CI) No CAD (95% CI)
Sensitivity 85.3% (83.6% - 86.9%) 87.3% (84.5% - 89.7%)
Specificity 91.6% (91.0% - 92.2%) 91.4% (90.6% - 92.0%)

CI, confidence interval.


CAD also failed to improve the overall cancer detection rate (4.1 per 10,000 women screened with and without CAD) or the invasive cancer detection rate (2.9 and 3.0 cancers per 1000 women screened with and without CAD, respectively). However, the detection rate for ductal carcinoma in situ was slightly higher with CAD than without (1.2 vs 0.9 cancers per 1000 women), which has been shown previously.

"Even more troubling," Dr Lehman notes in a news release, within the group of 107 radiologists who interpreted mammograms both with and without CAD, CAD was associated with decreased sensitivity (odds ratio, 0.53; 95% CI, 0.29 - 0.97).

It was actually worse with CAD. Dr Constance Lehman

"It wasn't just that it didn't improve detection rates, but it was actually worse with CAD," Dr Lehman told Medscape. "That concerns me," she said, "because while radiologists are taught you cannot be overly dependent on the computer, there is a concern that some radiologists are saying, 'Well, I noticed that it looked a little different, but it doesn't have a CAD mark, so I'm going to trust the computer.' "

Calls to Stop Paying for CAD

By conservative estimates, CAD costs over $400 million a year in current US healthcare dollars, with no added value and, in some cases decreased, performance, the researchers note in their article.

"In the era of Choosing Wisely and clear commitments to support technology that brings added value to the patient experience, while aggressively reducing waste and containing costs, CAD is a technology that does not seem to warrant added compensation beyond coverage of the mammographic examination," they write. "The results of our comprehensive study lend no support for continued reimbursement for CAD as a method to increase mammography performance or improve patient outcomes."

Joshua J. Fenton, MD, MPH, of the University of California, Davis Health System, Sacramento, who has also studied the value of CAD in mammography, agrees.

CAD was widely adopted before strong evidence of effectiveness, he notes in a linked commentary. "If the CMS [Centers for Medicare and Medicaid Services] were to consider a proposal for new CAD coverage at this time, the current evidence base would not support approval. Thus, we should question whether society should continue to pay for CAD use. Congress should therefore rescind the Medicare benefit for CAD use," he writes.

"I think this is finally the paper that people are going to have to take a long, hard look at," Robert Smith, PhD, vice president of cancer screening at the American Cancer Society, noted in an interview with Medscape Medical News.

"The data have been accumulating over time and show very disappointing results from CAD in the average way it is used by most radiologists. What's useful about this particular paper is that it is based on all-digital mammography," Dr Smith said. "And, I think, the most startling finding is that radiologists that read at sites with CAD have poorer performance in those sites than they do in the sites without CAD."

"The authors pose, I think, a really legitimate question ― what are we paying for?" Dr Smith added.

The study had no commercial funding. Dr Lehman has received grant support from General Electric (GE) Healthcare and is a member of the Comparative Effectiveness Research Advisory Board for GE Healthcare. No other significant financial relationships have been reported.

JAMA Intern Med. Published online September 28, 2015. Abstract, Commentary


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: