Computer-aided detection doesn’t improve mammography: study

Defying intuition as well as expectations, a new analysis of computer-aided detection (CAD) in nearly 626,000 screening mammography exams has found the technique no better than old fashioned “CAD-less” interpretations.

In fact, in the 271-radiologist study, sensitivity in image interpretations without CAD was actually somewhat better—87.3 percent to CAD’s 85.3 percent—while specificity was practically identical (91.6 percent with CAD and 91.4 percent without).

Study authors found no difference in the overall cancer detection rate (4.1 cancers per 1,000 women screened with and without CAD) or in the invasive cancer detection rate (2.9 vs. 3.0 cancers per 1,000 women screened with CAD or without).

Their work is running online in JAMA Internal Medicine.

Analyzing results gleaned from the National Cancer Institute’s Breast Cancer Surveillance Consortium database of digital screening mammograms, Constance Lehman, MD, PhD, of Massachusetts General Hospital, and colleagues examined image interpretations of 323,973 women screened between 2003 and 2009.

Comparing a CAD cohort of 495,818 screening exams with a non-CAD group of 129,807, Lehman et al. broke off a subset of 107 radiologists who interpreted mammograms both with and without CAD, finding the counterintuitive lag of CAD in sensitivity mentioned above.

The study arrives in an era when all things digital are assumed to be improvements over their analog predecessors.

However, this also being a time of belt-tightening in healthcare, the real-world utility of pricier CAD mammography—not to mention its cost-effectiveness—has been contested in recent years.

The Lehman study is sure to provide ammo for those seeking to advance the con side of the argument.  

“CAD is a technology that does not seem to warrant added compensation beyond coverage of the mammographic examination,” the authors 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.

An accompanying opinion piece in JAMA Internal Medicine amplifies this conclusion.

Congress should “rescind the Medicare benefit for CAD use,” writes Joshua Fenton, MD, MPH, of UC Davis. “If we could curtail use of many similarly ineffective tests and interventions, we could significantly reduce U.S. healthcare expenditures while augmenting resources for effective care or well-designed studies of promising innovations.”

The FDA approved CAD for mammography in 1998 and CMS upped reimbursement for it in 2002.

Read the study here and the commentary here