Computer-aided mammography might not help spot breast cancer

By Kathryn Doyle (Reuters Health) - Computer-aided detection, a part of almost all mammograms today, appears not to improve diagnostic accuracy for breast cancer screening, according to a large study. Computer-aided detection (CAD) marks areas of concern on mammogram readouts that radiologists might otherwise miss. It was approved by the Food and Drug Administration in 1998 and became common after the Centers for Medicare and Medicaid Services increased reimbursement for the technology in 2002, the authors write. Since then, it has been hard to measure how much difference CAD actually makes in spotting cancers. “There’s been a lot of controversy about whether or not CAD improves performance, and the largest studies haven’t supported improvement,” said Dr. Constance Lehman of Massachusetts General Hospital in Boston, who led the new study. “Prior studies were done on CAD technology on the very early curve of its adoption,” and radiologists may not have been comfortable using it yet, Lehman told Reuters Health by phone. For the new study, she and her coauthors only considered mammograms reviewed by radiologists who had been working with CAD for at least a year. Between 2003 and 2009, the nearly 324,000 women in the study had nearly 626,000 digital screening mammography exams, including 495,818 that were interpreted with CAD and 129,807 that were interpreted without it. The mammograms were interpreted by 271 radiologists at 66 facilities. Within a year after mammography, 3,159 women were diagnosed with breast cancer. Radiologists correctly detected a breast cancer about 85 percent of the time, regardless of whether CAD was involved. In other words, the cancer detection rate did not change based on CAD technology. When the researchers looked specifically at about 100 radiologists who interpreted mammograms with and without the computer readout, they found that the radiologists actually detected fewer cancers with CAD mammograms. CAD now costs more than $400 million yearly, without an established benefit for patients, the authors write. “Literally hundreds of thousands of women are undergoing this and in many cases paying out of pocket,” Lehman said. Radiologists should be thoughtful about how they are using CAD – it may be useful for spotting calcifications, but just because it does not mark an area as concerning doesn’t mean that area should be ignored, she said. Dr. Joshua J. Fenton of the University of California, Davis Health System in Sacramento, who wrote an editorial published with the findings in JAMA Internal Medicine, doesn’t think patients should be alarmed. “CAD seems to be having, on balance, little if any effect,” he told Reuters Health by phone. Fenton said he’s not surprised that CAD did not yield substantial benefits, and he questions whether Medicare should be reimbursing for it. “If it weren’t reimbursed then a lot of radiologists would have the freedom to say I don’t want to use CAD,” Fenton said. To change reimbursement, strong political leadership would need to make a compelling argument for why CAD is not a good use of taxpayer money, he said. “We need to be more rapid in assessing new technology,” Lehman said. “We are advancing so quickly in technology development, we need to get quicker at assessment.” “In my practice I’m not sure if I’m going to stop using CAD, but I question if it makes sense to charge for CAD if I’m using it to diagnose cancer,” Lehman said. SOURCE: http://bit.ly/1QIEzMs http://bit.ly/1KOjVqB JAMA Internal Medicine, online September 28, 2015.