Computers hinder mammogram readings, report finds
BOSTON (Reuters) - Computer-aided mammogram designed to help doctors spot cancer do not increase the chance of finding a tumor and, instead, heighten the risk that a woman will get an unnecessary biopsy, researchers reported on Wednesday.
"This study points out the need for the use of other techniques to find cancer at its earliest stages," said Dr. John Niederhuber, director of the National Cancer Institute, which helped pay for the study.
Dr. Joshua Fenton of the University of California, Davis, and colleagues studied more than 429,000 mammograms for their study, published in The New England Journal of Medicine.
About 24 million screening mammograms are taken in the U.S. each year. "We would guess maybe 25 to 30 percent of facilities have adopted this, maybe more in urban centers where they have a high volume," said Fenton.
The results "constitute a substantial hit to this technology" and will "surprise and disappoint" most doctors who read mammograms, Dr. Ferris Hall of Beth Israel Deaconess Medical Center, wrote in a commentary.
Hall said Medicare pays an extra $20 for mammograms that are read by computer, a financial incentive that "was mandated by a heavily lobbied Congress, despite little evidence-based data in support of its value at the time."
Complicating the issue is the fact that the field is changing so quickly, further research may not be practical, he said. "Such studies will be expensive, controversial, indeterminate, or quickly passe owing to the emergence of new technology," Hall wrote.
Three such computer-aided devices, costing $160,000 to $240,000, have been approved by the U.S. Food and Drug Administration.
Most of the facilities in the Fenton study used units from R2 Technology Inc. of Santa Clara, California, the first to get FDA approval, in 1998. R2 is owned by Hologic Inc.
Kodak and iCAD Inc. of Nashua, New Hampshire also make units.
The research team used mammograms taken from 1998 to 2002 at 43 medical facilities in three states, seven of which switched to computer-aided detection in the middle of the study.
With human-read mammograms, 98 out of every 1,000 women were mistakenly told they were free of cancer. When the readings were done with the help of a computer, the number rose to 128 out of 1,000, without significantly increasing the number of tumors that were spotted by X-ray.
In addition, the researchers said the computer programs tended to focus on the least-dangerous types of cancers.
"There was no clear benefit in terms of breast cancer detection," Fenton said in a telephone interview.
The researchers also estimated that if every medical center used computer-assisted detection, it would cost the U.S. health care system an extra $550 million, an increase of 18 percent in the cost of doing breast cancer screening exams.
There is a lot of pressure to improve detection.
Hall said missed tumors are the most common source of lawsuits against radiologists, and in as many as half of all cancer cases, doctors turned out to have missed the tumor in an earlier mammogram. And with many medical students avoiding the field because of the stress, there is a shortage of good mammographers.
Hall said one alternative would be magnetic resonance imaging. Although it may detect 10 times as many cancers than mammography or physical examination, it is also 10 times more expensive.
"In certain populations of women, MRIs are much more sensitive to picking up cancers than mammography," said the chairwoman of the American Cancer Society's Breast Cancer Advisory Group, Dr. Christy Russell.
For those women, the chance of an MRI finding a tumor is 70 percent or higher, compared to just 30 percent for mammography or ultrasound, she said.
What breastcancer.org says about this article…
Computers hinder mammogram readings, report finds
The study reviewed here found that using computers to help read mammograms leads to more false positives and more unnecessary biopsies—without finding a significant number of additional breast cancers.
While this isn't good news, it doesn't mean that we should completely give up on computer-aided detection.
It's important to know that computer-aided detection IS NOT digital mammography.
In computer-aided detection, a computer program is used to highlight areas on the mammogram image that may be abnormal. A radiologist looks at the highlighted areas and decides if you need more tests or not.
The computer highlights areas that MAY be abnormal. The radiologist has to decide if the areas really are abnormal. Computer-aided detection is not the same as having a second radiologist look at the mammogram. It's still only one radiologist looking at the mammogram. An additional read by a second radiologist improves detection and reduces false positives.
New technology offers much promise, but there is always a learning curve. In this case, doctors may be relying too much on the computer-aided detection results. Or they may be intimidated by the computer's results and may be hesitant to question what the computer says.
Computer-aided detection shouldn't be abandoned because of this study. Instead, we should use these results to study how doctors use computer-aided detection. By comparing the computer-aided detection results with the pathologic findings from any biopsies that are done, doctors can learn more about the best way to evaluate areas that computer-aided detection highlights. This could help doctors more accurately use computer-aided detection results and make mammograms better screening tools for all women.
