An Oregon insurance provider's decision to withdraw its proposal to cancel reimbursement for computer-aided detection is a short-term win for proponents of CAD technology. But whether the victory is longer lasting remains to be seen, as CAD proponents continue to grapple with the fallout from the New England Journal of Medicine article that prompted the Oregon cutbacks.
ODS Companies of Portland on July 18 told healthcare providers in its network that it had canceled its plans to stop paying for CAD performed with imaging studies, including screening mammography. The company in June had said it would stop paying for CAD due to a study published in the April 5, 2007, edition of the New England Journal of Medicine that found that CAD did not improve the diagnostic accuracy of sites using it (NEJM, April 2007, Vol. 356:14, pp. 1399-1409).
In announcing the change in policy, ODS medical director Dr. Csaba Mera wrote that ODS makes "every effort to base benefit coverage decisions on sound evidence and community medical standards that are based on valid, reliable scientific information. While some controversy remains around the value of CAD for mammography, we will await the results of additional, well-designed studies to make any changes in our coverage of CAD for mammography."
Although ODS's most recent announcement has restored the status quo to patients and providers in its network, the question is whether other payors are also looking at CAD as a service to be cut due to the NEJM findings. AuntMinnie.com spoke to a number of radiologists, imaging vendors, and others who expressed misgivings about the methodology used in the study, and of the long-term impact it could have on CAD's future.
Radiology strikes back
A number of radiologists have criticized the methodology used in the study, which was conducted by a multicenter research team led by Dr. Joshua J. Fenton of the University of California, Davis. Some even expressed anger that the editorial staff of NEJM would publish the article as written.
Dr. Robert A. Schmidt, professor of radiology of the University of Chicago Medical Center, said he was "sorely disappointed that the editors of the NEJM, possibly in their eagerness to have a 'hot' article that would generate a lot of press, chose to allow (the Fenton) article to be published without a discussion of prior results of CAD retrospective and prospective clinical trials."
Dr. Leonard Berlin, chairman of the department of radiology of Rush North Shore Medical Center in Skokie, IL, expressed a commonly shared opinion that while the Fenton article was "extremely interesting, provocative, and convincing, nonetheless it was not a rigorous, double-blinded scientific study."
"It is unclear how the article passed peer review in the first place," said Dr. Daniel Kopans, professor of radiology at Harvard Medical School and chairman of the breast imaging division of Massachusetts General Hospital, both in Boston. "By comparing interpretations without CAD to those with CAD, in the same patients at the same time, the study is the usual comparison of 'apples and oranges.' This use of 'historical controls' has been repeatedly shown to be flawed and is the reason for the development of more scientifically rigorous study designs such as randomized, controlled trials."
After noting that "the study was performed with early software and inexperienced users," Dr. Kathy Schilling, the medical director of breast imaging and intervention at the Center for Breast Care at Boca Raton Community Hospital in Florida, was among the most critical of the radiologists who contacted AuntMinnie.com. "It was perhaps irresponsible for the NEJM to publish this article in view of all previously published information in peer-review journals. One would wonder whether the journal continues to be an objective scientific resource or a mouthpiece for payors," she said.
In addition, more than 25 peer-review studies published prior to April 2007 have documented increased cancer detection rates when CAD is used with mammography, pointed out Dr. Ronald Castellino, a radiologist and chief medical officer of women's imaging and CAD vendor Hologic/R2 Technology of Bedford, MA.
Castellino cited five prospective sequential-read clinical studies published between 2001 and 2006 in the American Journal of Roentgenology and Radiology. Collectively, more than 53,500 screening mammograms were evaluated. When CAD was added, the average increase in recall rates was 11.8%, the average increase in biopsies was 9%, and the average increase in the number of breast cancers detected was 9.7%. "This is good, valid scientific information," Castellino said. "Unfortunately, the results of these studies were not publicized like the results of the Fenton study."
Castellino believes that the high recall rate of the seven mammography centers using CAD reported by the Fenton study was the result of the technology being new at the time of the research. In addition, the radiologists using CAD were either not well-trained or were overly sensitive about the marks made by CAD.
"R2's algorithm originally approved by the FDA operated at 90% sensitivity, with a 98% sensitivity for identifying cancers that manifest as microcalcifications and an 80% sensitivity for masses," Castellino said. "Nine years later, our current algorithm is at 90% to 92% sensitivity for masses. But an increase of 1% sensitivity is not going to be noticed by a practicing radiologist, who will see only five cancers out of every 1,000 mammograms read. A radiologist would have to read 50,000 mammograms to notice a 1% improvement."
What is of importance is the continuing reduction in the number of false-positive marks with each new CAD software release, and the ability to provide ancillary information about the marks that are displayed, according to Julian Marshall, Ph.D., director of product management and principal engineer of Hologic/R2.
Today's CAD programs also enable users to set individual threshold notifications, which may differ between the highly experienced mammographer and a radiologist who is not a specialist in breast imaging. "Radiologists are not burdened with a large number of false markers as they were when CAD was commercially released for use in the United States," Marshall said.
Dr. Terri Gizienski, medical director of breast imaging services at University of Pittsburgh Medical Center -- Passavant and UPMC Passavant Cranberry in Pittsburgh, agrees with Marshall. "There have been tremendous advancements in CAD technology. It is my belief that if the Fenton study were to be performed again, applying the more recent CAD technology ... the results would show statistically significant increased sensitivity for the detection of breast cancer," she said.
Dr. Judy C. Dean, a breast imaging specialist with a Santa Barbara-based private practice, conducted one of the five studies Castellino referenced. Dean noted that of the 9,250 consecutive mammograms interpreted in her study without and with the use of CAD over a two-year period, CAD increased detection of ductal carcinoma in situ by 14.2%. "Also, the additional invasive cancers detected by CAD were significantly smaller, averaging 5 mm in size compared to 10.5 mm for invasive cancers detected without CAD. Earlier detection will both save lives and decrease medical costs, but no cancer size or stage data was considered in the Fenton article," she said.
A landmark study published in Radiology in October 2006 comparing mammograms that were double read in 1996 and reread by eight radiologists in a single reading with CAD, conducted by Dr. Fiona J. Gilbert, professor of radiology at the University of Aberdeen in Scotland, determined that single reading with CAD led to a cancer detection rate increase of 6.5%.
The economic necessity of CAD
The increasing shortage of mammographers in the U.S., compounded by the steady increase of a population requiring mammograms, is making double reading increasingly unrealistic and expensive. "The incremental increase in detection that has been shown to be possible with CAD will not be available to many women as hospitals and clinics will not purchase the equipment to perform CAD," said Dr. Janet K. Baum, director of breast imaging at the Cambridge Health Alliance in Cambridge, MA.
The reimbursement rate for the current procedural terminology (CPT) codes that apply to CAD, 77051 and 77052, is approximately $16 to $18. Mera of ODS was adamant that the payor's decision was not one based on the goal of saving substantial sums of money, but rather to avoid spending money on inappropriate technologies. Reimbursement is believed by many industry experts to be one of the reasons that CAD technology has been adopted so rapidly.
In 2003, the Web site of CAD vendor iCAD of Nashua, NH, stated that "CAD can make both clinical and economic sense for the women's health center. (F)or a center with a caseload of fifty patients per day, every dollar invested in a CAD solution could return $7.54 in new revenues. On this basis, the payback period for a CAD solution could be under nine months."
But efforts to promote CAD's economic benefits may have backfired to some extent, tainting the technology as a revenue-generating gimmick. "The potential influx of funds from computer-aided detection may make screening mammography less of a financial burden," said Dr. Joanne G. Elmore, professor of medicine at the University of Washington in Seattle and a member of the research team of the Fenton study. In a February 2004 paper published in the Journal of the National Cancer Institute, she calculated that a facility reading 59,000 mammograms receiving "a conservative reimbursement of $15 for CAD" would receive $887,000 in total additional revenue."
Elmore went on to note that if the CAD device cost $200,000, "the practice would have generated more than $650,000 in direct income from computer-aided detection alone, with no discernable improvement in accuracy or improved outcomes for the women screened."
Many radiologists point out that money does play an important part in the acquisition and reimbursement of CAD procedures. CAD used with mammography saves time, increases productivity, and enhances accuracy.
Schmidt of the University of Chicago bluntly states that "a factor in continued use of CAD technology is that CAD makes reading mammograms easier, shortening the length of time needed to perform the most arduous part of screening, the slow foveal search for tiny microcalcification clusters."
"Since the major CAD programs in clinical use mark about 98% of malignant clusters, far better than any radiologist I've ever tested does in an observer study, foveal attention needs only to be focused on the few areas ... that have calcification marks for that part of the search. I have heard 'experts' horrified at this thought, but I believe this is very pragmatically true," Schmidt said.
Kopans of MGH is outspoken in his belief that cancers can be overlooked by even the most expert radiologists. "It is an apparently immutable fact of human perception that all of us, no matter how skilled we are, will periodically fail to see a significant abnormality on a mammogram or any other imaging study that is visible in retrospect," he said. With double reading become increasingly infeasible, CAD provides the safeguard.
Dr. Mark E. Klein of Washington Radiology Associates in Washington, DC, is so adamant about the use of CAD as a safeguard that he said the radiologists in his practice "will not allow our family members to undergo mammography without CAD. CAD cannot diminish mammography accuracy. It can only help. By definition, there can be no downside." Washington Radiology Associates has used CAD with 100% of its mammograms in the past five years and would never "wish to abandon this technology."
Based on his professional experience, Klein believes that if CAD stopped being reimbursed, women would be willing to shoulder the small cost of CAD in exchange for the technology's proven ability to find cancers."
Ken Ferry, CEO of iCAD, agrees that women are informed healthcare consumers. Citing the 25 clinical studies of CAD used with mammograms to identify breast cancer, the litigation associated with overlooked diagnoses, and the increased workflow productivity offered by CAD fully integrated with digital mammography, Ferry stated that some 90% of digital mammography systems use CAD. This is in comparison with a 33% rate of CAD use with analog mammography equipment.
"For all the right reasons, CAD should be utilized. For every cancer caught earlier by CAD, insurance companies save significant sums in cancer treatment," Ferry said. "Do you give patients the opportunity to have a much higher quality of life by detecting cancers earlier? Of course you should."
Dr. Rachel F. Brem, professor of radiology and director of the breast imaging and interventional center of the George Washington University Medical Center, was shocked by the original ODS decision. "To react in this manner as a response to a flawed study that will undoubtedly result in a decrease in the diagnosis of early, curable breast cancers is not only a gross disservice to women, but to all those impacted by a diagnosis of breast cancer." She added that she was pleased to learn that ODS had changed its opinion.
By Cynthia Keen
AuntMinnie.com contributing writer
July 24, 2007
Related Reading
Oregon payor reverses stand, reinstates CAD payments, July 23, 2007
NEJM study prompts Oregon payor to cancel CAD reimbursement, July 11, 2007
Studies show CAD matches up well with CR mammo, FFDM, May 24, 2007
NEJM study pans CAD, draws attention and criticism, April 5, 2007
CAD boosts single reading of mammograms, September 29, 2006
Copyright © 2007 AuntMinnie.com
