Screening mammography

Considering the dramatic 25% decline in breast cancer mortality rates in the U.S. in recent decades, mammography screening should be a big success story and source of national pride. The Mammography Quality Standards Act is a model for boosting standards across a populous country, and recommended screening intervals appear to be on the mark.

These results have been achieved, however, with little oversight of the actual interpretation of mammograms, little continuity in provision of breast care to women, and no central means of alerting eligible women when it's time for their mammogram. Although MQSA regulates equipment in detail, its requirements for practitioners could be seen as rather lenient. Mortality rates are available, but more detailed national data about breast cancers detected and physician performance are lacking, making it difficult to analyze the state of the field in the U.S. in detail.

MQSA regulations are now in the process of being reviewed and updated. Suggestions for changes were outlined in a 2005 report from the Institute of Medicine, Improving Breast Imaging Quality Standards. The report concluded that "questions remain about the quality of interpretation, which depends on human factors and is difficult to measure in practice. The available evidence indicates that interpretive performance is highly variable in the U.S."

Experts around the globe consider outcomes monitoring crucial for further improving the practice of mammography. But in a highly litigious society, malpractice fears have put a freeze on central collection of such sensitive data.

The absence of oversight has fueled stories in the mainstream media about problem doctors. In 2002, The New York Times ran a series of critical articles, including an article entitled "Spotting breast cancer: doctors are the missing link." It claimed that "many doctors and their clinics are compromising whatever precise value mammography has" and that MQSA "does not address the biggest problem of all-skill of doctors who interpret mammograms." The article pointed out that the government does not monitor doctors' performance and claimed those doctors who miss too many cancers are shielded.

Talk about kicking a profession when it's down. Although screening has been proven to save lives, that doesn't mean providers are satisfied with their lot. Reimbursement is low, and government regulation, though effective, is onerous, costly, and unfunded. About one-third of the roughly 1500 mammography facility closures from 2001 to 2004 were attributed to financial factors, according to the American College of Radiology.

Mammography work can be tedious, and malpractice is a constant threat. Many cancers are visible in retrospect, and even the best mammographers miss cancers, but juries might not take that fact into account when faced with a distraught woman who has suffered a delay in diagnosis at the hands of an expert. The legal risks partly explain high recall rates and an extremely high rate of biopsies of benign lesions in the U.S. relative to European countries. Attitudes in the U.S. are changing, not for the better.

"U.S. physicians were much more interested in providing mammography 15 years ago," said Swedish breast radiology expert Dr. Laszlo Tabar. "Now, many are depressed about it and want to get out of it."

BABY BOOMERS BOOM

Meanwhile, the baby boomers have come of screening age. The population of women over 40 will increase by 50% by 2025, while the number of practitioners per 10,000 women over 40 will dip by 23% during that period.

As women become more educated about the value of mammography, they can be expected to demand a higher quality of care that will be difficult to satisfy unless workforce supply issues are addressed. Despite the looming challenges, and amid ominous talk of a crisis, Tabar sees a bright side to the difficulties surfacing in U.S. mammography.

"This is a cleansing process. It may be that too many radiologists are reading mammograms. Not all radiologists can provide that perfect service women expect," he said.

In the view of Tabar, a professor of radiology at Uppsala University School of Medicine in Sweden who has given more than 300 courses in the U.S. since 1986, and others who specialize in breast radiology, mammography should be performed by experts in high-volume, centralized breast centers.

Just as the U.S. built up a network of high-quality heart institutes, it should create centers of excellence for breast care, Tabar said. The IOM report calls for the creation of such centers in the U.S., though at press time it was unclear whether funding would be forthcoming. The expert breast radiology model is not a popular suggestion in the U.S., mainly for practical reasons related to physician supply and financial limitations.

Such centers have been established since the 1970s in Sweden, which has one of the most enviable screening programs in the world. Compliance ranges from 85% to 95%, and mortality rates have been reduced by 43%. Tabar said he is confident that people will start to realize that breast cancer is a very complex disease-more complex than heart disease-that requires specialist attention.

"There are 25 different subtypes of breast cancers, which all behave differently and look different on a mammogram," Tabar said. "No other generation of physicians in the past 1000 years has seen as much ductal carcinoma in situ as we see today. Do we know everything about DCIS? No, we are the first generation that sees it."

Under the terms of MQSA, radiologists are required to read at least 480 mammograms a year. But if there are only four or five cancers per 1000 studies per year, those who read 500 studies might see only two or three cancers per year. In Swedish screening facilities, radiologists typically read 200 cases a day and are therefore exposed to more cancers.

Instead of relegating screening to someone who is on part-time mammography duty and dislikes the tedious and low-paid work, the images should be read by a person who is subspecialized in breast imaging, said Judith Wagner, a nurse and patient advocate.

"Who do you want reading your mammogram-the guy who reads 5000 a year and is committed to breast care or the guy who reads 480 because he has to?" Wagner said.

During visits to women's groups, Wagner educates consumers about assessing mammography quality, including recall rates, positive predictive value, size and stage of cancers detected, and radiologists' reading volume.

"We need to make breast care a priority in this country. Radiologists should be designating someone in the group as a clinical breast radiologist. Hospital systems need to work with radiology groups, and the government needs to look at raising reimbursement," Wagner said.

Dr. Kamilia Kozlowski, medical director at the Knoxville Comprehensive Breast Center in Knoxville, TN, expressed similar views.

"I think the 480 minimum is insignificant. Studies have shown that radiologists who read mammograms all the time have much better sensitivity and accuracy rates for finding breast cancers than general radiologists," she said.

Therefore, though MQSA has improved technical quality, second opinions are still necessary due to problems with interpretation, causing unnecessary anxiety and delays in diagnosis for patients. This in turn results in greater malpractice risk, she said.

Volume is just one factor that helps radiologists build up expertise. Tabar recommends at least 10,000, but preferably 25,000, studies per year. In the U.K., radiologists are expected to read a minimum of 5000 mammograms per year in a limited number of highly centralized screening programs. European guidelines also recommend 5000 per year for screening mammography.

Breast radiology in the U.K. is viewed as a highly professional, highly trained multidisciplinary specialty, said Dr. Nick Perry, director of breast screening at St. Bartholomew's Hospital in London.

"It is certainly a far too complex and risky specialty to just dabble in. Just because something is common does not mean that it is easy. The frequency of breast cancer and benign breast conditions, together with the level of anxiety for women and their families, do not in my opinion allow for anything less than a fully dedicated, trained, experienced, caring, and professional service," Perry said.

MQSA requirements reflect prevailing practice at the time the guidelines were developed in the early 1990s, when the practice was strongly generalist, said Dr. Kim Adcock, physician director of risk management for the medical group at Kaiser Permanente Colorado, where radiologists are required to read at least 4000 mammograms a year and must submit to regular assessment.

"It's tantalizing to suggest the federal government could simply legislate higher quality mammography. But that might have a strongly counterproductive effect of driving radiologists out of mammography by focusing scrutiny on one area of practice that simply does not exist elsewhere," Adcock said.

Others agree that raising the bar could decrease access for women.

"Any number you pick would be debated. Is 480 too low? It might be unless you are somewhere in northern Montana and the only radiologist in a small town. Not everyone works in a big city where you have the opportunity to read 10,000 mammograms a year," said Dr. Ellen Mendelson, a professor of radiology at Northwestern University.

A universally accepted approach (centralized, subspecialized, high-volume approach), as practiced in some European countries, may not be accepted or even desired at this time in the U.S., said Dr. Richard Ellis, codirector of the Norma J. Vinger Center for Breast Care in LaCrosse, WI.

"In the U.S., the issue of access seems to rule over a centralized, subspecialized, high-volume approach due to travel constraints. For example, the U.S. is a much larger country, and if women were asked to travel too far, attendance could be an issue," Ellis said.

Studies show benefits from not just reading a larger number of exams, but also from more education and reinforcement of skills and medical audits, according to Dr. Edward Sickles, a professor of radiology at the University of California, San Francisco.

"There is a lot of evidence that you get better results if you do all these things, but very little that one component-namely volume-is important in and of itself. The Europeans are convinced it is, but I don't think the evidence is there," he said.

The IOM report did not recommend an increase in the volume requirement and concluded that several factors contribute to improving physician performance. The report also explored workforce challenges, noting the current volumes of mammograms read by radiologists (see chart above). Raising the minimum level to 1000 would hit small metropolitan areas the hardest, as 37% would have to either increase volume or stop performing mammography.

The report also suggested, however, that new centers of excellence could test the importance of a range of factors, including increased volume, double reading, and feedback, in improving physician performance and service to women. Crucially, it also mentioned that more incentives are needed to encourage radiologists to specialize in mammography, such as greater financial rewards.

Although the U.S. does not appear to be moving toward a centralized system, there is a voluntary movement toward developing regional breast centers with highly trained, specialized breast radiologists, expedited care, and nurse navigators to help patients through the system.

Hundreds of centers across the country have embraced this approach, including Kozlowski's facility. Many are members of the National Consortium of Breast Centers, which represents almost 1000 breast care facilities in the U.S. Interest in this approach is growing, even though specialists are in very short supply.

Kozlowski says it is necessary to create a system that is vertically oriented and centered around a specialist breast radiologist who manages diagnosis and follow-up of breast disease. After an abnormal mammogram, the patient would be followed up immediately by a radiologist instead of being referred to a surgeon and then possibly back to the radiologist again for further imaging studies. Delays are avoided, follow-ups are less likely to fall through the cracks, and imaging procedures are always performed by skilled imaging specialists rather than nonradiologists practicing radiology on the side.

"Breast radiologists have better sensitivity in detecting breast cancer. If they were in charge of delivery of diagnostic breast care, we would have fewer cases of missed diagnoses and there would be fewer malpractice cases," she said. "Until clinical breast radiology is recognized as a subspecialty, we are not going to make a dent in changing the delivery of breast healthcare," Kozlowski said.

MAINTAINING ACCESS

In formulating its recommendations, the authors of the IOM report sought guidance from European sources. Perry, for example, was invited to share his experience with centralized screening in the U.K.

Some aspects of European-style centralized screening could be adopted in the U.S., Perry said. The U.K. set up a network of national training centers with a high-volume workload, and radiologists are required to visit these centers as part of their training. The U.S. centers of excellence mentioned in the IOM report could play a similar role in mammography training.

Access in underserved areas could be achieved partly through mobile mammography, which is routinely deployed in Sweden and the U.K., where vans stay on the road for months at a time. The IOM noted that a mobile mammography program based in Rapid City has been providing screening and diagnosis to sparsely served areas in central and western South Dakota for years.

Radiology assistants could also be used to a greater extent in breast imaging, possibly in prescreening or second reading under supervision, just as nonphysicians now read Pap smears. Demonstration programs can explore the feasibility and effectiveness of radiology assistants in mammography screening, according to the report.

In the U.K., most mammograms are read by radiologists, but some are also read by highly trained technologists, who are expected to meet the same standards, including the minimum screening volume of 5000 cases per year. All mammograms are double-read, preferably by a radiologist. Audits indicate the technologists' performance is broadly comparable to radiologists', although specificity was slightly lower, and the country is encouraging greater use of nonphysicians in screening. U.K. radiologists are also strongly encouraged to undergo self-assessment, including reading 60 mammograms in set conditions with instant feedback about whether cancers are being missed.

"Performance is monitored and remedial action taken if a radiologist performs below what is regarded as an acceptable level," Perry said.

The ACR offers a self-assessment program, but participation has not been strong.

STANDARDS VARY

Screening is still not standardized in Europe, where breast imaging is subject to a mixture of centralized and decentralized systems. In France, mammography is performed in small volumes at offices, but cases are then transmitted to a high-volume screening program at a center of excellence for a second reading. This approach has proven successful, but it is unclear whether it is as cost-effective as screening in high-volume centers directly.

Breast cancer mortality rates vary significantly in European countries, with Finland and Sweden at the low end and Belgium and Denmark at the high end. Data indicate that the U.S. has lower mortality rates than a number of European countries, possibly due to its recommendations for annual screening and for screening of women in their 40s. In contrast, most European countries recommend screening every two years for women over 50. Women in the U.K. are screened only every three years.

In an effort to even out the disparities between them, European countries have developed a pilot network of centers of excellence, seeking to replicate performance in countries with the most advanced centralized screening programs, such as Sweden and the U.K.

Detailed guidelines have been created to help assess performance and to establish national screening programs where they do not yet exist. The fourth edition, European guidelines for quality assurance in breast cancer screening and diagnosis, was published in April 2006. Perry is editor-in-chief of the 400-plus-page document. The guidelines are being promoted by the European Parliament, the legislative branch of the European Union, but they are not funded centrally.

The guidelines aim to reduce both mortality rates and unnecessary workup of benign lesions. One focus is on collecting data for screening results and outcomes. The guidelines advise screening every two years for women 50 to 69 and outline 39 performance indicators, including the following (also see chart, page 35):

- recall rate;

- benign-to-malignant biopsy ratio;

- proportion of screen-detected cancers that are invasive;

- size of cancers detected;

- interval cancer rates;

- proportion of cancers with lymph node metastases; and

- time in working days between screening mammography and result.

Ideally, the recall rate would be less than 5% at the time of initial screening and less than 3% for subsequent screening. The concept of recall rate ceilings does not sit well with some U.S. radiologists.

"Holding radiologists to arbitrarily low values saves dollars and contains costs, but it doesn't save lives," said Dr. Stephen Feig, chair of the committee on communications for the ACR Commission on Breast Imaging.

Tabar points out, however, that optimal recall rates vary greatly, depending on the patient population, and that it's not appropriate to compare recall rates in the U.S. with those of European countries such as Sweden. Recall rates in the first round of screening women aged 40 to 49, for example, will be very different from recall rates in subsequent rounds of screening in women aged 70 to 74 years.

In Sweden, a central registry notifies women when it is time for a mammogram, and participation is very high at 85% to 95%. Women show up on time for their study, and prior mammograms are available.

If a radiologist sees the same women over the course of decades, the recall rate can drop dramatically. In Tabar's practice, for example, the recall rate is 2.5% to 3%, and breast cancer mortality rates have been halved.

Compliance in the U.S. is estimated at about 70%, and it's unclear whether women are being screened at recommended intervals. Prior mammograms may or may not be available.

Similarly, it may not be fair to compare biopsy ratios. In the U.K. screening program for 2004-2005, there were 7.3 malignancies biopsied for every benign case.

"It is an incredibly good figure. With the state of litigation in the U.S., it would be hard to achieve that, and for good reason. But we don't have a blase attitude because we don't get sued as much. The ratio reflects centralization, a high degree of training, a high degree of experience at high volume, and accuracy of assessment techniques," Perry said.

Statistics in the U.K. are constantly monitored. All data on the national screening program are entered into a database and subject to regular audits from multiple sources at local, regional, and national levels. Audit data in 2002 showed that results were better in facilities with higher screening volume, and, consequently, the country began recommending a minimum volume of 9000 per year.

"Collection and publication of data are necessary for screening programs to be running optimally, as there cannot be a satisfactory quality loop system without it. How can you know what is going on unless you have the data to analyze? And, by the same token, how can you learn from your mistakes?" he said.

OUTCOMES AND GOALS

MQSA requires providers to track outcomes only for women referred for a biopsy after an abnormal mammogram. Data for screening and diagnostic mammography are not typically teased out.

Desirable goals for mammography were created in the early 1990s and incorporated into BI-RADS. The Breast Cancer Surveillance Consortium (BCSC) has collected a vast amount of data about performance parameters to provide a gauge of performance in the U.S., based on these goals. BCSC benchmark data for diagnostic mammography in 151 mammography facilities were reported by Sickles et al in Radiology in June 2005. Data for screening mammography in BCSC centers are in press.

The data will help contribute to understanding of best practice, which has not been fully defined in the U.S. It should be a mixture of audit parameters and practices that combine the best of U.S. and European experience, Ellis said.

"Whatever metrics you use, they must be centered around the mean and median size of invasive cancer found at screening, as this affects patient morbidity and mortality," he said.

Dr. Michael Linver, who served on the panel that established federal guidelines for quality standards in mammography, recalls that back in the 1980s, he learned about the value of gathering statistics to assess practice at a lecture by Tabar.

"It was like an epiphany. I thought: 'This is what I want to do for the rest of my life,'" he said.

But in the U.S., auditing is mostly voluntary. MQSA requirements for auditing were "watered down tremendously for a variety of reasons," according to Linver, staff radiologist at the Breast Imaging Center in Albuquerque. The biggest obstacle was that the data could not be protected from discoverability, or use by lawyers. There was a fear that audit data could be misconstrued and used to damage physicians.

In the U.S., data collected for peer review within institutions are immune from discoverability because courts recognize that if such information was publicly available, peer review would not exist. But anything a radiologist voluntarily turns over to a central database or other third party will certainly be discoverable, said Dr. Leonard Berlin, chair of the ACR Ethics Committee.

The IOM report suggested the possibility of carving out mammography and introducing no-fault liability for those radiologists proven to take part in quality assurance programs. But it is difficult to discriminate between various medical specialties. In the past, there was a movement to get a carve-out for obstetricians, but it didn't get to first base, Berlin said.

"I don't think there is much likelihood of carving out mammography, and I don't know if it would fair," he said.

The IOM report also proposed prorated malpractice insurance and loan repayment awards to encourage physicians to perform mammography, but it is unclear whether these suggestions are feasible.

"I don't believe 'prorated,' 'graduated,' or any other kind of differentiated malpractice insurance premiums for mammographers will ever occur. Even in the unlikely possibility that a state did pass such legislation, it would be challenged before the courts, which I predict might well rule that they are illegal," Berlin said.

Absent legislation, it's likely that the data collection will continue to be done on a voluntary basis. The IOM report has recommended that MQSA offer a voluntary advanced medical audit with feedback.

The screening program at Adcock's facility won a national Kaiser award for demonstrating a 10% increase in the number of mammographically detected cancers at stage 0 or 1. The improvement came through specialization, feedback, and outcomes analysis, he said. Each radiologist is tested on a mixture of known but subtle breast cancers three or four times per year.

Initiatives are also possible at the state level. Wisconsin has launched the Wisconsin Collaborative for Healthcare Quality, for example, a voluntary statewide consortium of organizations committed to measuring and improving health in a range of conditions. The collaborative will set performance standards and release results for physician group performance in screening for a range of cancers, including breast disease.

Meanwhile, the Norma J. Vinger Center is gearing up to publish five years' worth of its own audit data. The center plans to apply for a grant to analyze another five years of performance, this time with a cost-effectiveness component.

"Stage at diagnosis has a direct impact, not only for the outcome of the patient but for the cost of care," Ellis said.

Experience in such dedicated breast centers as well as facilities participating in the BCSC will help shape the future of mammography.

"This is a period of transition for the U.S. We are providing more and more research data regarding the early phase of this disease. I am here standing on the sidelines, hoping the evolution of mammography services in the U.S. goes in the right direction," Tabar said.

Ms. Hayes is feature editor of Diagnostic Imaging.