Have an account? Please log in.
Text size: Small font Default font Larger font
Radiology Daily
Radiology Daily PracticalReviews.com Radiology Daily

Mammogram Study: False-Positive Odds High

October 18, 2011
Written by: , Filed in: Breast Imaging
  • Comments

In 10 years of annual mammography screening, more than half of women—61.3 percent—will receive at least one false-positive result, according to a study published online today in Annals of Internal Medicine.

Switching to every-other-year screening, the study found, reduces the 10-year false-positive rate to 41.6 percent but, not surprisingly, may slightly increase the risk that any breast cancers that are detected will be in late stages.

The study found that the false-positive rates were similar whether women started screening at age 40 or 50. The researchers examined data from more than 169,000 screening mammograms in the registries of the Breast Cancer Surveillance Consortium, an offshoot of the federal National Cancer Institute.

Lead author Rebecca A. Hubbard, PhD, an assistant investigator with the Group Health Research Institute in Seattle, told HealthImaging that women and their doctors “can use this information along with a woman’s individual risk tolerance and information on her personal breast cancer risk to make informed decisions about the best screening strategy for her.”

Dr. Hubbard also noted:

An important finding of our study was the substantially decreased risk of false-positive results when comparison films were available. Breast imagers should encourage women to ensure that their previous mammogram is available.

In another study published at the same time, many of the same researchers found that for premenopausal women in their 40s with dense breasts, digital mammography improved cancer detection but also slightly increased the rate of false positives, as compared with film mammography.

In an accompanying editorial, Philippe Autier, MD, of the International Prevention Research Institute in Lyon, France, looks at this study and others, and concludes:

In summary, current evidence indicates that mammography screening every year is less efficient than screening every two years or more.

Which is exactly NOT the kind of language that will sell women on the idea of biennial screening. People care a lot more about whether the health-care system protects their health than about whether it’s efficient.

Annals did also publish a freely available summary of the main study, aimed at a general audience and written in nontechnical language. But mammography authorities and providers still have a long way to go in effectively communicating their recommendations to the public.

Of course, it doesn’t help that they don’t agree among themselves.

* * *

We think you should follow us on Twitter. Today, for example, you could find out why you should follow us on Twitter. And Facebook.

Related seminar: Pittsburgh Breast Imaging Seminar


Permalink: http://www.radiologydaily.com/?p=7434


  • No Related Posts
  • Comments

Would you like to keep current with radiological news and information?

Post Your Comments and Responses

One Response to “Mammogram Study: False-Positive Odds High”

  1. Daniel on October 21st, 2011 at 7:45 pm


    It is difficult to address the paper by Hubbard et al (1), which purports to analyze the false positive and biopsy rates for mammography screening, since critical information is lacking. It would appear that they compared women who were, apparently, screened every year and compared them to other women who chose to be screened every two years. This was not a randomized, controlled trial so the fact that women in one group, apparently, decided on their own to wait two years between screens, likely, introduces a major bias. The paper suggests that there were only small differences in the stage of the cancers detected with 1 year screening intervals compared to 2 years. However, as noted below, their data are misleading. They do not provide actual size distributions which is critical. If their stage distribution reflects the sizes of the cancers then they need to explain why the cancers in their study did not grow larger among the women who were screened every two years compared to those screened every year. It makes no biological sense that the cancers would be the same size when one group had a year longer for their cancers to grow.

    It also appears that they “modeled” (extrapolated) the recall rate and did not measure it directly. With such a large number of women, it is unclear why the recall rates were not measured, directly. In a study from the Massachusetts General Hospital screening program, rates were counted directly (with no extrapolation or “adjustments”) to provide the number of women actually recalled for additional evaluation over a 10 year period. We found that only 30% of women were recalled from screening. These authors report more than twice that number.

    There are numerous “adjustments” that the authors made in the data. It would seem that, given the large number of women in their database, they only needed to look at the actual frequency each woman was recalled, the frequency of the biopsies requested, and the frequency and size of the cancers diagnosed and whether or not lymph nodes were positive (and their number). The fact that they seem to have done a lot of “adjusting” is of some concern. Adjustments can be tailored to provide whatever outcome the adjuster wishes to write into the adjustment. It is unclear why there was any need to adjust the results. At a minimum, the unadjusted data should have been provided so that the effects of “adjustments” could be determined.

    The most important question remains unanswered – Are cancers found at a smaller size when women are screened every year vs. every two years? The authors provide some information on stage at diagnosis. These stage data, however, are insufficient to analyze the impact of screening. Staging is an effort to group cancers that may be similar, based on their size and whether or not lymph nodes in the axilla are involved with tumor, and whether or not the patient’s cancer has metastasized. Staging is a fairly crude way of grouping tumors that may, actually, behave quite differently. The actual data on the size of the cancers and the number of lymph nodes involved are critical for trying to determine the possible consequences of extending the time between screens (screening interval) to every two years rather than every year. This study was not a randomized, controlled trial, which is the only way to really determine the efficacy of an intervention. In the randomized, controlled trials the decrease in deaths was due to finding smaller cancers within stages. The key is to find smaller cancers. The greatest benefit is finding cancers at 1.5 cm. or smaller. Cancers found at 1.0 cm. do uniformly well (3). Lowering the size at diagnosis is what saves lives. As Hubbard et al point out, even using the crude measure of stage data, annual screening results in a higher percentage of earlier stage cancers. It would be important to look at the actual size of the cancers in relation to screening interval, and not the grouped stage data since the latter can be very misleading. The authors should have provided data on how screening interval influenced the size at diagnosis.

    It would also appear that the authors did not separate women who had more than one recall, or biopsy, from all the other women. It appears that these were averaged over the groups. Women who are at high risk, for example, might have a number of recalls and biopsies which, if not attributed to the same person, would inflate the numbers for the entire group. To exaggerate, one woman biopsied every year over 10 years is not the same as 10 women each having a biopsy.

    The United States Preventive Services Task Force (USPSTF), who were supported by Dr. Kerlikowske, the second author on this paper, used computer modeling to address the question of annual vs. biennial (every two years) screening. They clearly showed on page 742 (4) that, based on the available data, annual screening beginning at the age of 40 saves many more lives than biennial screening beginning at the age of 50. This was confirmed in a paper by Hendricks and Helvie who showed that, using the USPSTF models, biennial screening beginning at the age of 50, would cost the lives of as many as 100,000 women who are now in their thirties, that could be saved by annual screening beginning at the age of 40 (5).

    The main argument against annual screening has been called the “harms” of screening by those who oppose it. These “harms” are primarily the recall of women for additional evaluation for what proves to not be cancer. The problem is that those who oppose annual screening because of these “harms” fail to explain that most women who are recalled have a few extra mammographic images obtained or an ultrasound, that show that everything is fine. A small percentage of women are asked to return in 6 months just to follow-up on something that is, almost certainly, not cancer, but the radiologist would like to make sure it does not change with time. Only about 1-2% of women who are screened are advised to have a biopsy (usually a needle biopsy using local anesthesia as an outpatient). Of these latter women, 20-30% will be found to have breast cancer. This yield of cancers per biopsy is very reasonable when compared to other similar screening tests such as cervical cancer screening. The yield of cancers due to a mammographically instigated biopsy should be compared to when a patient or her doctor feels a lump that is biopsied by a surgeon. The yield of cancer in women with lumps is much lower (higher false positive rate) than biopsies performed for mammographically detected lesions, and when the palpable lesion turns out to be cancer, it is usually larger and at a latter stage, than cancers found by mammography.

    Because of the incomplete information in this paper, it is difficult to evaluate its importance. A simple way of looking at the question of how often to screen is the following. On some day a cancer becomes visible on a mammogram. If a woman has her screening mammogram the day before the cancer is visible is she better-off waiting for two years to have her next mammogram, or one year? Is it better to let a cancer grow for one year or two?

    Mammography screening, that began in the U.S. in the mid 1980’s, is the main reason that the death rate from breast cancer, that had been unchanged since 1940, began to decrease in 1990. Primarily as a result of mammography screening, the death rate from breast cancer has declined by over 30% since 1990. This means that there are more than 30% fewer women who die of breast cancer each year than would have been expected to die based on the pre-screening era. Therapies have improved, but the reason that oncologists have not been calling for a cessation to annual screening beginning at the age of 40 is because they well know that the best way for their therapies to cure breast cancer is to find it early. It is, probably, not a good idea to set the clock back by lengthening the screening interval.

    Daniel B. Kopans, M.D.

    Professor of Radiology Harvard Medical School
    Senior Radiologist Breast Imaging Division – Massachusetts General Hospital


    1. Hubbard RA, Kerlikowske K, Flowers CI, Yankaskas BC, Zhu W, Miglioretti DL. Cumulative Probability of False-Positive Recall or biopsy Recommendation After 10 Years of Screening Mammography. Ann intern Med. 2011;155:481-492.
    2. Blanchard K, Colbert JA, Kopans DB, Moore R, Halpern EF, Hughes KS, Smith BL,Tanabe KK, Michaelson JS. Long-term risk of false-positive screening results and subsequent biopsy as a function of mammography use. Radiology. 2006 Aug;240(2):335-42
    3. Tabar L, Fagerberg G, Day N, et al. Breast cancer treatment and natural history: new insights from results of screening. Lancet 1992; 339:412–414
    4. Mandelblatt JS, Cronin KA, Bailey S, Berry DA, de Koning HJ, Draisma G, Huang H, Lee SJ, Munsell M, Plevritis SK, Ravdin P, Schechter CB, Sigal B, Stoto MA,Stout NK, van Ravesteyn NT, Venier J, Zelen M, Feuer EJ; Breast Cancer Working Group of the Cancer Intervention and Surveillance Modeling Network. Effects of mammography screening under different screening schedules: model estimates of potential benefits and harms. Ann Intern Med. 2009 Nov 17;151(10):738-47
    5. Hendrick RE, Helvie MA. United States preventive services task force screening
    mammography recommendations: science ignored. AJR Am J Roentgenol. 2011