What We Can Learn from New ACS Guidelines on Screening Mammography

Since 2003 the American Cancer Society (ACS) has told women repeatedly to get screened with mammograms every year starting at age 40, or younger if they are at higher risk. The society also recommended clinical breast exams (CBEs) each year for women over age 40 and periodically for women younger than that. During this year’s National Breast Cancer Awareness Month [1], however, the ACS revised its screening guidelines. Based on an independent panel review of current evidence on screening mammography [2], the 2015 recommendations are specific to women of average risk.[3]

ACS guidelines Figure

JAMA, http://jama.jamanetwork.com/article.aspx?articleid=2463258

The new protocols published in JAMA advise average-risk women to wait until age 45 to begin screening but emphasizes that women should continue to have the option to screen at age 40 (and that insurance should continue to cover it). Since the data reviewed suggest that by age 55 the reduction in the average woman’s chances of dying from breast cancer would be the same even if she skipped screens every other year, the recommendation for women 55 and older is biennial screenings.

Other ACS recommendations are based on evidence that is inferred rather than observed directly. For women ages 45 to 54, the ACS recommends annual screens based on evidence that it decreases the number of more advanced cancers detected. Unfortunately, this judgment call is based on incomplete evidence since the survival benefits for this age group are not known [4] in the studies supporting that conclusion. Likewise, for women over age 74 who are in good health and have a life expectancy of more than 10 years, ACS recommends continuing their biennial screens even though there is no evidence of a screening benefit for this age group. Though there are few known harms to doing CBE, ACS does not advise the practice for average-risk women at any age to screen for breast cancer; this too is based on incomplete evidence on benefits. Thus, ACS is not consistent in how it weighs incomplete evidence.

USPSTF guidelines Figure

JAMA, http://jama.jamanetwork.com/article.aspx?articleid=2463258

The ACS position on screening differs from the US Preventive Services Task Force systematic analysis in 2009 that found insufficient evidence of survival benefit to screen women in their 40s or women aged 75 and older. The USPSTF recommendation to stop routine screening in these age groups would also counter the harms of false positives and unneeded biopsies, factors now well documented in a variety of studies. For women aged 50 to 74, the USPSTF found that biennial screens led to the same relative reduction in deaths from breast cancer overall but with half the radiation exposure. The independent panel also found insufficient evidence on benefits from clinical breast exam but stopped short of the new ACS position to drop CBEs entirely.

Reaction to the new advice from ACS was swift. With the media’s focus on the society’s removal of five years worth of mammograms for women aged 40 to 45 and halving the frequency to every other year at age 55, some radiologists and their professional organizations protested angrily as did some advocacy groups and health professionals. Others applauded the recommendations as a step closer to offering women in their late 40s an evidence-based protocol. But what undergirds the new recommendations provides fascinating object lessons on biomedicine. What might look like even-handed attention to different types of investigations—Randomized Controlled Trials (RCTs), statistical modeling, and observational studies—may obscure key assumptions about how experts judge evidence.

When asking what the benefit of mammography screening actually is, the answer depends on how you study it.

In round numbers, statistical models estimate that women of all ages who get screened regularly with mammograms reduce their chances of dying from breast cancer by 15 percent relative to the unscreened groups under study. The absolute (actual) risk of dying from breast cancer is much lower and differs by age. When based on RCTs, the relative reduction ranges from 0 to 25 percent across studies, with an overall reduction calculated at 20 percent. In contrast, the observational studies that ACS highlighted report a higher relative survival benefit, ranging from 25 to 54 percent.

Is the survival benefit of screening mammography 0 or 50 percent? Such a wide disparity in results should demand scrutiny of results and methods. What’s a good study compared to a pretty lousy one? Which experts draw on the best evidence?

A close reading of the ACS reports suggest bias toward the types of studies (observational in this case) that claim the highest survival benefits while the best designed RCTs – the types of large studies typically understood to be the gold standard for public policy – are minimized. For instance, the analysis buries two of the most useful studies done to date on screening a population of women with mammograms, one Canadian National Breast Screening Study (CNBSS 1) and the UK Age Trial that specifically evaluated women in their 40s. These RCTs were very large (hundreds of thousands of women) and sufficiently designed (e.g., adequate randomization, comparable groups, accurate cause-of-death records. They both found no statistically significant survival benefit to adding mammograms to quality care.

Such zeroes get lost when averaged together with RCTs showing higher numbers. The Cochrane Collaboration’s systematic analyses of screening trials from 2000 through 2013 for women in their 40s and older argue soundly that the RCTs showing any survival benefits for screening were inadequately randomized and thus of poor quality (i.e., study groups were not comparable, records were questionable). Yet the ACS accepted the findings of these studies and labeled the RCTs showing no benefit as outliers or anomalies.

Far more problematic than ignoring the quality of different RCTs is ACS’s prioritizing of observational studies as if equal to or better than clinical trial evidence. Observational studies, useful for a variety of purposes, do not have true control groups for comparing survival benefits and have less validity for inferring causation. As a result of these methodological limitations, researchers must estimate group comparisons, and they do so using a wide variety of methods. What’s more, ACS chooses to trust the observational studies that support screening benefits while overlooking those that do not.

Disagreements about the benefits and harms of screening mammography have persisted among experts publicly since 2000, and they need far more airing than the ACS provides. Advocates can and should use the ACS’s roiling of muddy waters to increase awareness of mammogram screening’s limitations and perhaps the usefulness of other screening tools such as low-tech CBEs accompanied with good quality health care.

1 ACS co-created NBCAM in 1984 with the National Cancer Institute, the American Society for Radiology, the Imperial Chemical Industries, and Zeneca, the ICI subsidiary that makes anti-cancer drugs.

2 The independent review panel included 4 clinicians, 2 biostatisticians, 2 epidemiologists, 1 economist, and 2 patient representatives (one a biostatistician and the other a retired physician).

3 Average-risk is defined loosely as lacking: “a personal history of breast cancer, a family history of breast cancer, a genetic mutation known to increase risk of breast cancer (such as BRCA), and women who had radiation therapy to the chest before the age of 30.”

4 Editorial in JAMA by Nancy Keating MD MPH and Lydia Pace MD MPH raises other issues glossed over by ACS’s new guidelines, such as important differences in accuracy between observational studies and RCTs, ways that relative risk benefits obscure far lower absolute numbers, and how differing outcomes such as recurrence, breast cancer deaths, and all-cause deaths tell different stories.

Be Sociable, Share!

Articles & Posts