Sorting Through the Arguments on Breast Screening FREE ONLINE FIRST

Michael G. Marmot, MBBS, MPH, PhD, FRCP, FFPHM

[+] Author Affiliations

JAMA. 2013;():1-2. doi:10.1001/jama.2013.6822.

Text Size: A A A

Published online May 30, 2013

Article

References

Views on the benefits and harms of breast cancer screening are sharply polarized and increasingly vocal. Allegations of harming women are flung in both directions. The antiscreeners claim that benefit is minimal and overdiagnosis is so frequent that women are being subject to unnecessary interventions and treatment. The proscreeners claim that if the critics win the day, women will be deprived of the benefits that screening brings of early diagnosis and reductions in mortality from breast cancer.

To those coming fresh to the argument, such disagreement seems surprising. Given the large body of evidence evaluating breast screening, it seems that the evidence would settle the issue. Such naiveté does not allow for the fact that people interpret evidence and, indeed, influence its generation. Judgments often reflect more about starting assumptions than they do about the nature of the evidence.

Into this somewhat raucous debate came the Independent UK Panel on Breast Screening.¹ Given that views about the benefits and hazards of breast screening are so polarized that one only has to look at the author of an article to make a reasonable guess whether the it will be pro or con, members of the breast screening panel were selected only if they had no prior publications on breast screening. It was hoped, therefore, that the only prior experiences shared by panel members were a desire to reach judgments based on sound science and to improve women's health. Characteristically, the United Kingdom has a more cautious screening program than the United States. Breast cancer screening in the United Kingdom is recommended for women aged 50 through 70 years, every 3 years—the focus of the UK panel—rather than annual screening starting at age 40 years that is usual in the United States.

The purpose of screening is to advance the timing of diagnosis and thereby improve prognosis. Even were screening not to alter the natural history of breast cancer, screening would lead to an apparent prolongation of the time between diagnosis and death—lead time. Therefore, the appropriate end point to evaluate the benefits of screening is mortality from breast cancer.

In the panel's judgment, the best evidence for the relative benefit of screening on mortality reduction is based on 11 randomized controlled trials (RCTs) of breast screening. Meta-analysis of these trials with 13 years of follow-up estimated a 20% reduction in breast cancer mortality among women invited for screening.¹ The panel based its meta-analysis on data reported in a Cochrane review.² The Cochrane review came to a similar estimate of relative reduction but, because of concerns about quality of some of the trials, reduced that estimate to 15%. The UK panel did not follow this procedure.

Some have claimed that being in the screening group of a trial could be harmful. The correct end point, therefore, should be not breast cancer mortality but all-cause mortality. The power of the trials simply does not allow that: a 20% reduction in breast cancer mortality would translate into a 3% reduction in all cancer mortality and a 1.6% reduction in all-cause mortality.¹

There has been a remarkable range in estimates of absolute benefit of screening, varying from 1 breast cancer death avoided for 2000 women invited to screening to 1 breast cancer death avoided for about 100 women screened, about a 20-fold difference. Major determinants of that large variation are the age of women screened and the durations of screening and follow-up. The panel applied the relative mortality reduction of 20% to achieve the observed cumulative absolute risk of breast cancer mortality over the ages of 55 through 79 years for women in the United Kingdom—assuming that women who began screening at 50 years would gain no benefit in the first 5 years but assuming that the mortality reduction would continue for 10 years after screening ended. This yielded the estimate that for every 235 women invited to screening, 1 breast cancer death would be prevented; correspondingly for 180 actually screened, 1 breast cancer death would be prevented.¹

This seems relatively straightforward—randomized controlled trials show benefit. Yet questions have been raised about the internal and external validity of the trials. One argument against relying on the trials is that they were conducted 20 to 30 years ago and are therefore out of date. However, this criticism applies to all trials—their findings apply to the past. It is a judgment call whether conditions in the future may be sufficiently different that what has been learned from trials no longer applies.

For this reason, some observers have turned to ecological studies comparing geographical areas, or time periods, when screening programs were and were not in place. These have generated diverse findings, partly because of the major advances in treatment of breast cancer, which have a demonstrably larger influence on mortality trends than does screening, and partly because of the difficulty of excluding imbalances in other factors that could affect breast cancer mortality. The panel did not consider these studies helpful in estimating the effect of screening on mortality. The panel noted that case-control studies provide a more contemporary estimate of benefit associated with screening and, in general, show screening to be associated with greater benefit than is shown by the trials. But because of concerns over confounding, the panel did not rely on the quantitative estimates from the case-control studies.

The major harm of breast screening considered by many is that of overdiagnosis—cancers detected by screening that would not have come to attention in the woman's life in the absence of screening. This area perhaps even more than the estimate of benefit provides ample evidence of the polarization of views. The central issue is that if screening is effective it will lead to increased incidence in women who are screened compared with those who are not. The challenge is in distinguishing this “true” increase from overdiagnosis. Various techniques have been used. Two recent reports that used time trends to estimate overdiagnosis of breast cancer provide evidence for the polarization of views. A US study found overdiagnosis of around 30%,³ whereas a European study estimates overdiagnosis at 7%.⁴ In each case, the findings fit with previous publications by these authors. The differences have much to do with the assumptions made, the age range, and the denominator used to define percent of overdiagnosis. A modeling exercise, conducted for the panel, showed that varying the assumptions in ecological studies of trends could yield estimates of overdiagnosis varying from the trivial to the frankly alarming. There being no strong reason to favor one set of assumptions over another, the panel concluded that this was not a helpful way to estimate overdiagnosis.

Given the definition of overdiagnosis, the best estimate should come from following up screened and unscreened women, who are otherwise comparable, to the ends of their lives and determining breast cancer incidence in the 2 groups. Were there no overdiagnosis the incidence rate in the screened group should approach that in the unscreened. Such evidence does not exist. For that reason, the panel relied on follow-up of 3 trials in which women in the control group were not offered screening at the end of the trial. Followed up for 6 to 15 years after the end of screening, these studies suggested an overdiagnosis rate of 11%. During the screening period this amounts to 19%. In other words, about 11% of breast cancer cases detected among screened women, followed up for 6 to 15 years after screening ended, could be overdiagnosis. During the screening period for 10 000 women screened, there would be 129 overdiagnosed cases. The panel emphasized uncertainty around this estimate, not only because of the small number of cases but because the studies depart from the ideal; in particular women are not followed up to the end of their lives.

For the panel, the bottom line was 2-fold: screening extends lives and should continue, and real effort should focus on the complicated task of conveying the information on risks and benefits to women so that they can be informed partners in the decision to screen or not.

It is a worthy aim to have guidelines by which to assess the merits, or otherwise, of screening.⁵ The UK panel's procedure did not differ greatly from these—with the single exception of feeling unable to gauge numerically the strength of the evidence. Reactions to the panel's report suggest that those who had hitherto been confused by the divergence of views found the report helpful—as close as could be to a dispassionate view of the evidence by an independent panel. It is doubtful, however, that the independent panel changed the minds of the principal proscreening and antiscreening groups in the debate over screening.⁶ Positions are too entrenched. But the evidence on breast screening is more extensive than in many other areas relevant to population health. If this is not enough for an independent group, coming fresh to the debate, to reach a reasonable judgment, then evidence-based policy is a good deal more difficult than many would believe.

AUTHOR INFORMATION

Corresponding Author: Michael G. Marmot, MBBS, University College of London Institute of Health Equity, University College of London, Department of Epidemiology and Public Health, 1-19 Torrington Pl, London, England WC1E 7HB (m.marmot@ucl.ac.uk).

Published Online: May 30, 2013. doi:10.1001/jama.2013.6822

Conflict of Interest Disclosures: The author has completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Members of the Independent UK Panel on Breast Cancer Screening include Chair: Sir Michael G. Marmot, MBBS, MPH, PhD, FRCP, FFPHM, FMedSci, Departments of Epidemiology and Public Health and Institute of Health Equity, University College, London; Douglas Altman, DSc, Center for Statistics in Medicine, University of Oxford, England; David Cameron, MSc, MD, Department of Oncology, Edinburgh University, and Edinburgh Cancer Research Centre, Edinburgh, Scotland; John Dewar, MD, Department of Clinical Oncology, Dundee University, Scotland; Simon Thompson, DSc, Department of Biostatistics, University of Cambridge, England; and Maggie Wilcox, patient advocate.

REFERENCES

Independent UK Panel on Breast Cancer Screening. The benefits and harms of breast cancer screening: an independent review. Lancet. 2012;380(9855):1778-1786
PubMed | Link to Article

Gøtzsche PC, Nielsen M. Screening for breast cancer with mammography. Cochrane Database Syst Rev. 2011;(1):CD001877
PubMed

Bleyer A, Welch HG. Effect of three decades of screening mammography on breast-cancer incidence. N Engl J Med. 2012;367(21):1998-2005
PubMed | Link to Article

Puliti D, Duffy SW, Miccinesi G, et al; EUROSCREEN Working Group. Overdiagnosis in mammographic screening for breast cancer in Europe: a literature review. J Med Screen. 2012;19(suppl 1) 42-56
PubMed | Link to Article

Ransohoff DF, Pignone M, Sox HC. How to decide whether a clinical practice guideline is trustworthy. JAMA. 2013;309(2):139-140
PubMed | Link to Article

Independent UK Panel on Breast Cancer Screening. The benefits and harms of breast cancer screening—authors' reply [letter]. Lancet. 2013;381(9869):803-804
PubMed