Although some critics continue to question the value of breast screening, the scientific evidence supporting it is rock solid.
|Jump to:||Choose article section... Seeing the issues more clearly Myth # 1: There is no benefit from screening before age 50|
Although some critics continue to question the value of breast screening, the scientific evidence supporting it is rock solid.
Despite more than 40 years of investigation, there are still those who doubt the value of screening mammography for breast cancer. Contrary to arguments from opponents, however, data from several reliable sourcesincluding randomized, controlled trialsclearly show that screening mammography is worthwhile.1-4 I believe that most of the opposition is due to a failure of many critics to understand the facts about screening mammography, as well as a failure of some to adhere to scientific principles when examining the data.
Of course, there's also a legitimate concern that women and their physicians should be informed of all aspects of breast cancer screening, including the negative effects. Since breast cancer screening involves healthy women, any negative consequence represents a "harm" that would not have been incurred had the patient not been screened. That's true of all screening tests. The potential harm includes the anxiety, discomfort, and inconvenience of having a mammogram. Positive test results that require additional evaluation generate even more anxiety. Many false-positive mammograms are easily resolved with additional mammographic views or ultrasound, but some questions can only be resolved by traumatic interventions. A number of women undergo biopsies for what prove to be benign reasons in order to find relatively few cancers. If the biopsy is positive, some women will undergo toxic treatment that can cause complications, and rarely death, without knowing whether the treatment has been beneficial.
Clearly, if screening didn't save lives, the test would cause nothing but harm. Since a screening test may make healthy individuals "ill," it's essential that strong scientific proof of benefit exist before recommending it. In fact, screening mammography has undergone more testing and more stringent evaluation than any other screening test we currently use. Although far from perfect, it does decrease the death rate from breast cancer.4-6 Not only has this been proven by RCTs, it's been confirmed in populations that have had access to routine screening. Before mammography came on the scene, the death rate from breast cancer had remained unchanged for over 50 years. With the introduction of widespread screening, the death rate has fallen steadily since 1990.
Several controversies surrounding screening mammography have emerged in recent years. In the 1990s the debate centered around whether or not women between 40 and 49 could benefit from screening.7 In 2000 and again in 2001, two members of the Cochrane Collaboration questioned whether or not screening was efficacious for women at any age.8,9 Despite the fact that all of the issues that have been raised have been addressed through scientific analysis, some have ignored the science and continue to suggest that we still can't be certain about the value of screening.10,11
When there's a disagreement over the interpretation of information, rather than providing a detailed and scientific analysis of the conflicting points of view, too many analysts dispense with the issues altogether by stating that experts can look at the same information and come to different conclusions. "We agree to disagree" has replaced systematic review. Furthermore, in the area of breast cancer screening, the peer review process has failed to pick up faulty science, and has permitted unsubstantiated conclusions and flawed analyses to gain credibility through publication.12-14
Unfortunately, certain journalsparticularly those that are monitored by the mediaseem to harbor undeclared biases.14 One well-known journal has even gone so far as to suggest that it feels no obligation to provide balanced information on both sides of controversial topics.15 This makes it difficult for clinicians, who assume that the general medical journals are unbiased, to determine the truth of a matter. A great deal of misinformation has been promulgated during the course of the mammography debate.16 The goal of this review is to shed light on some of these controversies.
Some analyses are scientifically valid and others are not. Unfortunately, in the screening debates both have been given equal weight. A prime example of this is the debate over the efficacy of screening women aged 40 to 49. Opponents of screening younger women have never explained that comparing women 49 and younger to those 50 and over was an arbitrary decision that's not supported by good science. The age of 50 has no biological significance. It was initially chosen as a surrogate for menopause to see if the results from the RCTs on screening were different in premenopausal versus postmenopausal women.17 In fact, opponents of screening have neglected to point out that none of the screening parameters change abruptly at 50or any other age for that matteryet they have continued to use this age as if it had some real significance.18
There are no accurate data that link age 50, or menopause, directly to any changes in the sensitivity or specificity of mammography, or its benefit. Every result that has been linked to age 50 has been the product of artificially grouping women 49 and younger as if they are a uniform group, and comparing them to women aged 50 and over as if they, too, are a uniform group. For example, since the incidence of breast cancer increases with age, it should come as no surprise that the detection of breast cancers also increases with age. Some analysts have made the detection rate appear to change suddenly at the age of 50 by dichotomously grouping women and comparing detection rates above and below 50. This makes it appear that detection suddenly changes at age 50 when there is no abrupt change at any age.19
The fact is that all of the arguments that have been used against screening for women under the age of 50 could be made for any age simply by dividing women into two groups and comparing women above and below that age. Taking a variable that changes steadily with increasing age and comparing it by grouping everyone above and below an arbitrarily chosen age will make the variable appear to change suddenly at that age. For example, if we wanted to determine the age at which hair turns gray and we compared two groups that encompassed everyone age 45 and younger to everyone 46 and older, it would appear that hair suddenly turns gray at age 45. This type of grouping and dichotomous analysis was used to make it appear that the age of 50 was an important age.19
I believe that evaluating data as if the age of 50 had some real significance was initially an innocent mistake, but it now appears that health planners use it to limit the costs of screening. My experience suggests that some countries abruptly curtail discussions of screening women under the age of 50 because national health plans do not want to shoulder the added economic burden of screening these women. Despite the overwhelming data that support a screening benefit, many countries continue to hide behind the observation that, since there is not uniform agreement on a benefit, they don't have to support screening of younger women.
Those who oppose screening for women under the age of 50 have either ignored the facts, as happened with the Consensus Development Panel in 1997, or misinterpreted and even manipulated the data to suggest there is no benefit.19,20 If women under the age of 50 are told that screening can save lives, but it is not being offered because it's too expensive, then the issue would be open for debate. However, if they are told that screening does not work based on the scientific studies, then discussion is precluded. I believe that's the reason why many European countries as well as Australia, Canada, and New Zealand avoid promoting screening before the age of 50.
I suspect that most physicians are not aware of the fact that RCTs of screening mammography were not designed to permit the retrospective evaluation of women aged 40 to 49 as a separate subgroup. Even the National Breast Screening Study of Canada (NBSS), which was touted as having been designed to evaluate these women, was underpowered and not executed properly.21 The other trials were never intended to permit analysis of women under age 50 separately and did not include enough women to permit these analyses.13 Not only was the time between screens in most of the trials probably too long for younger women, who tend to have faster growing cancers, but experts warn against making medical recommendations based on retrospective subgroup analysis using data that lack statistical power.22
Experts on clinical trial analysis caution that underpowered analysis can be very misleading.23,24 Were statistical power not important, then trials would only need two womenone screen and one control. Obviously in order to show a benefit, trials have to enroll enough patients to show that if there is a benefit, the difference in deaths between the screened women and the controls reaches statistical significance. Since the trials were not designed to analyze women aged 40 to 49 as a separate subgroup, they lacked the statistical power individually or even when combined to permit accurate analysis of women aged 40 to 49 as a separate group in the early years of follow-up.13
The arguments against screening younger women have been based on scientifically unsupportable, retrospective subgroup analyses. Although the trials demonstrated a benefit for women who began screening at age 40, opponents of screening advised women in their 40s that screening was of no benefit because the reduction in deaths among these women, when analyzed separately, did not reach statistical significance.7,25 What they failed to tell women was that it was mathematically impossible for the trials, in the early years of follow-up, to show a statistically significant decrease in deaths at the expected level of 25%. By violating a basic rule of trial data analysis, opponents of screening advised women using scientifically invalid analyses.
Opponents to this day have never provided scientific justification for advising women using data from unplanned subgroup analysis that lacked any statistical power, yet the analysis has been promulgated around the world and is the basis for the persistent disagreement. Because this misinformation was so widely disseminated and repeated, even now when the benefit for younger women has become statistically significant with longer follow-up, opponents ignore the scientific evidence and continue to say that the question remains unresolved.26
One of the more recent examples of poor data analysis is a review by two members of the Cochrane Collaboration who reanalyzed the results from the RCTs on mammographic screening and concluded that there was no benefit from screening for women at any age.8,9 They arrived at this conclusion by deciding that they did not like the way five out of the seven RCTs of screening were performed. Not only were their criticisms of the five trials unjustified, but discarding results from trials that they did not like violates one of the basic rules of trial analysis.
Compounding their misinterpretation was the fact that they claimed that the remaining two trials (which met their approval) showed no benefit when, in fact, they used outdated information from a clinical trial that actually shows a benefit.27 Unfortunately, this misguided review received a great deal of attention in the New York Times and was followed by a succession of biased reports and editorials in the newspaper.11 The report stimulated a number of large-scale re-reviews of the screening trials (The US Preventive Services Task Force review was already underway).28-32 These reviews all disagreed with the Cochrane Review analysis and concluded that there is a benefit from mammographic screening. The reviews concluded that the concerns raised by the Cochrane reviewers were either due to their misunderstanding of screening trials, or were inconsequential.
RCTs are the most convincing way to show a benefit from a test such as mammographic screening. However, once it has been shown in scientific studies that a test can save lives, the final proof comes when the test is introduced into general health care. A similar decrease in deaths in the general population once the new test is introduced confirms the test's benefits. This has now been shown in four studies and suggested by national health statistics in the US.3,33-36 When large-scale mammographic screening was introduced into the general population, the death rate from breast cancer dropped. Since the death rate is essentially the number of deaths from breast cancer divided by the number of women (or women years) in the population, the death rate is not influenced by cancer detection rates so that nonlethal cancers have no influence.
Although it's impossible to directly measure the percent reduction in deaths that can be attributed to screening versus better health care and treatment, the data suggest that at least two thirds of the decrease in the breast cancer death rate can be attributed to screening mammography. In a large review of the breast cancer death rate in Sweden, a 44% decrease was seen when the death rate among women in the era before screening was available was compared to the death rate among women after screening was made available to all women aged 40 to 69. The investigators found a 16% drop in breast cancer deaths among women who were too young to be screened or who refused screening. One can attribute this reduction to better health among Swedish women, and better cancer treatment.
This implies that there was a 28% decrease in deaths due to screening (which closely agrees with the data from the RCTs).34 Thus, not only has the mortality rate from breast cancer been decreased in the screening RCTs, but screening decreases the death rate among women in the general population.
Another myth that has developed over the years is the belief that early detection of breast cancer, for some women, may result in their earlier demise.37 This was based on the observation that in the early years of follow-up of several of the RCTs of screening mammography, there were more deaths among the screened women than among the control women. Some interpreted this to mean that early detection might be detrimental. Some investigators looked to the animal literature, and convinced themselves that the increased growth rate of some metastatic lesions that had been observed in laboratory animals might occur in humans, and that the earlier detection and removal of a primary breast cancer resulted in faster growth of metastatic lesions and earlier death. There are, indeed, some animal data that show that metastatic lesions grow more rapidly for a period of time when the primary tumor is removed.38 Complex explanations have been used to explain why finding a cancer earlier might lead to an earlier death. Unfortunately, this is an example of using incomplete information to develop a false theory.
There are many reasons why this interpretation of the data is wrong. To begin with, only a few of the trials showed an excess of cancer deaths among screened women in the early years of follow-up. In one of the most recent papers, describing the long-term follow-up from the Gothenburg Trial, the mortality reduction ranged from 22% for women aged 39 to 44 to 49% for women aged 45 to 49. The survival curves begin to open virtually immediately and there is no evidence of an excess of cancer deaths among the screened women at any time.39
An RCT is designed to include identical groups. If there are sufficient numbers of women who are randomly assigned to a screened group and a control group that was left alone, both groups would have the same number of women develop breast cancer and the same number of women would die from breast cancer each year as the groups were followed over time. If the only difference between two randomly developed groups is that one group is screened, and there are fewer breast cancer deaths over time in that group than in the unscreened control group, and if the numbers are large enough that the difference reaches statistical significance, then one can conclude that screening reduced the rate of death.
However, although properly designed and executed RCTs develop two "identical" groups, the women do not behave precisely the same in both groups. The groups can never be so identical that women will die at precisely the same time in both groups. This is why statistical analysis is required. A flipped coin may come up heads five times in a row, but as the number of flips increases, the true 50/50 probability becomes increasingly apparent. Similarly, in the early years of follow-up of a screening trial, the number of deaths is so small that there is enormous statistical fluctuation, with chance playing the greatest role. This is clearly seen in the graph of the Malmo Trial deaths where there are a few more deaths in one group versus the other that fluctuates back and forth until the numbers become large enough for the pattern to stabilize.40 Chance plays a large role in biological systems and that's why large trials are needed with long-term follow-up to determine a true pattern. Analyzing data prematurely can be very misleading. It is unlikely that the excess of deaths that was seen in the early years of follow-up was anything more than statistical fluctuation.
In addition to the fact that early analysis can be grossly misleading, those who have suggested that screening can lead to premature deaths ignore the fact that there are no published data (with the exception of the flawed National Breast Screening Study of Canada) that indicate which women actually accounted for these early deaths. What many do not realize is that every trial except the Canadian trials involved women who were "invited" to be screened. They were chosen through a blinded, random allocation process before anything was known about the women. Once a woman was allocated to one group or the other she was counted with that group even if she did not comply with the allocation. Thus, if a woman who had been allocated to the screened group died from breast cancer but had refused to be screened, her death was still counted with the screened group. If she was allocated to the control group and her life was saved by a mammogram that she obtained outside of the trial, she was still counted with the unscreened control group. Those who suggest that the data show an adverse consequence of screening actually have no way of knowing whether the few excess deaths in the early years of follow-up in the screened group were among women who had actually been screened, or among those who had refused to be screened.
The final mistake in the argument is found in the animal data. Although it appears that some metastatic lesions grow more rapidly once the primary tumor has been excised, this is a transient phenomenon that lasts for only a few hours to a few days. There are no data that show that it leads to earlier death.41
Unfortunately, a great deal of misinformation has been widely disseminated on the subject of breast cancer screening. In my opinion, this misinformation has been perpetuated by journals that have undeclared publication biases and poor peer review. The media have also played a role by reporting on studies without understanding the scientific issues.
When asked why the New York Times had not covered the latest data from Sweden showing a decrease in deaths in the screening era, the paper said it did not believe the results because screening is more likely to detect nonlethal cancers and this would dilute the death rate (e-mail from Cornelia Dean, Science Editor of the New York Times, to Richard Moore, Director of Research in the Breast Imaging Division of the Department of Radiology at the Massachusetts General Hospital, May 2003). What she clearly did not understand is that the death rate is independent of what is detected. Nonlethal cancers have no effect on the death rate. It is of some concern that those who do not understand the basic scientific issues are deciding what information is given to the public.
Screening mammography is far from perfect. It does not find all cancers and does not find all cancers early enough to permit a cure, but it offers major benefits. Screening can save many livesit has already. No one involved in the screening effort believes that mammography solves the problem, but it's a major step forward in the battle against cancers that have resisted therapeutic interventions for over 50 years. Women aged 40 and over should be urged to attend annual screening while strong support needs to be given to efforts to find cures, discover safe methods of prevention, and develop additional ways to find cancers earlier.
1. Seidman H, Gelb SK, Silverberg E, et al. Survival experience in the breast cancer detection demonstration project. CA Cancer J Clin. 1987;37:258-290.
2. Linver MN, Paster SB. Mammography outcomes in a practice setting by age: prognostic factors, sensitivity, and positive biopsy rate. J Natl Cancer Inst Monogr. 1997;(22):113-117.
3. Tabar L, Vitak B, Chen HH, et al. Beyond randomized controlled trials: organized mammographic screening substantially reduces breast carcinoma mortality. Cancer. 2001;91:1724-1731.
4. Nystrom L, Andersson I, Bjurstam N, et al. Long-term effects of mammography screening: updated overview of the Swedish randomised trials. Lancet. 2002;359:909-919.
5. Shapiro S, Venet W, Strax P, et al. Ten- to fourteen-year effect of screening on breast cancer mortality. J Natl Cancer Inst. 1982;69:349-355.
6. Tabar L, Fagerberg CJ, Gad A, et al. Reduction in mortality from breast cancer after mass screening with mammography. Randomised trial from the Breast Cancer Screening Working Group of the Swedish National Board of Health and Welfare. Lancet. 1985;1:829-832.
7. Fletcher SW, Black W, Harris R, et al. Report of the International Workshop on Screening for Breast Cancer. J Natl Cancer Inst. 1993;85:1644-1656.
8. Gotzsche PC, Olsen O. Is screening for breast cancer with mammography justifiable? Lancet. 2000;355:129-134.
9. Olsen O, Gotzsche PC. Cochrane review on screening for breast cancer with mammography. Lancet. 2001;358:1340-1342.
10. Hendrick RE, Smith RA, Rutledge JH 3rd, et al. Benefit of screening mammography in women aged 40-49: a new meta-analysis of randomized controlled trials. J Natl Cancer Inst Monogr.1997;(22):87-92.
11. Kopans DB. The most recent breast cancer screening controversy about whether mammographic screening benefits women at any age: nonsense and nonscience. AJR Am J Roentgenol. 2003;180:21-26.
12. Welch HG, Black WC. Using autopsy series to estimate the disease "reservoir" for ductal carcinoma in situ of the breast: how much more breast cancer can we find? Ann Intern Med. 1997;127:1023-1028.
13. Kopans DB, Halpern E, Hulka CA. Statistical power in breast cancer screening trials and mortality reduction among women aged 40-49 years of age with particular emphasis on the National Breast Screening Study of Canada. Cancer. 1994;74:1196-1203.
14. Kopans DB. The breast cancer screening controversy: lessons to be learned. J Surg Oncol. 1998;67:143-150.
15. Kassirer JP, Angell M. Controversial Journal editorials. N Engl J Med. 1997;337:1460-1461.
16. Davis DL, Love SM. Mammographic screening. JAMA. 1994;271:152-153.
17. Shapiro S, Venet W, Strax P, et al. Periodic Screening for Breast Cancer: The Health Insurance Plan Project and its Sequelae, 1963-1986. Baltimore, Md: The Johns Hopkins University Press; 1988.
18. Kopans DB, Moore RH, McCarthy KA, et al. Biasing the interpretation of mammography screening data by age grouping: nothing changes abruptly at age 50. The Breast Journal. 1998;4:139-145.
19. Kerlikowske K, Grady D, Barclay J, et al. Positive predictive value of screening mammography by age and family history of breast cancer. JAMA. 1993;270:2444-2450.
20. Kopans DB. The breast cancer screening controversy and the National Institutes of Health Consensus Development Conference on Breast Cancer Screening for Women Ages 40-49. Radiology. 1999;210:4-9.
21. Kopans DB, Feig SA. The Canadian National Breast Screening Study: a critical review. AJR Am J Roentgenol. 1993;161:755-760.
22. McNamee D. Consensus reached at new European breast cancer meeting. Lancet. 1998;352:1200.
23. Lachin JM. Introduction to sample size determination and power analysis for clinical trials. Control Clin Trials. 1981;2:93-113.
24. Moher D, Dulberg CS, Wells GA. Statistical power, sample size, and their reporting in randomized controlled trials. JAMA. 1994;272:122-124.
25. Shapiro S. Screening: assessment of current studies. Cancer. 1994;74:231-238.
26. Hendrick RE, Smith RA, Rutledge JH rd, et al. Benefit of screening mammography in women aged 40-49: a new meta-analysis of randomized controlled trials. J Natl Cancer Inst Monogr. 1997;(22):87-92.
27. Miettinen OS, Henschke CI, Pasmantier MW, et al. Mammographic screening: no reliable supporting evidence? Lancet. 2002;359:404-405.
28. U.S. Preventive Services Task Force. Screening for breast cancer: recommendations and rationale. Ann Intern Med. 2002;137:344-346.
29. International Agency for Research On Cancer (IARC). Available at: http://www.iarc.fr/pageroot/top1.html. Accessed August 11, 2003.
30. Duffy SW, Tabar L, Smith RA. The mammographic screening trials: commentary on the recent work by Olsen and Gotzsche. CA Cancer J Clin. 2002;52:68-71.
31. Health Council of the Netherlands. The benefit of population screening for breast cancer with mammography. The Hague. Health Council of the Netherlands, 2002, Publication No. 2002/03E.
32. European Institute of Oncology, Global Summit on Mammographic Screening. Milan, Italy, June 3-5, 2002
33. Duffy SW, Tabar L, Chen HH, et al. The impact of organized mammography service screening on breast carcinoma mortality in seven Swedish counties. Cancer. 2002;95:458-469.
34. Tabar L, Yen MF, Vitak B, et al. Mammography service screening and mortality in breast cancer patients: 20-year follow-up before and after introduction of screening. Lancet. 2003;361:1405-1410.
35. Otto SJ, Fracheboud J, Looman CW, et al. Initiation of population-based mammography screening in Dutch municipalities and effect on breast-cancer mortality: a systematic review. Lancet. 2003;361:1411-1417.
36. Kopans DB. Beyond randomized controlled trials: organized mammographic screening substantially reduces breast carcinoma mortality. Cancer. 2002;94:580-581.
37. Retsky M, Demicheli R, Hrushesky W. Breast cancer screening: controversies and future directions. Curr Opin Obstet Gynecol. 2003;1:1-8.
38. Van Dierendonck JH, Keijzer R, Cornelisse CJ, et al. Surgically induced cytokinetic responses in experimental rat mammary tumor models. Cancer. 1991;68:759-767.
39. Bjurstam N, Bjorneld L, Warwick J, et al. The Gothenburg Breast Screening Trial. Cancer. 2003;97:2387-2396.
40. Andersson I, Janzon L. Reduced breast cancer mortality in women under age 50: updated results from the Malmo Mammographic Screening Program. Natl Cancer Inst Monogr. 1997;(22):63-67.
41. Gunduz N, Fisher B, Saffer EA. Effect of surgical removal on the growth and kinetics of residual tumor. Cancer Res. 1979;39:3861-3865.
Dan Kopans. Three common myths about mammography.
Aug. 1, 2004;49:49-58.