Modern mammography screening and breast cancer mortality: population study
BMJ 2014; 348 doi: https://doi.org/10.1136/bmj.g3701 (Published 17 June 2014) Cite this as: BMJ 2014;348:g3701
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Jørgensen and Gøtzsche seem to think that effectiveness of mammography screening is best assessed by ecologic studies, where the timing of implementing screening programmes in a population is related to breast cancer mortality, or better still, to total mortality. Their reason for suggesting total mortality as the appropriate outcome, is their claim that overtreatment of indolent cancers detected by screening might eventually kill a certain proportion of the patients. They seem to forget that breast cancer deaths only constitute a small proportion of all deaths, and that the separation of any screening effect from random variation in total mortality will have too low statistical power.
Further, the ecologic approach to mammography screening suffers from two major short-comings that limit their value. First, the crudeness of the timing of implementing screening cannot properly capture the distinction between screened and un-screened women. Second, the long survival of many breast cancer patients means that a screening effect will appear gradually at the population level, and be almost impossible to separate from other variations in long term mortality trends. The figure below illustrates this gradual effect, and indicates that any clear correlation between mammography screening and breast cancer mortality cannot be expected to appear in ecologic studies, even in the presence of a substantial screening effect.
In our study, the gradual implementation of mammography screening in the 19 counties in Norway, and the near random distribution of invitations within each county, provided a setting similar to that of a natural experiment, and created distinct and unconfounded contrasts between invited and not (yet) invited women. Using exact and individual information about invitation to screening, date of breast cancer diagnosis, and date of death, we compared breast cancer mortality according to invitation status, and made a clear distinction between women who were diagnosed before (without potential benefit from screening) and after invitation (with a potential benefit) to screening. We used invitations (”intention to screen”) instead of attendance to avoid selection bias in the analyses.
We found a mortality benefit attributed to screening that is a likely result of earlier diagnosis and more effective treatment. As mentioned in the paper, breast cancer care was reorganised in parallel with the national screening programme, in order to optimise clinical management. This factor has benefited all breast cancer patients, and not only participants in the screening programme. Nonetheless, better organisation of care may be an important mediator for the observed screening effect.
Jørgensen and Gøtzsche assume that breast cancer mortality trends are similar across age groups in the absence of screening. The incidence of breast cancer subtypes differs by age, and responses to treatment strongly differ between subtypes (1,2). To illustrate, using two age groups with little or no influence of screening; from 1995 to 2009, breast cancer mortality declined by 51% among Norwegian women younger than 50 years of age, and by 8% among women 80 years and older (3).
Stavros Saripanidis is concerned about bias that may arise due to selective screening attendance. We share his concern, and used invitations to screening in the analysis to eliminate selective attendance as a source of bias.
Ellen CG Grant questions whether the results could be influenced by different use of menopausal hormone treatment. To account for this potential confounding factor, as well as for changes in breast cancer treatment, we included a non-linear breast cancer mortality trend variable in the regression model. We also observed that the estimated screening effect persisted across age groups and over calendar time, indicating that confounding by variations in hormone treatment is not an important factor for our findings.
Alain Braillon misses information about over-diagnosis. There is little doubt that some cancers detected by mammography screening have limited malignant potential. However, the proportion of indolent cases is unknown, and a study of over-diagnosis and over-treatment was beyond the scope of our paper. However, such a study should also use individual data with exact information about screening status to provide reliable results.
Harald Weedon-Fekjaer (I, II, III)
Pål R. Romundstad (I)
Lars J. Vatten (I, IV)
I) Department of Public Health, Norwegian University of Science and Technology, Trondheim, Norway
II) Oslo Center for biostatistics and epidemiology, Department of Biostatistics, University of Oslo, Oslo, Norway
III) Oslo Center for biostatistics and epidemiology, Research Support Services, Oslo University Hospital, Oslo, Norway
IV) Harvard School of Public Health, Department of Epidemiology, Boston, USA
References
1. Li CI, Daling JR, Malone KE. Incidence of invasive breast cancer by hormone receptor status from 1992 to 1998. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. Jan 1 2003;21(1):28-34.
2. Early Breast Cancer Trialists' Collaborative G, Davies C, Godwin J, et al. Relevance of breast cancer hormone receptors and other factors to the efficacy of adjuvant tamoxifen: patient-level meta-analysis of randomised trials. Lancet. Aug 27 2011;378(9793):771-784.
3. NORDCAN; http://www.ancr.nu/nordcan.asp
Competing interests: No competing interests
In an Italian study on prevention of breast cancer there was no significant difference between tamoxifen treatment and placebo in the whole group. However, when high-risk women (taller than 160 cm, at least one functioning ovary, age 13 or less at menarche, no completed pregnancy before age 24) were separated out, the protective effect of tamoxifen was large and significant.[1] In a similar way, it is possible that breast screening has greater benefit for high risk women, while being less beneficial for low risk women. This could be very important. When each woman’s risk factors are taken into account, it is possible that screening frequency could be adjusted according to risk, as it currently is for cervical cancer screening. More research is needed to help women and their doctors make decisions about screening.
1. Veronesi U, Maisonneuve P, Rotmensz N, et al. Italian randomized trial among women with hysterectomy: tamoxifen and hormone-dependent breast cancer in high-risk women. J Natl Cancer Inst 2003;95(2):160-5
Competing interests: No competing interests
Some of the strongest proponents of mammography have accused several critics of the test of relying on numerous "terms of imprecision" - such as "assumptions" and "estimates" - in the analysis of some of their screening-unfavorable studies [1,2] and thus, the mammogram advocates claimed, such work "should not have been published in peer-reviewed journals" [2].
Yet this argument of "imprecision" hasn't been raised with this observational study by Weedon-Fekjær and colleagues [3], despite that it is also quite seriously affected by a number of "assumptions" and statistical "estimates". Instead, a perusal of the submitted commentaries here reveals, it is defended as a rather solid piece of scientific investigation.
This suggests arbitrary judgment: the value of a mammogram study loses or gains scientific credibility based on a preconceived screening ideology instead of an impartial integration and comparison of it against the total body of medical evidence on mammography.
References:
[1] Dean PB, Tabár L, "Mammography's alleged harms: Separating fact from fiction", auntminnie.com, 13-Dec-2012
[2] Tabár L, Dean PB, Chen TH, Yen AM, Chiu SY, Tot T, Smith RA, Duffy SW, "The Impact of Mammography Screening on the Diagnosis and Management of Early-Phase Breast Cancer", Ch. 2, Pp. 31-78, in 'Breast Cancer: An New Era of Management', Francescatti DS, Silverstein MJ. (Eds.), 2014
[3] Weedon-Fekjær H, Romundstad PR, Vatten LJ, "Modern mammography screening and breast cancer mortality:population study", BMJ 2014;348:g3701
Competing interests: Author of the (e)book "The Mammogram Myth" (2013)
Weedon-Fekjær et al. reported a 28% reduction in breast cancer mortality associated with an invitation to breast screening [1]. However, they also say that they “found no association of the time that mammography screening was implemented with breast cancer mortality.” This means that there were the same reductions in breast cancer mortality rates in screened and non-screened areas of Norway. The reduction was 23% during 1989 to 2006 for women aged 50-69 years, and 35% for women <50 years who were never invited to screening [2]. This shows that the decline in mortality is mainly or fully explained by adjuvant therapy. Accordingly, a study that compared three pairs of very similar neighbouring countries, including Norway and Sweden, that had introduced screening 10–15 years apart, found no relation between start of screening and the reduction in breast cancer mortality (Figure 1)[3,4]. The fall in breast cancer mortality started around 1990 in all countries, which was when effective treatments were introduced, and the rate of decline was about the same in all countries. When results of observational studies seem too good to be true, we should look for other explanations.
Better organisation of care, which was systematically introduced alongside breast screening, also plays a part. Kalager et al. found a non-significant reduction of 10% in Norwegian women that had been invited to screening and a non-significant 8% reduction in older women who were not invited to screening in the same areas, but had access to the breast cancer centres [5], something which Weedon-Fekjær and colleagus “cannot exclude” [1] as a limitation of their study. Kalager et al. also found that the incidence rate for stage III or IV cancer had declined by 24% both in screened and non-screened areas in Norway [6], which suggests that screening cannot have had any effect.
Denmark is similar to Norway in many respects, but despite less screening, we have had a 26% reduction in breast cancer mortality in the same time period and age group as in Norway [2]. For 17 years, till late 2007, screening was offered to only 20% of the Danish population. Despite this, a study by Olsen et al. reported a 25% reduction in breast cancer mortality that they attributed to screening [7] whereas we were unable to find any effect when we compared breast cancer mortality rates in screened and non-screened areas [8]. Again, the raw data presented in a graph [8] said a lot more than the results of complicated statistical models [1,8].
While a screening effect on population mortality rates is initially “diluted” by deaths from breast cancers detected before screening was introduced, this cannot explain the lack of a visible effect after 17 years. By then, practically all breast cancer deaths would be from cases detected within the screening era. A graph should show two increasingly separating lines if screening had any worthwhile effect, but this is not the case [8]. The Forrest report from 1986 [9], which laid the foundation for breast screening in the UK, noted:
”Monitoring the effectiveness of screening. This can be done approximately by examining trends in age-specific breast cancer mortality available from routine statistics.” This is what we did [8].
Whatever the effect of screening on breast cancer mortality, it is the wrong outcome. Treatment of healthy overdiagnosed women increases their mortality, which means that total mortality is the most meaningful outcome. Further, the psychological consequences of false positive findings are substantial and affect a large proportion of invited women. It is clear that a utility analysis that takes account of the harms is likely to come out negative [10,11].
References
1. Weedon-Fekjær H, Romundstad PR, Vatten LJ. Modern mammography screening and breast cancer mortality:population study. BMJ 2014;348:g3701
2. Autier P, Boniol M, LaVecchia C, Vatten L, Gavin A, Héry C, et al. Disparities in breast cancer mortality trends between 30 European countries: retrospective trend analysis of WHO mortality database. BMJ 2010;341:c3620.
3. Autier P, Boniol M, Gavin A, Vatten LJ. Breast cancer mortality in neighbouring European countries with different levels of screening but similar access to treatment: trend analysis of WHO mortality database. BMJ 2011;343:d4411 doi: 10.1136/bmj.d4411.
4. Bleyer A. US breast cancer mortality is consistent with European data. BMJ 2011;343:d5630
5. Kalager M, Zelen M, Langmark F, Adami HO. Effect of Screening Mammography on Breast-Cancer Mortality in Norway. N Engl J Med 2010;363:1203-10.
6. Kalager M, Adami HO, Bretthauer M, Tamimi RM. Overdiagnosis of invasive breast cancer due to mammography screening: results from the norwegian screening program. Ann Intern Med 2012;156:491-9.
7. Olsen AH, Njor SH, Vejborg I, Schwartz W, Dalgaard P, Jensen MB et al. Breast cancer mortality in Copenhagen after introduction of mammography screening: Cohort study. BMJ 2005;330:220-4.
8. Jørgensen KJ, Zahl PH, Gøtzsche PC. Breast cancer mortality in organised mammography screening in Denmark. A comparative study. BMJ 2010;340:c1241
9. Forrest P. Breast Cancer Screening. Report to the Health Ministers of England, Wales, Scotland & Northern Ireland. London: Department of Health and Social Science, 1986.
10. Biller-Andorno N, Jüni P. Abolishing mammography screening programs? A view from the Swiss Medical Board. N Engl J Med 2014;370:1965-7.
11. Swiss Medical Board. Systematisches Mammographie-Screening. February 2014. Available at: http://www.medical-board.ch/fileadmin/docs/public/mb/Fachberichte/2013-1...
Competing interests: No competing interests
The ongoing debate about screening for breast cancer versus no screening, is like fishing for wild salmon in open rivers as opposed to fishing for farm yard salmon in protected home grown ponds. You get less wild ones because there are less per unit volume of water. You get loads of farmed ones because they are in a closed system. Breast cancer these days is like the farmed salmon. They are home grown and there are loads of them. Why say this? Because the databases of all reputable cancer site show very high levels of breast cancer which is increasing in invasiveness and spread especially amongst the young. They are home grown because they are iatrogenically grown. The accelerant is female sex hormones as oral contraceptives and HRT. This response is more a reaction to the silence surrounding the cause of the rising incidence of invasive breast cancer than about screening per se. But what do you do if you don't screen? Wait for it to present to late?. The attached article pulls no punches!
Competing interests: No competing interests
Comparing the mortality benefit of screening with the harms of screening through over-screening, such as psychological distress and unnecessary biopsies, is not at par because it compares the worst outcome for those diagnosed with breast cancer to comparatively minor outcomes in those falsely diagnosed with breast cancer. Morbidity parameters have to be added to the outcome parameters in women with breast cancer such as 1) stage of diagnosis 2) number of treatment options available (assuming that having the choice between treatment modalities is by itself positive)3) treatment side effects 4) need for long-term follow-up 5) psychological well being of patients after diagnosis of breast cancer. If those are not added to breast cancer outcome, it seems not valid to take into considerations harms of screening. Of note, on an individual patient base, there is currently almost no reliable way to tell whether and when an early breast cancer (including stage 0) would progress. The financial aspect should be discussed separately and compared to the costs of other clinical measures that reduce a bad outcome such as treating hypertension, Type 2 DM, hyperlipidemia which are mostly preventive measures.
With the current state of knowledge, the complexity of cancer screening, be it for breast, prostate, lung, anal or liver cancer, seems to be in need of a well informed, open minded and individualized approach respecting patient autonomy to the fullest.
Competing interests: No competing interests
Women value more than just mortality reduction. Early detection of smaller tumours may reduce the need for mastectomy, axillary lymph node dissection, and chemotherapy. Even when these aspects of treatment do not affect mortality statistics, they do affect women’s quality of life and the cost of their care.
We need more data on who benefits most from breast screening. Women who have more or stronger risk factors for breast cancer, including family history, early menarche, higher age at first birth, taller than average, HRT users,[1] and those with moderately dense or very dense breasts on mammography,[2] may benefit more from screening than those with less or no risk factors. They may need to be screened more frequently than women at low risk.
Advice to women about whether to take up breast screening or not may need to include an evaluation of their risk factors.
1. Veronesi U, Maisonneuve P, Rotmensz N, et al. Italian randomized trial among women with hysterectomy: tamoxifen and hormone-dependent breast cancer in high-risk women. J Natl Cancer Inst 2003;95(2):160-5
2. Houssami N, Kerlikowske K. The impact of breast density on breast cancer risk and breast screening. Current Breast Cancer Reports 2012;4(2):161-68
Competing interests: No competing interests
Compliments to the Authors. Reduction in 28% of breast cancer deaths is an important finding in this observational study with modern mamography screening in Norway. It adds more data to the current debate. Some critics of mammography screening consider that their benefits could be, at least, modest, and harms should be clear to the patients. Of course, the advances in breast cancer treatment will progressively reduce the impact of screening on survival. And this is good! But it does not, at least until now, reduce the importance of the current screening model. As surgeons, we are doing more breast conserving surgeries, avoiding axillary dissection, and doing more immediate breast reconstructions, thanks to mammography screening. So, we should evaluate in future screening studies the impact on breast cancer treatment. Survival is important, but is not enough.
Competing interests: No competing interests
Mammography or fall in hormone use reduces breast cancer mortality?
Does the claimed 28% breast cancer mortality benefit of modern mammography in Norway from 2002 to 2009 allow for the reduction in breast cancer incidence and mortality due to the fall in HRT use since 2002 post WHI?1,2
Evaluations of the effectiveness of mammographic screening on breast cancer mortality must be confounded by changes in hormone use. Women on HRT the main age group screened. Breast cancers increase in incidence within one year of using HRT while use for more than 5 years trebled the risk in both WHI and MWS.3,4 Since 1962 large increases in breast cancer have matched increases in hormone use (Fig 1). There have been falls in hormone use in the 1970s, mid 1990s, and, since 2002 when the Women’s Health Institute randomised controlled trial confirmed HRT increased breast cancers and also breast cancer mortality.5-7 In 2007 the IARC confirmed that combined estrogen-progestogen contraceptives and combined estrogen-progestogen menopausal therapy were highest level (Group 1) carcinogens.8
These facts should not be ignored.
1 Weedon-Fekjær H, Romundstad PR, Vatten LJ. Modern mammography screening and breast cancer mortality: population study. BMJ 2014;348:g3701
2 Rossouw JE, Anderson GL, Prentice RL, LaCroix AZ, Kooperberg C, Stefanick ML, Jackson RD, Beresford SA, Howard BV, Johnson KC, Kotchen JM, Ockene J; Writing Group for the Women's Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women's Health Initiative randomized controlled trial. JAMA. 2002 Jul 17;288(3):321-33
3 Chlebowski RT, Hendrix SL, Langer RD, et al, for the WHI Investigators. Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women's Health Initiative randomised trial. JAMA 2003: 289: 3243-53.
4 Beral V, Banks E, Reeves G, Bull D, on behalf of the Million Women Study Collaborators. Breast cancer and hormone-replacement therapy: the Million Women Study. Lancet 2003; 362: 1331.
5 Chlebowski RT, Anderson GL, Gass M, et al. Estrogen plus progestin and breast cancer incidence and mortality in postmenopausal women. JAMA. 2010; 304(15):1684–1692
6 Grant ECG. Fall in HRT use would have reduced breast cancer mortality. BMJ (Published 28 January 2005)
7 Colditz GA. Decline in breast cancer incidence due to removal of promoter: combination estrogen plus progestin. Breast Cancer Res. 2007;9:108.
8 IARC. Combined estrogen-progestogen contraceptives and combined estrogen-progestogen menopausal therapy. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 2007; Volume 91
Competing interests: No competing interests
Re: Modern mammography screening and breast cancer mortality: population study
Dear Editor,
In Norway public mammography screening of women aged 50-69 years was gradually implemented in the period 1996-2005. Weedon-Fekjær et al have studied the effect of invitation to public screening on the breast cancer mortality and conclude that “Invitation to modern mammography screening may reduce deaths from breast cancer by about 28%” [1]. However, the total reduction in the breast cancer mortality in the age group and the period they study is about 23% [2], which must mean that there is hardly any effect of better treatment among women invited to screening. Furthermore, other authors using the same data set have reported 10% non-significant reduction related to public mammography screening [3]. The results in this paper can only be explained by statistical modelling, and a discussion of the statistical modelling is imperative.
The authors have used an age-period-cohort model which is known to be notoriously unstable [4,5] and hardly not used in medicine, but by using natural splines the authors seem to have stabilized the parameters so statistical artefacts due to over-parameterization is unlikely. The authors also write that they have done multiple sensitivity analyses and “these additional procedures did not substantially influence the estimated effect and yielded mortality rate ratios ranging from 0.71 to 0.75”. The screening program was implemented simultaneously with the introduction of multidisciplinary breast cancer treatment centers [3]. After the start-up of the public screening program, all women are treated at the same centers. An interesting sensitivity analysis is to repeat the analyses on the age group 30-49 years. Then the parameter “invited to screening at new interdisciplinary breast cancer centers” has another interpretation: “treatment by interdisciplinary teams”. What is the effect of the parameter they study in this sensitivity analyses?
The authors agree that the model cannot distinguish between better treatment and invitation to screening because these two events were introduced simultaneously [1,6]. In the authors reply below [6], they suggest that better treatment is a mediator [6] but what is the biological evidence for this theory? Chemotherapy and anti-estrogens are being used to kill micro metastases, while screening is supposed to prevent dissemination to occur. In my mind it does not make sense to assume that better treatment is a mediator of mammography screening because if early diagnosis prevents dissemination, then there is no need for adjuvant therapy or chemotherapy.
One important assumption in this paper is that the effect of treatment varies grossly with age [1,6]. In meta-analysis of breast cancer medicines there is no age difference in the effect of adjuvant hormone therapy on the mortality [7] and the effect of polychemotherapy is 38% under age 50 and 20% in the age group 50-69 years [7]. The proportions of ER+ tumors are 40% (under age 50 years) and 50% (age 50-69 years), and about 20% of all patients receive polychemotherapy [7]. A rough estimate is that there should be 20-30% improved treatment effect in the invited age group. What was the effect of improved treatment actually in this study?
Finally, the authors argue there is no major screening activity before the start of the screening program because there is no incidence increase before public screening started [1]. In the 10-year period before public screening started the breast cancer incidence increased from about 150 to 200 per 100 000 women in the invited age group [8] and the number of mammograms increased from about 10 000 per year in the early 1980s to 220 000 in 1993 [9], and most of these mammograms were done in areas that started public screening in 1996-7. In 1996-97 about 40% of all eligible women in Norway were invited to public screening and the annual number of mammograms were 127 000 in the screening program [10]. Furthermore, there were diagnosed 479 ductal carcinoma in situ (DCIS) in the invited age group in the period 1996-9 but only half of them (257) were diagnosed among women attending the public screening program. The assumption that there was no mammography screening before the public screening program started is not justified. It was a switch from opportunistic screening to organized screening, and that is what they have studied.
References
1. Weedon-Fekjær H, Romundstad PR, Vatten LJ. Modern mammography screening and breast cancer mortality: population study. BMJ 2014;348:g3701
2. Autier P, Boniol M, LaVecchia C, Vatten L, Gavin A, Héry C, et al. Disparities in breast cancer mortality trends between 30 European countries: retrospective trend analysis of WHO mortality database. BMJ 2010;341:c3620.
3. Kalager M, Zelen M, Langmark F, Adami HO. Effect of Screening Mammography on Breast-Cancer Mortality in Norway. N Engl J Med 2010;363:1203-10.
4. Greenland S, Robins J. Invited commentary: ecologic studies—biases, misconceptions, and counterexamples. Am J Epidemiol1994;139:747-60.
5. Greenland S. A review of multilevel theory for ecologic analyses. Stat Med2002;21:389-95.
6. Weedon-Fekjær H, Romundstad PR, Vatten LJ. Authors reply: Modern mammography screening and breast cancer mortality: population study . http://www.bmj.com/content/348/bmj.g3701/rr/760451
7. Early Breast Cancer Trialists' Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials. Lancet 2005;365:1687-717.
8. Zahl P-H, Strand BH, Mæhlen J. Breast cancer incidence in Norway and Sweden during introduction of nation-wide screening: prospective cohort study. BMJ 2004;328:921-4.
9. Widmark A, Olsen JB. Mammografivirksomhet i Norge. Oslo: Norwegian Radiation Protection Authority, Strålevernrapport 1995:5.
10. Mammografiprogrammet i Norge. Evaluering av prøveprosjektet 1996-2000. Forskningsrapport nr. 2-2000. Oslo: Kreftregisteret, 2001.
Competing interests: No competing interests