Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial

BMJ 2014; 348 doi: http://dx.doi.org/10.1136/bmj.g366 (Published 11 February 2014)
Cite this as: BMJ 2014;348:g366

Recent rapid responses

Rapid responses are electronic letters to the editor. They enable our users to debate issues raised in articles published on bmj.com. Although a selection of rapid responses will be included as edited readers' letters in the weekly print issue of the BMJ, their first appearance online means that they are published articles. If you need the url (web address) of an individual response, perhaps for citation purposes, simply click on the response headline and copy the url from the browser window.

Displaying 11-20 out of 60 published

Unacceptable mammographic image quality in the early 1980s in Canada must not influence our decision about early detection in 2014.

The attention given by the media to the “update” of the failed Canadian trials and their unjustified conclusion that early detection of breast cancer does not reduce mortality from the disease is surprising, especially after the publication of the Independent UK Panel on Breast Cancer Screening which concluded in a meta-analysis of 11 randomised trials, that “the relative risk of breast cancer mortality for women invited to screening compared with controls was 0•80 (95% CI 0•73–0•89), which is a relative risk reduction of 20%” (1). The two Canadian trials are the conspicuous outliers with no mortality decrease accomplished. Dr. Miller made no response to his co-worker Dr. Boyd’s prophetic statement, “…the results of these trials should not be used to change the prevailing scientific view of the potential benefits of screening with mammography”(2).

Professor Day and colleagues pointed out in 1989 in their landmark publication on monitoring breast cancer screening programs (3) that a reduction in breast cancer mortality can occur only after a significant decrease in advanced cancers. Since the CNBSS trials are the only ones which failed to reduce the rate of advanced breast cancers, the lesson to learn from this colossal failure is that they failed to find the potentially fatal cancers, no matter how many cancers were reported. Day and colleagues summarized in their Table III “a minimum set of measures that an information system should monitor to evaluate the effectiveness of the programme in reducing severity of and mortality from the disease. A favourable value for each of these measures is necessary, however, if an acceptable effect on breast cancer mortality is to be achieved. Poor performance indicates where remedial action is required”(3).

The authors of the CNBSS make numerous misleading and erroneous comments on the Two-County Swedish Trial which do not bring us any closer to finding out why the Canadian trials are unique among all the randomized controlled trials in their failure to reduce breast cancer mortality.

To inform the reader of the published facts concerning the Two-County Swedish Trial:

• This trial was designed and run by the Swedish National Board of Health and Welfare with Professor Gunnar Eklund as the Principal Investigator. It has never been my trial, contrary to Dr. Baines’ claim; Dr. Gunnar Fagerberg was the project leader for the E-county arm and I was the project leader for the W-county arm, as has been repeatedly mentioned in the publications from the trial.

• Drs. Miller and Baines are incorrect that there was no informed consent in the Two-County Swedish trial, since “the controls were only contacted after screening ceased in the active study population”. On the contrary, the National Board of Health and Welfare saw to it that the entire population of the two counties was extremely and repeatedly well informed by the media and by activities at all levels of the local governmental agencies. The execution of the randomization process was carried out by the local politicians according to the instructions and under the supervision of Professor Gunnar Eklund.

• It is embarrassing to read the authors’ statements that the randomisation in the Swedish Trial took place on the “county level” and that the tumor size in the control group was 28 mm when in reality it was 22 mm.

• Dr. Baines notes that the Canadian trials used two-view mammography at 12 month intervals whereas the Two-County trial used one-view mammography at 33 month intervals. This only emphasizes the remarkable difference in the performance of early detection between the two studies, since the Swedish trial reduced both the advanced cancer rate and mortality from breast cancer, while the Canadian trials failed on both counts.

• Dr. Baines correctly mentions that “an external audit of mammography based on stratified sampling” took place in the Canadian trials, but she does not acknowledge the impact of the unacceptable quality found by these external audits, my own included. It is an inescapable fact that poor quality mammography means poor quality detection of the potentially fatal cancers

• Dr. Baines comments that “Trial A [Canadian] has an external death review panel to determine cause of death in all cases of deaths in participants known to have breast cancer during the trial or suspected of having breast cancer after linkage with a national data base. Not so for Trial B” [Swedish]. This is patently untrue because so far there have been two external death review panels of the Swedish Two-County Trial, reviewing every breast cancer patient’s chart combined with the linkage to the National Death Registry (4,5). The individual patient charts of all breast cancer cases are still preserved in the county archives and the data are currently under a third external review directed by Professor Richard Peto at the University of Oxford, UK. On the other hand, Dr. Miller sent me an e-mail on Aug 28, 2012, in which he declared that “it was impossible to maintain all the paper records in storage, and it would be impossible to re-conduct a death review for the CNBSS”. Imagine my surprise at the publication of this “Twenty five year follow-up” without a new death review and long after the destruction of the original documents from the CNBSS trials.

In summary: women, the medical profession and public health decision makers should rest assured that the successful early detection of breast cancer and treatment in an early stage significantly improve the outcome and life quality of women afflicted by this disease. There is ample evidence that the modern therapeutic regimens are most effective against non-palpable, screen detected, early stage breast cancer. Instead of attempting to “prove” that early detection makes no difference to the outcome of breast cancer patients, our concerted efforts should use the evidence from well conducted randomized controlled trials (1) to provide even earlier diagnosis and better treatment to achieve further improvements in the outcome of breast cancer.

László Tabár, MD.

No conflict of interest declared

References
1. Marmot MG, Altman DG, Cameron DA, Dewar JA, Thompson SG, Wilcox M. The benefits and harms of breast cancer screening: an independent review. Br J Cancer. 2013 Jun 11;108(11):2205-40.
2. Boyd NF, Jong RA, Yaffe MJ, Tritchler D, Lockwood G, Zylak CJ. A critical appraisal of the Canadian National Breast Cancer Screening Study. Radiology. 1993 Dec;189(3):661-3.
3. Day NE, Williams DR, Khaw KT. Breast cancer screening programmes: the development of
a monitoring and evaluation system. Br J Cancer. 1989;59(6):954–8.
4. Nyström L, Rutqvist LE, Wall S, Lindgren A, Lindqvist M, Rydén S, Andersson I, Bjurstam N, Fagerberg G, Frisell J, et al. Breast cancer screening with mammography: overview of Swedish randomised trials. Lancet. 1993 Apr 17;341(8851):973-8. Erratum in: Lancet 1993 Nov 27;342(8883):1372.
5. Holmberg L, Duffy SW, Yen AM, Tabár L, Vitak B, Nyström L, Frisell J. Differences in endpoints between the Swedish W-E (two county) trial of mammographic screening and the Swedish overview: methodological consequences. J Med Screen. 2009;16(2):73-80. doi: 10.1258/jms.2009.008103.

Competing interests: None declared

Laszlo K Tabar, retired radiologist

Dept of Mammography, Falun Central Hospital Swden, 4429 East Spur Dr

Click to like:

This is in response to Eugene G Breen, who wrote that in countries where screening is not performed the incidence rate is similar to the mortality and thus it is an evidence that screening decreases mortality. He mentioned that performing screening will decrease it by 20%. He also mentioned Pakistan as an example.

Since I belong to Pakistan, I would like to add that the relationship between screening and mortality is not so simple where the treatment part is missing. Since you will only benefit by screening if all you screen are treated as well. The problem for patients in poor countries like Pakistan is that either they do not have access to proper treatment or they can't afford it. Compounding the problem are social taboos which do not let women seek treatment despite being diagnosed with the disease. Thus the mortality rate in these countries will not be reduced automatically by screening unless adequate treatment is ensured as well.

Competing interests: None declared

Afsheen Zafar, General Surgeon

Islamic International Medical College, Rawalpindi, Pakistan, Unit 2, Department of Surgery, Railway hospital, Rawalpindi. Pakistan.

Click to like:

If existing data is consistent with conflicting interpretations, if it does not command consensus amongst reasonable disinterested qualified people, then the problem is not the quality of the analyses but the quality of the data.

Data that cannot clinch the matter is no basis for a screening programme.

It is uncontroversial that if a woman attends mammography screening, no conscientious practitioner can give her useful information about what her results mean.

Nobody can tell her that a negative mammogram means she doesn’t have cancer, still less that she won’t have it tomorrow.

If the mammogram shows something suspicious, diagnostic tests may yield a negative result. All anyone can say is that the bit that caused concern probably wasn’t cancerous, according to this competent pathologist, on balance. They can’t say she has no cancer, only that nothing else showed on the mammogram, and they can’t say that she won’t have cancer tomorrow.

Diagnostic tests may show cancer. Whether invasive or noninvasive, nobody can tell her that it will progress without treatment since some, but they don’t know how many, and they don’t know which, do not. They can tell her that estimates of numbers vary widely. So they don’t know if she has progressive cancer or not, and on this basis it’s time to decide about treatment.

She’ll want to know whether, if her cancer is progressive, earlier treatment will prolong her life. Nobody can tell her there is a good chance it will because nobody knows if there is any chance it will. Nobody can honestly say that anyone declining screening, or declining treatment at point of diagnosis after screening, will die sooner because they have not had earlier treatment.

They can tell her the wide range of estimates of the other treatment outcomes: life shortened, no difference to lifespan, unnecessarily treated. But they can’t say which will affect her. They can say how different the situation is now from when the poor to middling quality trials were carried out, making the estimates inapplicable to her present predicament. They can tell her about more recent research, but they’d have to say they can’t be more precise because the experts are still – er – debating – its implications and it would be misleading to give only one side of the debate.

She will wonder if anything she has heard constitutes information. For all she’s been through, she knows no more about her situation with regard to breast cancer than before she was screened. She could have read the tea leaves.

A person who declines screening is not risking death. Screening, if it worked, would offer a chance of longer life than fate decrees at a price, so declining it would be forgoing a chance of extra time and not paying the price, not risking earlier death. As things stand, being screened is risking grave harm for no known benefit. Declining it is opting out of Russian roulette.

No conscientious expert would deny these things.

Competing interests: Diagnosed through screening

Miriam Pryke, PhD student

King's College London, WC2R 2LS

Click to like:

A screening program monitors a population for disease over an extended period of time. The prevalent (first) round of screening contains many confounding variables and so relying on that first round of screening when evaluating a detection technology can be misleading. Tumours caught in the first round of screening that lead to patient death contribute to a degradation in the mortality associated with the screening technology being evaluated. When a lethal tumour has been caught in the first round of screening, the technology being relied upon (in this case mammography) did not have a chance of catching that malignancy during its treatable stage as it already reached a lethal stage of growth by the time the screening technology was applied to the patient. Thus combining tumours detected in the first round of screening with those detected at a later stage biases the analysis against mammography.

The confounding variables associated with the first round of screening are reason enough to focus one’s evaluation of a disease screening technology on subsequent screening rounds in order for the experiment to better reflect the reality of an on-going screening program. If a screening program detects many lethal tumours in its first round of screening it is justification for changing the parameters of the screening program so that it has a chance to catch those lethal tumours earlier on in their development (for example: by inviting patients to join screening at a younger age when their tumours would have, on average, been less developed or did not yet exist).

In a study such as this [1] where serious concerns have been raised regarding potential problems in patient randomisation [2, 3], relying on data from the first round of screening is liable to increase the inherent bias against screening described above. The results from the second round of screening onwards [1] indicate a mortality benefit from mammographic screening. Pooling the data from the first round of screening with subsequent rounds as has been relied upon in the author’s analyses biases the results against mammography, may have contributed to misleading the public and is based on the author’s assumptions, not facts.

Miller et al. consider the hazard ratio for death from breast cancer caught in the first round of screening (1.47) to be expected. There does appear to be considerable range in the literature that could influence expectation, however, it is entirely plausible that some of this hazard ratio’s large value was due to inappropriate randomisation. The authors have countered claims that their randomisation was inappropriate by citing the external review of their study which didn’t find any problems. It should be noted that every peer reviewed scientific study that has ever been retracted was first subjected to a review process that overlooked the study’s critical errors. That this study cleared an external review process does not mean it is free from inappropriate randomisation.

Jacob Levman, PhD
Institute of Biomedical Engineering
University of Oxford

[1] A B Miller, et al. Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial, British Medical Journal, 2014;348:g366.
[2] D B Kopans, Re: Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial, British Medical Journal, 2014;348:g366, Feb. 12th 2014.
[3] L K Tabar, Re: Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial, British Medical Journal, 2014;348:g366, Feb. 18th 2014.

Competing interests: None declared

Jacob Levman, Researcher

Institute of Biomedical Engineering, University of Oxford, Parks Road, Oxford, OX1 3PJ

Click to like:

The terms `breast self-examination` (BSE) and `breast awareness` (BA) mean different things to different people: there is much confusion. [1] [2] Publication of the updated Canadian Study has raised many difficult issues, including the practice of physical breast examination. [3] Authors, in their rapid response 10th March 2014, in particular to Malgrem, provide some clarification about the ethics and practice of breast examination of various kinds at the time the study was launched.

‘Touching and finding’ can occur in different situations motivated by different attitudes of mind. ‘Chance detection’ can occur when women who are sensibly alert are showering, bathing or dressing; many cancers are found this way. ‘Deliberate detection’ can occur when women purposefully and routinely practise BSE or BA with the intention to check for abnormalities. This intervention is not without harms. [4]

But BSE continues in spite of evidence that it cannot now be recommended. A Cochrane Review concludes: `Data from two large trials do not suggest a beneficial effect of screening by breast self-examination but do suggest increased harm in terms of increased numbers of benign lesions identified and an increased number of biopsies performed. At present, screening by breast self-examination or physical examination cannot be recommended`. [4]

It is likely that women who are ‘sensibly alert’ to the possibility of finding an abnormality by chance will be less anxious than those who deliberately and routinely practice BSE or BA.

It is vital that clarification is provided so that women are not left without proper guidance. [5]

[1] Thornton H. Pillarisetti RR. ‘Breast awareness’ and ‘breast self-examination’ are not the same. What do these terms mean? Why are they confused? What can we do? Eur. J. Cancer 2008; 44:2118-2121

[2] McMenammin M, Barry H, Lennon AM, Purcell H, Baum M, Keegan D, McDermott E, O`Donoghue D, Daly L, Mucahy H. A survey of breast cancer awareness and knowledge in a Western population: lots of light but little illumination. Eur J Cancer. 2005 Feb;41(3):393-7.

[3] Miller AB, Wall C, Baines CJ, Sun P, To T, et al. Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomized screening trial. BMJ 2014;348:g366

[4] Kösters JP, Gøtzsche PC. Regular self-examination or clinical
examination for early detection of breast cancer. Cochrane
Database of Systematic Reviews 2003;(2). Art. No.: CD003373. doi:
10.1002/14651858. CD003373 (No change, Update, Issue 3, July
2008)

[5] Dillner L. Breast cancer: should I examine myself for lumps? | Women are often advised to check their breasts once a month. But there is strong evidence that self-examination doesn't work. The Guardian. 30th September 2012.
www.theguardian.com/lifeandstyle/2012/sep/30/breast-cancer-examine-mysel...

Competing interests: None declared

Hazel Thornton, Honorary Visiting Fellow, Department of Health Sciences

None

University of Leicester, "Saionara", 31 Regent Street, Rowhedge, Colchester, CO5 7EA

Click to like:

Jacob Levman (RR 10 March 2014) has misunderstood our data. He states: “The hazard ratio comparing breast cancer mortality from women in the mammography arm to women in the control arm during the first round of screening was reported as 1.47, implying that being assigned to the mammography arm was associated with substantially increased risk of dying of breast cancer (based on the results of the first round of screening only).” He has overlooked the fact that the prevalence screen has added to the mammography arm the cancers detected by mammography alone, half of which we have demonstrated are not over-diagnosed. Thus there will inevitably be more deaths from breast cancer in the mammography arm than the control arm in a sub-group analysis restricted to the prevalence screen. We also reported the hazard ratio for deaths from cancers diagnosed in years 2 to 5 was 0.90 (0.69 to 1.16; P=0.40), thus indicating that the apparently additional deaths from the prevalence screen were compensated by reduced deaths in subsequent screens, i.e. as we demonstrated in other analyses no overall breast cancer mortality reduction from mammography screening. We reported these results to emphasise the fact that the additional cancers, brought forward by the lead time from mammography, increases the period of life lived with cancer without benefit.

Competing interests: None declared

Anthony B. Miller, Professor Emeritus

Baines CJ, To T, Wall C Narod SN

Dalla Lana School of Public Health, University of Toronto, 155 College Street, Toronto, ON, M5T 3M7, Canada

Click to like:

The screening battle has now achieved cold war proportions. Switzerland as a vanguard has pulled its screening program and others seem to be considering the same, based on responses to this article.

What happens when you don't screen? They don't screen in Iran or Pakistan or most of Asia and what is found is that the prevalence of breast cancer is up to 80% less than in Europe or USA. They find that breast cancer presents much later and the mortality figures are similar to those of EU and USA. What is starkly different is the ratio of breast cancer incidence to breast cancer mortality. In non screened countries the death rate from breast cancer is very close to the incidence rate, presumably because they present late and do not have comprehensive treatments. In the screened countries the incidence rates are multiples of the death rates because the treatments are available and also because they present earlier. If you were to take the screening out of the EU and USA the death rates would climb up to the incidence rates, as is the natural process in those countries with no screening. So screening works. It successfully reduces the death rates (as published on Cancer Research UK website) to 18 per 100,000 population from an incidence of 90 per 100,000 population in EU.

Examples are: the breast cancer incidence in East Africa is 19 per 100,000 population and the death rate is 11.4. In South Africa the incidence is 38 per 100,000 population and the death rate is 19. The incidence in South Eastern Asia is 26 per 100,000 population and the death rate is 14. In USA the incidence is 76 per 100,000 population and the death rate is 15. In Western EU the incidence is 90 per 100,000 and the death rate is 18.

Even allowing for incomplete data these figures are impressive and consistent. They suggest that if screening was introduced into third world countries the death rate would fall to 20% of the incidence rate.

The other aspect of these data is why is breast cancer so common in developed countries and much less common in third world countries. Deaths are very countable no matter what country you look at and late breast cancer is easily detected, and so the raw data of breast cancer deaths is a good surrogate for incidence in non-screened countries. This implies that the incidence actually is as low as is reported.

The three consistent robust and parallel patterns seen world wide are increasing breast cancer incidence, reduced birth rate and increased use of exogenous female hormones.1.

1. The Screech Owls of Breast Cancer. Author House. 2013.

Competing interests: None declared

Eugene G Breen, Physician/psychiatrist

Mater Misericordiae University Hospital, Dublin 7

Click to like:

In this Rapid Response, we take the opportunity to respond both to recent comments on our study (1), as well as previous comments to which we have not already responded.

Ana Rosengurtt (RR 5 March 2014) is concerned about the radiation implicit in mammography. So were we when planning the study. However our computations suggested benefit would exceed any risk from radiation (2), and a subsequent study showed that women exposed to multiple fluoroscopies over the age of 35 had little increased risk of breast cancer (3). During the study, our reference physicist took care to ensure the mammography machines did not emit too much radiation, while maintaining adequate image quality (4). Nevertheless, we agree that unnecessary exposure to radiation should be avoided as far as possible.

Mohammad Tahir (RR 1 March 2014) suggests that screening mammography would offer “survival advantage” if DCIS cases had not been excluded from our analyses. That is not so. Any woman with a previously diagnosed DCIS who developed invasive breast cancer was included in our analyses from the date of diagnosis of the invasive cancer. Any deaths from such cancers were included in our analyses of mortality. There is good evidence that DCIS is not a breast cancer precursor but a marker of future risk of breast cancer. In no breast screening trial has it been demonstrated that detection and treatment of DCIS following screening results in a reduction in subsequent incidence of breast cancer. The results of the follow-up of cases of in situ cancer ascertained in the trial has just appeared on line in the International Journal of Cancer (5).

Mona P. Tan (RR 26 February 2014) is concerned that our study design required informed consent from all participants and that the women in the control arm were not blinded as to their allocation. Implicitly she raises the question as to whether the breast examinations and the teaching of breast self-examination to all women resulted in the null effect of mammography on breast cancer mortality. Application of the Miscan model to our data suggested the breast examinations might have resulted in a 20% reduction in breast cancer mortality (6). However, as the modellers used data from the Two-county trial, the absence of modern adjuvant therapy in that trial (7) may have led to an over-estimation of the effect of the breast examinations. Dr Tan is concerned that absence of mammography in screening could lead to litigation because of delay in diagnosis. Perhaps more critical is that over-diagnosis by mammography screening may lead to litigation for over-treatment and its consequences.

Elsebeth Lynge and Sisse Njor (RR February 2014) comment on small increases in numbers of screen-detected breast cancers in our succeeding reports on mortality in the trial. They are correct in that the increases from 1992 to 2000-2 resulted from delayed registration of cancers in some provincial cancer registries. The difference from 2000-2 to the present report arose from re-coding of histology to ensure that we only considered invasive breast cancers.

J. Tristram C Dammin (RR 19 February 2014) is concerned that we do not make any reference to breast size, shape or "lumpy, bumpy breast" as possible confounding issues that could have nullified our data.
 Although we did not specifically characterise women as to whether they had “lumpy” breasts, at the first screen referral by the examiners to review by the study surgeon in the centres for abnormalities detected on breast examination was almost identical in the two arms, 14.1% in the mammography arm and 14.6% in the control arm for women age 40-49 (8), and 11.0% and 11.2%, respectively, for women age 50-59 (9). Thus there was no excess in the mammography arm which would have been anticipated had women suspected of having breast cancer been shifted from the control to the mammography arm as postulated by Kopans (RR 12 February 2014). Dr Dammin also asks if our data can be extrapolated to all postmenopausal women. We believe so, there is no major shift in the behaviour of breast cancer at the age of 60, although incidence and competing causes of death (and therefore over-diagnosis) increase with increasing age.

Peter Sasieni (RR 17 February 2014) had asked whether there was a typographical error in Table 3 and that the number presented of fatal breast cancers incident “in year 5” should have been “in years 2-5”. Dr Sasieni is correct, we regret the error.


Martin Heilweil (RR 17 February 2014) comments “The study is silent on treatment for these discovered conditions”. We have no reason to believe treatment decisions were differential between the arms. However, we have data on the initial surgery applied and we performed a special study of treatment given for women age 40-49. Treatment quality was assessed by experienced oncologists blind as to allocation (10). There was no evidence that treatment was inferior in the mammography arm.

Judith Malmgren (RR 14 February 2014) pointed out that at no point in the RCT was there an arm without either initial clinical breast examination, training in self breast examination, or usual care. That is correct, such an arm was considered unethical at the time we initiated the study, in part because of the results of the HIP trial (11), and because the Canadian Cancer Society, a major sponsor of the trial, had adopted a policy of advocating breast self examination (BSE). Many consider there is no benefit from BSE, though there are good observational data from Finland (12) as well as from a nested case-control study within the CNBSS (13) that indicate a benefit. Dr Malmgren comments that a relatively small number of cancers, 32%, were found by mammography alone in the mammography arm of the study. That is true, but she has failed to include the cancers detected by both mammography and breast examination, which brings the total found by mammography to 69% of the screen-detected breast cancers among women age 40-49 (8) and 81% among women age 50-59 (9). Dr Malmgren also states that the mean tumour sizes found in the control arm are not as expected in a study with a nonintervention control arm. However, we suspect that the patient/physician detected (palpable) tumour mean size of 3 cm in her study was based upon a symptomatic group of women, not an asymptomatic group as enrolled in our trial. Her statement that fewer deaths from breast cancer were seen among the mammography detected tumors compared to the palpable tumors is true, but the figures she quotes should be corrected for the over-diagnosis resulting from mammography. She also comments “All breast cancer discovered in this RCT had a hefty breast cancer death rate of 20 to 33% presenting a situation in which it does not appear any of the breast cancer discovered was not going to present harm to the patient during her lifetime.” When the estimated number of over-diagnosed breast cancers is deducted from the denominator in the mammography arm, the proportion of women with screen-detected breast cancers who died from breast cancer during the 25-year follow-up becomes 32% in the mammography arm compared to 33% in the control.

Carmen Vidal and Montse Garcia (RR 14 February 2014) wondered why we did not stratify the analysis according to age groups. In fact we reported under the heading Breast Cancer Mortality: “For women aged 40-49 at assignment the hazard ratio for 25 year breast cancer specific mortality associated with mammography was 1.09 (95% confidence interval 0.80 to 1.49; P=0.58) and for women aged 50-59 at assignment was 1.02 (0.77 to 1.36; P=0.88).” (1) As the results were similar, we did retain this stratification throughout our report. Vidal and Garcia also suggested we should use a “population approach” to the estimation of over-diagnosis. We were interested in their table 1, but this does not give an adequate depiction of the extent of over-diagnosis in relation to screen-detected cancers. They ask why we chose the 15-year point to compute the extent of over-diagnosis, as did Lynge and Njor (RR 19 February 2014). Fifteen years was when the differential between the numbers of breast cancers ascertained in the two arms became stable, indicating that the effect of lead time in the mammography arm was past. Subsequently, towards the end of the period of follow-up, the numbers of breast cancers newly ascertained falls off in each arm, and the differential between the arms suffers from random fluctuation. Computing over-diagnosis at the end of the period of screening results in confounding by lead time. The proportion of screen-detected cancers over-diagnosed at 10 years is 21%, and at the end of follow-up 24%.

References
1. Miller AB, Wall C, Baines CJ, Sun P, To T, Narod SA. Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial. BMJ 2014.348:g366 doi: 10.1136/bmj.g366.

2. Howe GR, Sherman GJ, Semincew RM, Miller AB. Estimated benefits and risks of screening for breast cancer. CMA Journal 1981; 124:399-403.

3. Miller AB, Howe GR, Sherman GJ et al Mortality from breast cancer after radiation during fluoroscopic examination in patients being treated for tuberculosis. New Engl. J. Med. 1989; 321:1285-89.

4. Yaffe M, Mawdsley GE, Nishikawa RM. Quality assurance in a national breast screening study. Proc Soc Photo-Optical Instrum Eng 1983; 419: 23-30.

5. To T, Wall C, Cornelia J. Baines CJ, Miller AB. Is carcinoma in situ a precursor lesion of invasive breast cancer? Int. J. Cancer 2014; 00: 00–00.

6. Rijnsburger AI, van Oortmarssen GJ, Boer R, Draisma G, To T, Miller AB, de Koning
HJ. Mammography benefit in the Canadian National Breast Screening Study-2:
a model evaluation. Int J Cancer 2004;110:756–62.

7. Tabar L, Chen H-HT, Duffy SW, Kruesmo UB. Primary and adjuvant therapy, prognostic factors and survival in 1053 breast cancers diagnosed in a trial of mammography screening. Jpn J Clin Oncol 1999;2129:608–16.

8. Miller AB, Baines CJ, To T, Wall C. Canadian National Breast Screening Study: 1. Breast cancer detection and death rates among women aged 40 to 49 years. CMAJ 1992; 147:1459-76.

9. Miller AB, Baines CJ, To T, Wall C. Canadian National Breast Screening Study: 2. Breast
cancer detection and death rates among women aged 50 to 59 years. CMAJ 1992; 147:1477-88.

10. Kerr M: A Case-Control Study of Treatment Adequacy and Mortality from Breast Cancer for Women of Age 40-49 Years at Entry into the National Breast Screening Study (master's thesis), U of Toronto, Toronto, 1991.

11. Shapiro S, Strax P, Venet L. Periodic breast cancer screening in reducing mortality from breast cancer. JAMA 1971;215:1777-85.

12. Gastrin G, Miller AB, To T, Aronson KJ, Wall C, Hakama M et al. Incidence and mortality from breast cancer in the Mama program for breast screening in Finland, 1973-1986. Cancer 1994; 73:2168-74.

13. Harvey BJ, Miller AB, Baines CJ, Corey PN. Effect of breast self-examination techniques on the risk of death from breast cancer. Can Med Assoc J 1997;157:1205-12.

Competing interests: None declared

Anthony B. Miller, Professor Emeritus

Cornelia Baines, Claus Wall, Teresa To, Ping Sun, Steven Narod

Dalla Lana School of Public Health, University of Toronto, 155 College Street, Toronto, ON, M5T 3M7, Canada

Click to like:

Miller’s recent article on the results of the Canadian National Breast Screening Study [1] has been heavily criticized for errors in the randomisation employed, leading to more challenging patients being disproportionately assigned to the mammography arm of the trial [2, 3]. Miller has made the claim that randomisation was done appropriately [4]. However, it should be noted that Miller’s own mortality data [1] is supportive of the criticisms raised against his study [2, 3]. The hazard ratio comparing breast cancer mortality from women in the mammography arm to women in the control arm during the first round of screening was reported as 1.47 [1], implying that being assigned to the mammography arm was associated with substantially increased risk of dying of breast cancer (based on the results of the first round of screening only). This is a remarkable imbalance regarding the hazard associated with death from breast cancer between the mammography arm and the control arm and the most likely explanation is that the criticisms regarding the randomisation employed in this trial are valid [2, 3]. Women presenting to the trial with pre-existing concerns regarding their breast health were most likely regularly assigned to the mammography arm of the trial for this study’s results to demonstrate such a marked increase in the hazard of breast cancer mortality in the mammographic screening arm based exclusively on the first year of the trial.

It is not ideal to assess a disease screening technology based on the first round of screening (even in trials that have randomized their subjects appropriately). When we consider a patient who died of cancer detected in the first round of a screening trial, we simply won’t know whether the technology would have contributed to saving that patient’s life had the patient been enrolled in screening earlier on in the progression of their disease. As such, it is more reliable to form conclusions regarding the efficacy of a screening technology based on its results from the second round of screening onwards. Incidentally, the authors provide this data which indicates a mortality benefit (hazard ratio 0.9) from the use of mammographic screening after the first year of the trial. This indicates that mammography has contributing to lowering breast cancer mortality after the initial overly challenging cases were attended to in the first year of screening. This hazard ratio (0.9) reporting a mortality benefit from mammography after the first year of screening did not achieve a p-value below 0.05 and as such was not deemed statistically significant and was apparently ignored when establishing the study’s conclusions. However, it is doubtful that this calculation included enough breast cancer mortality samples to achieve sufficient statistical power for the statistical significance test to be reliably relied upon. Had the authors collected many more samples that continued to produce an identical hazard ratio of 0.9, the finding would have been statistically significant as it is known that the amount of separation required between two groups in order to achieve statistical significance diminishes as the number of samples increases. It is unfortunate that the total number of samples contributing to this study’s mortality calculations from cancer detected in screening rounds 2 through 5 is so limited, as this study’s mortality results (much like this study’s survival results) appear positive for mammography.

Jacob Levman, PhD
Institute of Biomedical Engineering
University of Oxford

[1] A B Miller, et al. Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial, British Medical Journal, 2014;348:g366.
[2] D B Kopans, Re: Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial, British Medical Journal, 2014;348:g366, Feb. 12th 2014.
[3] L K Tabar, Re: Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial, British Medical Journal, 2014;348:g366, Feb. 18th 2014.
[4] A B Miller, Re: Twenty five year follow-up for breast cancer incidence and mortality of the Canadian National Breast Screening Study: randomised screening trial, British Medical Journal, 2014;348:g366, Feb. 17th 2014.

Competing interests: None declared

Jacob Levman, Researcher

Insitute of Biomedical Engineering, University of Oxford, Parks Road, Oxford, OX1 3PJ

Click to like:

Dear Dr. Jerome S. Schroeder,

Please check your references.

You accuse BMJ of being one-sided in the breast screening question and cite “recently published results” that favor the intervention claiming they support your argument. However, neither the 25% breast cancer mortality reduction, nor the 4% overdiagnosis rate in Denmark that you mention is “recently published results”. The paper you allude to [1] is in fact a position paper that mostly sums up previously published results. In fact, the above mentioned results were published in two separate papers in BMJ [2,3].

We have also published results based on Danish data, including in the BMJ, but reached quite different results [4,5]. In short, the full 25% reduction was claimed to occur within the first 3 years after screening was introduced [2] and therefore cannot be related to screening, as the effect takes 5-10 years to materialise [4]. In the period where breast screening could have an effect, mortality rates declined to a similar extent in both screened and non-screened areas and we could not detect an effect of Danish breast screening in our analyses [4]. The low estimate of overdiagnosis do not fit the observed data [5] and is likely caused by a diluting effect caused by long follow-up with inclusion of all cancers detected long after screening stops. It is simply wrong when you state that the position paper shows that breast screening has lowered mastectomy rates in Denmark, as the paper contains no formal analyses of this and therefore cannot show anything about it [1]. We published data on Danish mastectomy rates in 2011 [6]. After large increase in mastectomy use in the screened areas at introduction, mastectomy rates declined in a similar fashion in both screened and non-screened areas alike due to national changes in treatment policy unrelated to screening [6]. We showed the same for Norway [7].

I would like to thank Dr. Schroeder for demonstrating so clearly to us all how emotion and good intentions can overcome a rational and courteous approach to the debate on breast screening. This can make individuals jump to conclusions; make them accept messages that they like without attention to detail; and make them be quick to dismiss messages they do not like along with those who send them and ascribe them hidden malicious motives.

1. Christiansen P, Vejborg I, Kroman N, Holten I, Garne JP, Vedsted P et al. Position paper: Breast cancer screening, diagnosis, and treatment in Denmark. Acta Oncol 2014; Early Online:1-12.
2. Njor SH, Olsen AH, Blichert-Toft M, Schwartz W, Vejborg I, Lynge E. Overdiagnosis in screening mammography in Denmark: population based cohort study. BMJ 2013;346f1064.
3. Olsen AH, Njor SH, Vejborg I, Schwarz W, Dalgaard P, Jensen MB, Tange UB, Blichert-Toft M, Rank F, Mouridsen H, Lynge E. BMJ 2005;330:220.
4. Jørgensen KJ, Zahl PH, Gøtzsche PC. Overdiagnosis in organsed breast screening in Denmark. A comparative study. BMC Womens Health 2009;9:36.
5. Jørgensen KJ, Zahl PH, Gøtzsche PC. Breast cancer mortality in organised mammography screening in Denmark. A comparative study. BMJ 2010;340:c1241.
6. Jørgensen KJ, Keen J, Gøtzsche PC. Is Mammographic Screening Justifiable Considering its Substantial Overdiagnosis Rate and Minor Effect on Mortality? Radiology 2011;260:621-7.
7. Surhke P, Mæhlen J, Schlichting E, Jørgensen KJ, Gøtzsche PC, Zahl PH. Mammography screening and surgical breast cancer treatment in Norway: comparative analysis of cancer registry data. BMJ 2011;343:d4692.

Competing interests: None declared

Karsten J. Jørgensen, MD, DrMedSci

The Nordic Cochrane Centre, Rigshospitalet, Department 3343

Click to like:

THIS WEEK'S POLL