Re: Lives saved by breast screening outnumber cases of overdiagnosis, review says
The EUROSCREEN Working Group published a summary paper in a supplement of Journal of Medical Screening on 14 September (1), which received a lot of media attention. Many people have asked me to post a critique of it on our centre's website, which I have done (2). I shall briefly explain the major issues here.
The authors use inappropriate methods, including extrapolations far beyond the data. They accept results of case-control studies, despite previous consensus among breast cancer screening experts that such studies cannot say anything reliably about the effect of screening (3). The bias introduced by such studies is huge, which has been demonstrated by data from the Malmö randomised screening trial. When properly analysed as a randomised trial, the reduction in breast cancer mortality was 4%, but when analysed as a case-control study (comparing breast cancer mortality in attenders with non-attenders within the screening arm), the reduction in breast cancer mortality was 58% (4). The flaw is referred to as the 'healthy screenee effect.'
Using flawed methods, the EUROSCREEN Working Group arrive at two estimates of the effect of screening on breast cancer mortality, a 38% and a 48% reduction. What is remarkable about these estimates is that they correspond fairly well to the declines in breast cancer mortality we have seen in many countries, which have occurred at the same time as highly effective adjuvant therapy was introduced. Thus, the EUROSCREEN Working Group seems to say that there has been no effect of adjuvant therapy.
The most detailed systematic reviews that have been carried out of the randomised trials by independent researchers are the Cochrane reviews and the reviews prepared by the US Preventives Services Task Force. The two teams came up with a similar estimate, a 15-16% reduction in breast cancer mortality (5,6).
Using the implausible estimates of the effect of screening, the EUROSCREEN Working Group performs a stunning act of hocus-pocus. Their starting point is the number of breast cancer deaths occurring in a 30-year period when the women are between 50 years and 79 years of age. The authors then calculate that over 30 years, so many lives will be saved that it corresponds to saving 7-9 out of every 1000 women screened. The average of this, 8 lives, is an overestimate of a factor of 16, or 1500%, compared to one woman saved out of 2000 that follows directly from the estimate derived from the randomised trials (5,6; the calculation is very simple and is explained in (5)).
We actually know a good deal about the effect when we screen for longer than the 10 years in the randomised trials without needing to use inappropriate extrapolations. Denmark is unique in the world for observational studies of mammography screening because it has a concomitant non-screened control group throughout 17 years, where screening was only offered in about 20% of the country, measured as population size (7). We found that the decline in breast cancer mortality was 1% per year in women who could benefit from screening (ages 55–74 years) in the screening areas during the 10-year period when screening could have had an effect (1997–2006), whereas it was 2% per year in the non-screening areas. The decline was larger in women who were too young to benefit from screening (ages 35–55 years), namely 5% per year in the screened areas, and 6% per year in the non-screened areas in the same time period.
These observations fit very well with recent papers in the BMJ by Philippe Autier et al. These researchers have described the breast cancer mortality trends in 30 European countries and found that the declines in young, non-screened age groups had been almost twice those in screened age groups (8). Further, they found that there was no discernible difference in the declines when comparing neighbouring countries that had introduced screening 10-15 years apart, and there was no relation whatsoever between start of screening and the reduction in breast cancer mortality (9). The fall in breast cancer mortality was not only about the same in all countries; it was also about the same as that seen in the United States (10). These observations fit very well with the fact that screening has not reduced the occurrence of large tumours (bigger than 20 mm) or of tumours in stages III and IV (11,12), which means that screening cannot have an effect on breast cancer mortality.
Even Stephen Duffy, who was quoted in BMJ's news piece about the EUROSCREEN Working Group's papers (13), acknowledges that, 'The key feature of a successful mammographic screening program is a reduction in the incidence rate of advanced tumors' (14). Duffy owes us an explanation how screening can have a dramatic effect without reducing the rate of advanced tumours. What the screening advocates have done time and again is to provide misleading information about this (15). They say that there are now relatively fewer large tumours than before screening. Of course there is, as all the small overdiagnosed tumours have been added to the denominator when such a percentage is calculated. I have illustrated why this manipulation is highly misleading (15):
Imagine a town with a certain level of crime. You divide the crimes into serious ones and less serious ones. Over a period of time, the rate of serious crime increases by 20% and the rate of less serious crime increases by 40%. This is clearly a development for the worse. But although more people are exposed to serious crime and more people are exposed to less serious crime as well, a trickster would say that, as there are now relatively fewer cases of serious crime, the situation has improved.
So, whatever the effect of screening was when the randomised trials were performed many years ago, it seems to have disappeared. This is due to much better treatments and because women attend a doctor much earlier today when they have noticed anything unusual in their breast.
The EUROSCREEN Working Group's summary article says about overdiagnosis that, 'In the absence of over-diagnosis, the initial increase in breast cancer occurrence in the screened group would be fully compensated by a similar decrease in cancers among older age groups no longer offered screening - the so-called "compensatory drop".20 The compensatory drop method requires that the screening programme has been running long enough to achieve a full adjustment for lead time.' Our study of overdiagnosis in Denmark lives fully up to these requirements, as we had a control group for 17 years without screening (16). We found 33% overdiagnosis in Denmark after adjustment for a small compensatory drop. Although our study is the most ideal there is, according to their own criteria, the EUROSCREEN Working Group's summary article does not mention our paper.
What the EUROSCREEN Working Group has documented so clearly is the crucial necessity of an independent and honest assessment of the evidence for health care interventions, free from conflicts of interest. Duffy says in an interview in the BMJ that 'it is good news that lives saved by screening outweigh
overdiagnosed cases by a factor of two to one' (13). Eugenio Paci, another author from the EUROSCREEN Working Group's paper, says: 'We believe that not only should our conclusions be communicated to women offered breast screening in Europe but that, in addition, communication methods should be improved in order to raise women’s awareness and to make information more accessible, relevant, and comprehensible' (13).
I believe the women in Europe should be spared these terribly false results. Allow me to draw attention to our leaflet about mammography screening, which we first published in the BMJ (17). It was revised in 2012 and is available at our homepage, www.cochrane.dk. The Center for Medical Consumers in the United States has described it as 'the first honest mammography information for women written by health professionals' (15). We think this is the reason that volunteers have translated it so that it now exists in 14 languages, and soon in 16, including Arabic, Chinese and Urdu. We conclude in our leaflet that it 'no longer seems reasonable to attend for breast cancer screening. In fact, by avoiding going to screening, a woman will lower her risk of getting a breast cancer diagnosis.'
1 EUROSCREEN Working Group. Summary of the evidence of breast cancer service screening outcomes in Europe and first estimate of the benefit and harm balance sheet. J Med Screeen 2012;19 Suppl 1:5-13.
2 Gøtzsche PC. Why the results from the EUROSCREEN Working Group are false. 11 Oct 2012. www.cochrane.dk/screening/EuroScreen-2012-critique.pdf.
3 Vainio H, Bianchini F. IARC Handbooks of Cancer Prevention. Vol 7: Breast Cancer Screening. Lyon: IARC Press, 2002.
4 Jørgensen KJ. Flawed methods explain the effect of mammography screening in Nijmegen. Br J Cancer 2011;105:592-3.
5 Gøtzsche PC, Nielsen M. Screening for breast cancer with mammography. Cochrane Database Syst Rev 2009;4:CD001877.
6 Humphrey LL, Helfand M, Chan BK, et al. Breast cancer screening: a summary of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med 2002;137(5 Part 1):347–60.
7 Jørgensen KJ, Zahl P-H, Gøtzsche PC. Breast cancer mortality in organised mammography screening in Denmark: comparative study. BMJ 2010;340:c1241.
8 Autier P, Boniol M, Gavin A, et al. Breast cancer mortality in neighbouring European countries with different levels of screening but similar access to treatment: trend analysis of WHO mortality database. BMJ2011;343:d4411.
9 Autier P, Boniol M, La Vecchia C, et al. Disparities in breast cancer mortality trends between 30 European countries: retrospective trend analysis of WHO mortality database. BMJ 2010;341:c3620.
10 Bleyer A. US breast cancer mortality is consistent with European data. BMJ 2011;343:d5630.
11 Autier P, Boniol M, Middleton R, et al. Advanced breast cancer incidence following population-based mammographic screening. Ann Oncol 2011;22:1726-35.
12 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.
13 Hawkes N. Lives saved by breast screening outnumber cases of overdiagnosis, review says. BMJ 2012;345:e6155.
14 Smith RA, Duffy SW, Gabe R, et al. The randomized trials of breast cancer screening: what have we learned? Radiol Clin North Am 2004;42:793-806.
15 Gøtzsche PC. Mammography screening: truth, lies and controversy. London: Radcliffe Publishing; 2012.
16 Jørgensen KJ, Zahl P-H, Gøtzsche PC. Overdiagnosis in organised mammography screening in Denmark: a comparative study. BMC Women's Health 2009;9:36.
17 Gøtzsche P, Hartling OJ, Nielsen M, Brodersen J, Jørgensen KJ. Breast screening: the facts - or maybe not. BMJ 2009;338:446-8.
Competing interests: No competing interests