The debate on breast cancer screening with mammography is important
2004; Elsevier BV; Volume: 1; Issue: 1 Linguagem: Inglês
10.1016/s1546-1440(03)00017-6
ISSN1558-349X
Autores Tópico(s)Cancer Risks and Factors
ResumoThe issue of breast cancer screening is worthy of continued scrutiny and scientific and public debate [1Horton R Screening mammography—an overview revisited.Lancet. 2001; 358: 1284-1285Abstract Full Text Full Text PDF PubMed Scopus (70) Google Scholar]. The quality of the scientific evidence is variable, and new evidence appears all the time. Furthermore, the effect of breast cancer screening is uncertain [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar], and screening leads to overdiagnosis and overtreatment [3Olsen O Gøtzsche P.C Cochrane review on screening for breast cancer with mammography.Lancet. 2001; 358: 1340-1342Abstract Full Text Full Text PDF PubMed Scopus (616) Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar]. Hence, the balance between possible benefits and harms is delicate, which underlines the need for honest information to women also on harms and individual decision making rather than blanket recommendations [5Thornton H Edwards A Baum M Women need better information about routine mammography.Br Med J. 2003; 327: 101-103Crossref PubMed Scopus (80) Google Scholar]. We [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar, 3Olsen O Gøtzsche P.C Cochrane review on screening for breast cancer with mammography.Lancet. 2001; 358: 1340-1342Abstract Full Text Full Text PDF PubMed Scopus (616) Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar] and others, such as the Early Breast Cancer Trialists’ Collaborative Group [6Early Breast Cancer Trialists’ Collaborative GroupEffects of radiotherapy and surgery in early breast cancer an overview of the randomized trials.N Engl J Med. 1995; 333: 1444-1455Crossref PubMed Scopus (1069) Google Scholar, 7Early Breast Cancer Trialists’ Collaborative GroupFavourable and unfavourable effects on long-term survival of radiotherapy for early breast cancer an overview of the randomised trials.Lancet. 2000; 355: 1757-1770Abstract Full Text Full Text PDF PubMed Scopus (1195) Google Scholar], have pointed out why bias in the assessment of cause of death in screening trials would be expected to favor screening. The clearest example of bias has been provided by the Swedish Two-County Trial (Table 1). The 2002 overview of the Swedish trials reported only a 10% reduction in breast cancer mortality for the Östergötland part of the Two-County Trial [8Nyström L Andersson I Bjurstam N Frisell J Nordenskjöld B Rutqvist L Long-term effects of mammography screening updated overview of the Swedish randomised trials.Lancet. 2002; 359: 909-919Abstract Full Text Full Text PDF PubMed Scopus (961) Google Scholar] (data were not made available for the Kopparberg part), whereas the lead investigator of the Two-County Trial in 2000 reported a 24% reduction [9Tabar L Vitak B Chen H.H et al.The Swedish Two-County Trial twenty years later. Updated mortality results and new insights from long-term follow-up.Radiol Clin North Am. 2000; 38: 625-651Abstract Full Text Full Text PDF PubMed Scopus (411) Google Scholar]. With a slightly longer follow-up, there were 10 fewer breast cancer deaths in the study group and 23 more in the control group [9Tabar L Vitak B Chen H.H et al.The Swedish Two-County Trial twenty years later. Updated mortality results and new insights from long-term follow-up.Radiol Clin North Am. 2000; 38: 625-651Abstract Full Text Full Text PDF PubMed Scopus (411) Google Scholar].Table 1Two-County study, Östergötland partNo. of Breast Cancer DeathsRelativeRiskInvitedControlSwedish overview 20021771900.90Two-County 20001672130.76Same age group (40–74). Same statistics (evaluation model). Similar follow-up (1.2 vs 1.3 mio women-years). Open table in a new tab Same age group (40–74). Same statistics (evaluation model). Similar follow-up (1.2 vs 1.3 mio women-years). This large discrepancy of 33 deaths in favor of screening is difficult to explain, because the data come from the same randomized trial and from the same age group of women. The attempts to explain the discrepancy by Professor Peter Dean [10Dean P.B Gøtzsche’s quixotic antiscreening campaign nonscientific and contrary to Cochrane principles.J Am Coll Radiol. 2004; 1: 3-7Abstract Full Text Full Text PDF PubMed Scopus (4) Google Scholar] or the trial’s authors [11Tabár L Smith R.A Duffy S.W Update on effects of screening mammography.Lancet. 2002; 360: 337Abstract Full Text Full Text PDF PubMed Scopus (28) Google Scholar, 12Duffy S.W Tabár L Smith R.A The mammographic screening trials commentary on the recent work by Olsen and Gøtzsche (authors’ reply).J Surg Oncol. 2002; 81: 164-166Crossref Google Scholar] have not been elucidating. The most likely explanation is that the assessment of cause of death was not performed blindly in the Two-County Trial [8Nyström L Andersson I Bjurstam N Frisell J Nordenskjöld B Rutqvist L Long-term effects of mammography screening updated overview of the Swedish randomised trials.Lancet. 2002; 359: 909-919Abstract Full Text Full Text PDF PubMed Scopus (961) Google Scholar]; this fact was recently confirmed by one of the investigators closely involved with the study [13Crewdson J Swedes doubt mammography trial. 2002Google Scholar]. The issue is one of bias, not one of “conspiracy” or “deliberate bias,” as Dean asserts [10Dean P.B Gøtzsche’s quixotic antiscreening campaign nonscientific and contrary to Cochrane principles.J Am Coll Radiol. 2004; 1: 3-7Abstract Full Text Full Text PDF PubMed Scopus (4) Google Scholar]. Dean argues, with reference to data from the Two-County Trial, that screening saves lives because all-cause mortality among those with diagnoses of breast cancer is significantly lower when the women have been screened. This argument is common in the screening literature, but it is fallacious [14Berry D.A The utility of mammography for women 40 to 50 years of age (Con).in: DeVita V.T Hellman S Rosenberg S.A Progress in oncology. Jones and Bartlett, Sudbury, UK2002: 346-372Google Scholar, 15Gøtzsche P.C Mammography service screening and mortality.Lancet. 2003; 362: 329-330Abstract Full Text Full Text PDF PubMed Scopus (1) Google Scholar]. The purpose of randomization is to create two comparable groups. Thus, provided the randomization is adequate, one can compare all-cause mortality among all those randomized without bias. However, Dean’s argument relates to two subgroups of the randomized women, those with diagnoses of breast cancer. These two groups of women are not comparable. Screening leads to overdiagnosis (see below) and predominantly identifies slow-growing tumors (length bias), including carcinoma in situ, which most often does not progress and is not detected clinically. Breast cancers diagnosed in a group invited for screening are therefore different from those identified in a control group, and these women have much better prognoses, on average, than those in the control group because relatively more of them are healthy. This bias is so large that it can explain breast cancer mortality reductions of 50% or even more [14Berry D.A The utility of mammography for women 40 to 50 years of age (Con).in: DeVita V.T Hellman S Rosenberg S.A Progress in oncology. Jones and Bartlett, Sudbury, UK2002: 346-372Google Scholar]. Lead-time bias also invalidates such comparisons, although probably to a lesser extent [14Berry D.A The utility of mammography for women 40 to 50 years of age (Con).in: DeVita V.T Hellman S Rosenberg S.A Progress in oncology. Jones and Bartlett, Sudbury, UK2002: 346-372Google Scholar]. In our Cochrane review [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar] we compared all-cause mortality among all randomized women. This is not only an unbiased analysis, it is also the most relevant analysis for the women, because it might not help them to get screened if they do not live longer but die of something else, for example, of complications to the treatment of their breast cancer or from other cancers when they have two cancers [3Olsen O Gøtzsche P.C Cochrane review on screening for breast cancer with mammography.Lancet. 2001; 358: 1340-1342Abstract Full Text Full Text PDF PubMed Scopus (616) Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar]. It is therefore important to point out that the screening trials have failed to demonstrate an effect not only on all-cause mortality but also on all-cancer mortality [3Olsen O Gøtzsche P.C Cochrane review on screening for breast cancer with mammography.Lancet. 2001; 358: 1340-1342Abstract Full Text Full Text PDF PubMed Scopus (616) Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar]. Dean writes that the criteria we used to assess the quality of the screening trials were devised by me, that they were “homemade,” that they have not been used by other investigators, and that they have limited relevance to population screening trials. These allegations are all incorrect. Cochrane reviews are based on well-established quality criteria [16Clarke M Oxman A.D Cochrane reviewers’ handbook 4.2.0.in: The Cochrane Library. Update Software, Oxford, UK2003Google Scholar, 17Guyatt G.H Sackett D.L Cook D.J Users’ guides to the medical literature. II. How to use an article about therapy or prevention. A. Are the results of the study valid? Evidence-Based Medicine Working Group.J Am Med Assoc. 1993; 270: 2598-2601Crossref PubMed Scopus (984) Google Scholar, 18Guyatt G.H Sackett D.L Cook D.J Users’ guides to the medical literature. II. How to use an article about therapy or prevention. B. What were the results and will they help me in caring for my patients? Evidence-Based Medicine Working Group.J Am Med Assoc. 1994; 271: 59-63Crossref Scopus (730) Google Scholar], which we followed. We classified the available data according to (1) whether the randomization was adequate and led to comparable groups, (2) whether postrandomization exclusions were few or unbiased, and (3) whether reliable outcome data were available [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar, 3Olsen O Gøtzsche P.C Cochrane review on screening for breast cancer with mammography.Lancet. 2001; 358: 1340-1342Abstract Full Text Full Text PDF PubMed Scopus (616) Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar]. These criteria are highly relevant for trials of screening as well as for other interventions. Screening is an intervention the main aim of which is to improve health. Thus, it can and should be evaluated according to the same strict principles as trials on drugs, in particular because screening addresses the healthy population and therefore has a delicate balance between possible benefits and harms. Dean sees our arguments as circuitous, but it is difficult to find the basis for this interpretation. It is quite linear—and in accordance with good scientific practice—to first plan a systematic literature search, set quality criteria for studies found, use these criteria systematically, and then look at studies of higher and lesser quality separately [16Clarke M Oxman A.D Cochrane reviewers’ handbook 4.2.0.in: The Cochrane Library. Update Software, Oxford, UK2003Google Scholar]. We found that trials of higher quality not only failed to show effects on breast cancer mortality but that the effect estimates were significantly different from those of the lower quality trials (Table 2). This should be cause for concern.Table 2Mortality ascribed to breast cancer after 13 years Dean criticizes the Canadian trials and notes that they were not included in the World Health Organization report’s “final evaluation.” He fails to say, however, that the report noted that “these trials were valid” [19Vainio H Bianchini F IARC handbooks of cancer prevention Vol. 7. Breast cancer screening. IARC Press, Lyon, France2002Google Scholar, p. 94). It would indeed be a mistake not to include valid trials in a meta-analysis of screening. Dean notes that we should have included the extended data from the Malmö study in our review. However, as we have shown [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar], these data are not reliable. In particular, there was no formal protocol for the extension, and there are unexplained imbalances that should not have occurred if the study had been correctly randomized [20Gøtzsche P.C Update on effects of screening mammography.Lancet. 2002; 360: 338-339Abstract Full Text Full Text PDF PubMed Google Scholar]. Like so many other screening advocates, Dean ignores the harms of screening. Cancer screening without overdiagnosis and overtreatment is not possible, and the 30% overtreatment we reported [3Olsen O Gøtzsche P.C Cochrane review on screening for breast cancer with mammography.Lancet. 2001; 358: 1340-1342Abstract Full Text Full Text PDF PubMed Scopus (616) Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar] agrees well with U.S. epidemiological data that suggest 26% overdiagnosis of invasive cancers (and about 35% if carcinoma in situ is added) [21Gøtzsche PC. On the benefits and harms of screening for breast cancer. Int J Epidemiol. In press.Google Scholar]. That overdiagnosis occurs was reported by one of the authors Dean quotes to show the opposite [22Fletcher S.W Elmore J.G Mammographic screening for breast cancer.N Engl J Med. 2003; 348: 1672-1680Crossref PubMed Scopus (315) Google Scholar], and this author drew attention to the U.S. data [23Ries L.A.G Eisner M.P Kosary C.L et al.SEER cancer statistics review, 1973–1999. National Cancer Institute, Bethesda, MD2002Google Scholar, Table IV-1). Overdiagnosis does not only reflect the fact that most cases of carcinoma in situ do not progress. There is also considerable overdiagnosis of invasive cancers [24Miller A.B To T Baines C.J Wall C The Canadian National Breast Screening Study—1 breast cancer mortality after 11 to 16 years of follow-up. A randomized screening trial of mammography in women age 40 to 49 years.Ann Intern Med. 2002; 137: 305-312Crossref PubMed Google Scholar]. This may seem surprising, but it is because the average age of breast cancer diagnosis in the screening trials is about 60 years [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar] and because many of the invasive cancers detected by screening grow quite slowly [25Fox M.S On the diagnosis and treatment of breast cancer.J Am Med Assoc. 1979; 241: 489-494Crossref Scopus (105) Google Scholar]. Some of these cancers are therefore detected only if the women are screened, because if they are not screened, the cancers will not become apparent before the women die from other causes. This very basic and important fact—that many cancers are histologically malignant, but biologically benign [25Fox M.S On the diagnosis and treatment of breast cancer.J Am Med Assoc. 1979; 241: 489-494Crossref Scopus (105) Google Scholar, 26Doll R Peto R The causes of cancer quantitative estimates of avoidable risks of cancer in the United States today.J Natl Cancer Inst. 1981; 66: 1191-1308Crossref PubMed Scopus (3981) Google Scholar]—is either ignored or not understood by many screening advocates. But it has been described in studies of tumor biology [27Spratt J.S Meyer J.S Spratt J.A Rates of growth of human neoplasms part II.J Surg Oncol. 1996; 61: 68-83Crossref PubMed Scopus (118) Google Scholar] and in studies of breast cancer survival. As an example, in one study, the majority of invasive cancers grew slowly, with a relative annual mortality rate of only 2.5% [25Fox M.S On the diagnosis and treatment of breast cancer.J Am Med Assoc. 1979; 241: 489-494Crossref Scopus (105) Google Scholar]. The overdiagnosis necessarily leads to mastectomies that would not have been performed in the absence of screening [3Olsen O Gøtzsche P.C Cochrane review on screening for breast cancer with mammography.Lancet. 2001; 358: 1340-1342Abstract Full Text Full Text PDF PubMed Scopus (616) Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar]. In the southeast Netherlands, for example, when screening was introduced from 1990 to 1998, the number of women who underwent breast-conserving surgery increased by 71%, but the number of women who underwent mastectomies increased by 84% [28Ernst M.F Voogd A.C Coebergh J.W.W Repelaer van Driel O.J Roukema J.A The introduction of mammographical screening has had little effect on the trend in breast-conserving surgery a population-based study in southeast Netherlands.Eur J Cancer. 2001; 37: 2435-2440Abstract Full Text Full Text PDF PubMed Scopus (29) Google Scholar, 29Gøtzsche P.C Trends in breast-conserving surgery in the southeast Netherlands comment on article by Ernst and colleagues Eur J Cancer 2001, 37, 2435–2440.Eur J Cancer. 2002; 38: 1288Abstract Full Text Full Text PDF PubMed Scopus (1) Google Scholar]. If the study had included carcinoma in situ there would have been even more mastectomies [30West Midlands NHS Breast and Cervical Screening Quality Assurance Reference Centre. An audit of screen detected breast cancers for the year of screening April 1999 to March 2000. BASO Breast Audit 1999/2000, also available at http://www.cancerscreening.nhs.uk/breastscreen/publications.html.Google Scholar]. Dean quotes three papers to support his view that screening leads to less, not more, mastectomies [31de Koning H.J van Oortmarssen G.J van Ineveld B.M van der Maas P.J Breast cancer screening its impact on clinical medicine.Br J Cancer. 1990; 61: 292-297Crossref PubMed Scopus (32) Google Scholar, 32Foster Jr, R.S Farwell M.E Costanza M.C Breast-conserving surgery for breast cancer patterns of care in a geographic region and estimation of potential applicability.Ann Surg Oncol. 1995; 2: 275-280Crossref PubMed Scopus (60) Google Scholar, 33Paci E Duffy S.W Giorgi D et al.Are breast cancer screening programmes increasing rates of mastectomy? Observational study.Br Med J. 2002; 325: 418Crossref PubMed Google Scholar]. However, the first paper did not report actual data but built on a simulation model with unverified assumptions [31de Koning H.J van Oortmarssen G.J van Ineveld B.M van der Maas P.J Breast cancer screening its impact on clinical medicine.Br J Cancer. 1990; 61: 292-297Crossref PubMed Scopus (32) Google Scholar]. The authors predicted that the level of overdiagnosis would be 3% after 10 years, whereas concrete data have shown that it is 10-fold higher [3Olsen O Gøtzsche P.C Cochrane review on screening for breast cancer with mammography.Lancet. 2001; 358: 1340-1342Abstract Full Text Full Text PDF PubMed Scopus (616) Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar, 21Gøtzsche PC. On the benefits and harms of screening for breast cancer. Int J Epidemiol. In press.Google Scholar, 23Ries L.A.G Eisner M.P Kosary C.L et al.SEER cancer statistics review, 1973–1999. National Cancer Institute, Bethesda, MD2002Google Scholar]. The second paper reported that the use of breast-conserving surgery for invasive cancer in Vermont increased from 9% in 1975–1984 to 43% in 1989–1990 [32Foster Jr, R.S Farwell M.E Costanza M.C Breast-conserving surgery for breast cancer patterns of care in a geographic region and estimation of potential applicability.Ann Surg Oncol. 1995; 2: 275-280Crossref PubMed Scopus (60) Google Scholar]. However, it also noted that 67% of all patients in 1975–1984 might have been treated by breast-conserving surgery and that most of the variation in type of surgery was related to other factors than stage of disease, probably to local community factors and physician attitudes. Second, percentages cannot say anything about whether there is a true increase in the mastectomy rate (or a slower decrease than what one would have found if there had been no screening). The third paper [33Paci E Duffy S.W Giorgi D et al.Are breast cancer screening programmes increasing rates of mastectomy? Observational study.Br Med J. 2002; 325: 418Crossref PubMed Google Scholar] built on a false assumption [34Gøtzsche P.C Misleading paper on mastectomy rates in a screening programme. 2002Google Scholar]. The authors asserted that if screening increases the number of mastectomies, populations in which screening has been introduced should see a subsequent increase. This is not correct. The mastectomy rate has gone down steadily throughout many years, even in countries without screening. The question therefore is whether this decline is slower when there is screening than when there is no screening. The study did not address this question but merely noted a decrease in the mastectomy rate in the period when screening was introduced. This is not surprising and does not prove anything. Dean correctly points out that modern radiotherapy might be less harmful than previous methods. However, the study that has claimed this [35Hojris I Overgaard M Christensen J.J Overgaard J Morbidity and mortality of ischaemic heart disease in high-risk breast-cancer patients after adjuvant postmastectomy systemic treatment with or without radiotherapy analysis of DBCG 82b and 82c randomised trials. Radiotherapy Committee of the Danish Breast Cancer Cooperative Group.Lancet. 1999; 354: 1425-1430Abstract Full Text Full Text PDF PubMed Scopus (317) Google Scholar] had too short a follow-up period and too little power to prove it [36EBCTCG SecretariatBreat cancer survival advantage with radiotherapy.Lancet. 2000; 356: 1271Abstract Full Text Full Text PDF PubMed Google Scholar]. Furthermore, it is not likely that any type of radiotherapy would be completely free of adverse, and even lethal, effects on the heart and vessels. Women cannot provide truly informed consent for screening when they are only informed about the possible benefits, not about the major potential harms [5Thornton H Edwards A Baum M Women need better information about routine mammography.Br Med J. 2003; 327: 101-103Crossref PubMed Scopus (80) Google Scholar]. A recent editorial noted, in response to another paper [37Humphrey L.L Helfand M Chan B.K Woolf S.H Breast cancer screening a summary of the evidence for the U.S. Preventive Services Task Force.Ann Intern Med. 2002; 137: 347-360Crossref PubMed Scopus (253) Google Scholar] Dean quotes as evidence of the absence of harm, “Most women are unaware that mammography increases the risk for lumpectomy and mastectomy. Ignoring these extra surgical interventions creates the illusion of almost complete physical safety from mammography, tilting the cost-benefit calculus in a more favorable direction” [38Goodman S.N The mammography dilemma a crisis for evidence-based medicine?.Ann Intern Med. 2002; 137: 363-365Crossref PubMed Google Scholar]. Dean insists that I had not made a comprehensive analysis of these organizations. He then describes his own search and even draws a conclusion based on it, without giving any details of his methods and naming only one of the organizations to which he refers. In fact, we had done a systematic search for Web sites and had collected all the data from the 27 sites we located well in advance of the meeting in Helsinki to which Dean refers. Our research shows that the informational material provided by professional advocacy groups and governmental organizations is biased in favor of screening [39Jørgensen KJ, Gøtzsche PC. How possible benefits and harms related to screening for breast cancer are presented to women on web sites: cross-sectional study. Br Med J. In press.Google Scholar] (Table 3). Table 3ScreeningThe most balanced and comprehensive information on breast cancer screening comes from consumer Web sitesWeb sites from cancer charities and governmental agencies are persuasive, not informative Open table in a new tab Concerning Dean’s own specialty, radiology, we found one Web site, RadiologyInfo, that is funded by the American College of Radiology and the Radiological Society of North America [40RadiologyInfo. Home page. Available at: http://www.radiologyinfo.org. Accessed October 4, 2002.Google Scholar]. This Web site notes, The use of screening mammography increases the detection of small abnormal tissue growths confined to the milk ducts in the breast, called ductal carcinoma in situ (DCIS). These early tumors cannot harm patients if they are removed at this stage and mammography is the only proven method to reliably detect these tumors. Women are not told that the detection and removal of “these early tumors” can harm them and that their detection presents a dilemma for medicine because we do not know what the best strategy is (wait and see or immediate action, and if so, what action, lumpectomy or mastectomy, and plus or minus radiotherapy?). The main reason we decided to publish a version of our systematic review in The Lancet in 2001 [3Olsen O Gøtzsche P.C Cochrane review on screening for breast cancer with mammography.Lancet. 2001; 358: 1340-1342Abstract Full Text Full Text PDF PubMed Scopus (616) Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar] simultaneously with the publication of our Cochrane review [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar] was that the Cochrane Breast Cancer Review Group deleted our data on overtreatment from the Cochrane review, although these data were envisaged in the published protocol for the review they had themselves approved [41Gøtzsche P.C Screening mammography setting the record straight.Lancet. 2002; 359: 440-441Abstract Full Text Full Text PDF PubMed Google Scholar]. We felt that this information is important for decision making. When we published an earlier Lancet review in 2000 [42Gøtzsche P.C Olsen O Is screening for breast cancer with mammography justifiable?.Lancet. 2000; 355: 129-134Abstract Full Text Full Text PDF PubMed Scopus (706) Google Scholar], the word Cochrane appeared only as our affiliation and in our search strategy. Nonetheless, the editors of the Cochrane Breast Cancer Review Group received complaints requesting that we should not be allowed to complete our Cochrane review on this topic. The editors very properly withstood these attempts at censorship. Similarly, The Lancet’s editor called for an open scientific debate [43Horton R Screening mammography setting the record straight (editor’s reply).Lancet. 2002; 359: 441-442Abstract Full Text Full Text PDF Google Scholar]. In contrast, Dean follows his line of personal attacks by putting the blame for any problem on me, but by so doing, he misrepresents the historical events. Others see them—and in particular their causes—quite differently [1Horton R Screening mammography—an overview revisited.Lancet. 2001; 358: 1284-1285Abstract Full Text Full Text PDF PubMed Scopus (70) Google Scholar, 41Gøtzsche P.C Screening mammography setting the record straight.Lancet. 2002; 359: 440-441Abstract Full Text Full Text PDF PubMed Google Scholar, 43Horton R Screening mammography setting the record straight (editor’s reply).Lancet. 2002; 359: 441-442Abstract Full Text Full Text PDF Google Scholar, 44Gøtzsche P.C Olsen O More on mammography (authors’ reply).Lancet. 2000; 356: 1276Abstract Full Text Full Text PDF Google Scholar, 45The C.B.C.G Screening mammography setting the record straight.Lancet. 2002; 359: 439-440Abstract Full Text Full Text PDF PubMed Scopus (2) Google Scholar]. Dean suggests that we should have consulted the Cochrane Screening and Diagnostic Tests Methods Group. However, this is not part of review procedures, because methods groups do not have the capacity to consult individual authors. Furthermore, reviewing the quality of reviews is the task of the review groups, not the methods groups. Dean notes that there is an annual subscription fee of £500 for access to the Cochrane Library. But he can get access through his hospital district at no personal charge. And a growing number of countries have national subscriptions with free access for all citizens, for example, the United Kingdom, Ireland, Norway, Australia, and various Latin American nations (Cochrane reviews have been translated into Spanish). Dean complains that in a short report on screening we submitted to the Danish National Board of Health on May 10, 1999, we had not paid attention to a letter we received from a trialist. The only letter we have that fits his description came from an investigator involved with the Two-County Trial, in response to three questions we asked. However, there would be no possible way this letter could have been included, because we wrote to this investigator in June, after the submission. It is unclear to this day how the clusters were randomized in the Kopparberg part of the Two-County Trial. The letter noted that the research group had published all the details regarding the comparability of the group invited to screening and the control group, with special emphasis on socioeconomic status and age, and gave a number of references we were recommended to read. However, we already knew that there was a difference in age, and none of the references contained any data on comparability of socioeconomic status or other factors of major prognostic importance. We asked again about the missing data on socioeconomic status and were told that they were in an enclosed reprint, but this paper did not contain the data either [46Tabar L Fagerberg G Duffy S.W Day N.E Gad A Grontoft O Update of the Swedish two-county program of mammographic screening for breast cancer.Radiol Clin North Am. 1992; 30: 187-210PubMed Google Scholar]. We have had good contacts with the lead investigators from all the other screening trials, who have provided answers to our questions. The lead investigator of the Two-County Trial has withdrawn his collaboration with the other Swedish trialists [8Nyström L Andersson I Bjurstam N Frisell J Nordenskjöld B Rutqvist L Long-term effects of mammography screening updated overview of the Swedish randomised trials.Lancet. 2002; 359: 909-919Abstract Full Text Full Text PDF PubMed Scopus (961) Google Scholar], but it is perhaps about time that the Two-County Trial authors start responding adequately to the important criticisms they have received from other researchers throughout the years, for example, about the randomization [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar], the cause of death classification [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar, 8Nyström L Andersson I Bjurstam N Frisell J Nordenskjöld B Rutqvist L Long-term effects of mammography screening updated overview of the Swedish randomised trials.Lancet. 2002; 359: 909-919Abstract Full Text Full Text PDF PubMed Scopus (961) Google Scholar, 11Tabár L Smith R.A Duffy S.W Update on effects of screening mammography.Lancet. 2002; 360: 337Abstract Full Text Full Text PDF PubMed Scopus (28) Google Scholar], the many unexplained discrepancies in numbers of randomized women and numbers of deaths [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar, 4Olsen O Gøtzsche P.C Systematic review of screening for breast cancer with mammography. 2002Google Scholar, 21Gøtzsche PC. On the benefits and harms of screening for breast cancer. Int J Epidemiol. In press.Google Scholar], and the serious length and lead-time biases in their subsequent observational studies [15Gøtzsche P.C Mammography service screening and mortality.Lancet. 2003; 362: 329-330Abstract Full Text Full Text PDF PubMed Scopus (1) Google Scholar, 47Tabar L Yen M.-F Vitak B Chen H.-H.T Smith R.A Duffy S.W Mammography service screening and mortality in breast cancer patients 20-year follow-up before and after introduction of screening.Lancet. 2003; 361: 1405-1410Abstract Full Text Full Text PDF PubMed Scopus (597) Google Scholar]. When these biases were last pointed out to the authors, they “dismissed” the criticism by referring to an analysis their paper did not contain, without providing any data to support their claim [48Tabar L Duffy S.W Smith R.A Mammography service screening and mortality (authors’ reply).Lancet. 2003; 362: 330Abstract Full Text Full Text PDF Google Scholar]. Dean writes that we made no further contacts to the “breast cancer screening community” after May 10, 1999, until a press conference in January 2000. This is not correct. We had contacts with trialists and other knowledgeable people before, during, and after this interval, and I have not participated in any press conference at any time. Dean misrepresents our research in the same way as he has done before [49Dean P.B Screening med mammografi, en pålitlig undersökningsmethod.Läkartidningen. 2001; 98: 5924-5926PubMed Google Scholar], although it has been pointed out to him [50Gøtzsche P.C Misvisende fremstilling af vort arbejde om screening med mammografi.Läkartidningen. 2002; 99: 75-76PubMed Google Scholar]. He quotes from the results section of the abstract of our Cochrane review [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar]: “If data from all eligible trials (excluding flawed studies) are considered then the relative risk for breast cancer mortality after 13 years is 0.80 (95% CI 0.71-0.89).” But the sentence before his quote notes, “The best trials failed to show a significant reduction in breast cancer mortality with a relative risk of 0.97 (95% CI 0.82–1.14).” And the sentences after are as follows: “However, breast cancer mortality is considered to be an unreliable outcome and biased in favour of screening. Flaws are due to differential exclusion of women with breast cancer from analysis and differential misclassification of cause of death.” Furthermore, our conclusion in the abstract is that the currently available reliable evidence does not show a survival benefit of mass screening for breast cancer (and the evidence is inconclusive for breast cancer mortality). Women, clinicians and policy makers should consider these findings carefully when they decide whether or not to attend or support screening programs. Dean is correct that we had no a priori opinion on the effect of screening for breast cancer when we started to review the trials. Neither do we have any conflict of interest on the issue. Dean calls for an explanation of two conflicting statements, but there is no conflict. One statement concerns the effect (i.e., a scientific issue), and the other concerns possible harms and whether cancer screening is good or bad overall, which is a judgmental issue. We have not stated that the study and control groups in the Swedish trials received different levels of care, as Dean suggests. Dean’s quotation of a letter by the Two-County Trial’s authors [51Duffy S.W Tabar L Smith R.A Screening for breast cancer with mammography.Lancet. 2001; 358: 2166Abstract Full Text Full Text PDF PubMed Scopus (32) Google Scholar] is also misleading, because these authors—in contrast to Dean—did not suggest that we stated this. Our argument was of a general nature. We wrote, Control group women in the trials received usual care whereas women in the screened group, in particular those with screen-detected cancers, would be more likely to be treated in specific centres [ref. 92]. This could have resulted in better management and treatment, but there was not enough information in the trial reports to reveal any bias.Our reference 92 was to the Canadian study whose authors gave this information. In a meeting in Stockholm, according to Dean, I “had no arguments to prove [my] accusations of bias.” However, I presented the evidence, it was given to the participants in written format, and my abstract is still available on the home page of the organization that arranged the meeting [52Vilka effekter har mammografiscreening?2003Google Scholar]. Dean also misrepresents the discussion in Helsinki, where he failed to give evidence for his views. I show in this paper the four slides I presented in Helsinki that were related to mammography screening (Table 1, Table 2, Table 3, Table 4). Dean complains that in press interviews, I frequently refer to myself as a biologist—which I also am. I have two full academic educations, the other as a physician. Dean claims that we should not have responded to “elementary analytical errors” in our review. Dean is not concrete, but I assume he refers to a criticism by Djulbegovic and Hozo, who accused us of “elementary analytical errors.” This criticism was published with our Cochrane review, and we have refuted it [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar]. Dean mentions other “factual and methodological errors” but does not specify them, which makes it impossible to respond. I have similarly refuted criticisms by Tabar, Duffy, and Smith [53Duffy S.W Tabár L Smith R.A The mammographic screening trials commentary on recent work by Olsen and Gøtzsche.J Surg Oncol. 2002; 81: 159-162Crossref PubMed Scopus (12) Google Scholar], who suggested arithmetic inconsistencies, although there were none [54Gøtzsche P.C Invited response to “The mammographic screening trials commentary on recent work by Olsen and Gøtzsche.”.J Surg Oncol. 2002; 81: 162-163Crossref Google Scholar]. None of the criticisms I have received, including the most detailed discussion of our research [55Freedman DA, Petitti D, Robins JM. On the efficacy of screening for breast cancer. Int J Epidemiol. In press.Google Scholar], have had consequences for the validity of our review [21Gøtzsche PC. On the benefits and harms of screening for breast cancer. Int J Epidemiol. In press.Google Scholar]. Dean is correct that meta-analysis, like any statistical technique, can be misused. And it often has been. For example, screening advocates have repeatedly omitted the Canadian trials from their meta-analyses of breast cancer screening, and they have only occasionally paid any attention to the quality of the trials [2Olsen O Gøtzsche P.C Screening for breast cancer with mammography (Cochrane review).in: The Cochrane Library, Issue 4. Update Software, Oxford, UK2001Google Scholar]. Our methods and analyses are transparent and can therefore provide a basis for a sound and informed scientific debate.
Referência(s)