Overdiagnosis in screening mammography in Denmark: population based cohort studyBMJ 2013; 346 doi: https://doi.org/10.1136/bmj.f1064 (Published 26 February 2013) Cite this as: BMJ 2013;346:f1064
All rapid responses
It is not possible from the brief description in the letter to disentangle the methodology used by Jørgensen et al. If Jørgensen et al. would state which birth cohorts from which periods, which regions and which diagnosis they are studying I can tell them why they cannot recreate our conclusions. As long as birth cohort, periods, regions and diagnosis are hidden this is not possible.
Jørgensen et al. argue that: “The authors discuss their estimate in relation to other studies without explaining this difference between increase in incidence in the screening period and increase in incidence in screening + post screening period, essentially comparing incomparable results.” We did for instance not comment on this difference in method, since this do not explain the difference between the Jørgensen study that merged data from Copenhagen, Funen and Frederiksberg, used age-groups instead of birth cohorts, population instead of individual data, and a short follow-up period and our study using individual follow-up. We did present data from both screening period and post screening period, as well as absolute numbers; it should therefore be quite easy to calculate whatever proportion one wants to based on our data.
Regarding the last paragraph, I do not have “views” on harms and benefits of screening; I provide evidence on screening outcomes.
Competing interests: No competing interests
Njor and colleagues estimated that breast screening in Copenhagen and Funen caused 2.3% overdiagnosis . Their result is in stark contrast to evidence from the randomised trials (relative risk 1.31 , relative risk 1.23 ), estimates of overdiagnosis in Norway which also has a contemporary control group of non-screened areas (relative risk 1.5), Sweden (relative risk 1.45) and North America (relative risk 1.45) [5,6,7].
One reason for the difference is that the percentage estimates are not addressing the same question. While the estimates from the randomised trials , Norway , Sweden , and USA  refer to the increase in incidence in the screening period, Njor et al estimate the increase in risk after including 8 years of follow-up after screening has stopped (they call this the “lifetime risk” ). This will provide a much smaller (“diluted”) percentage because it includes many cancers diagnosed after screening stopped in both the nominator and the denominator. The authors discuss their estimate in relation to other studies without explaining this difference, essentially comparing incomparable results . Furthermore, their percentage is entirely dependent on both the length of the screening period and the length of follow up after screening has stopped, which will usually vary considerably from study to study. Such calculations are therefore essentially meaningless because estimates cannot be compared.
The “diluting” effect cannot explain all the difference. In the paper by Njor et al., we missed a graphical presentation of the dataset they used, which would have allowed us to see whether the estimated level of overdiagnosis fits observations. We have therefore produced a graph based on the same dataset from the Danish Breast Cancer Group that Njor et al. used (Figure 1). We followed two closed cohorts over 10 years during which one of them was screened, and for ten more years without screening. The graph depicts five closed cohorts screened while in the age group 60-69 years, where the follow-up after screening is even longer (10 years) than in the study by Njor et al. (8 years). It seems that the results of Njor et al. do not fit the observed data. We find a relative risk of 1.37 (95% CI: 1.27-1.49) in the screening period when we compensate for a 10% pre-screening difference between regions, and a relative risk of 0.94 (95% CI: 0.85-1.03) in previously screened women aged 70-79 years in the 10 years after screening stops. When we take this drop into account, we find 86 overdiagnosed cases per 100,000 women, or that screening increases breast cancer incidence by 30% in the screening period due to overdiagnosis. This is in good agreement with our previous estimate of 33% overdiagnosis in Denmark .
Given the vast discrepancy in the views on harms and benefits of current breast screening programmes, and the relation between these views to the affiliation of authors with screening [Jørgensen 2007, Rasmussen 2013 in press], we suggest that assessments of breast screening programmes be performed by independent researchers.
1. 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;346:f1064.
2. Gøtzsche PC, Nielsen M. Screening for breast cancer with mammography. Cochrane Database of Systematic Reviews 2009, Issue 4. Art. No.:CD001877.
3. Independent UK Panel on Breast Cancer Screening. The benefits and harms of breast cancer screening: an independent review. Lancet 2012;380:1778-86.
4. Zahl PH, Mæhlen J. Overdiagnosis of breast cancer after 14 years of mammography screening. Tidskr Nor Laegeforen 2012;132:414-17.
5. Bleyer A, Welch HG. Effect of three decades of screening mammography on breast cancer incidence. New Engl J Med 2012;1998-2005.
6. Jørgensen KJ, Gøtzsche PC. Effect of screening mammography on breast cancer incidence. N Engl J Med 2013;368:677-8.
7. Jørgensen KJ, Gøtzsche PC. Overdiagnosis in publicly organised mammography screening programmes: systematic review of incidence trends. BMJ 2009;339:b2587.
8. Jørgensen KJ, Zahl PH, Gøtzsche PC. Overdiagnosis in organised mammography screening in Denmark. A comparative study. BMC Women’s Health 2009;9:36.
9. Jørgensen KJ, Klahn A, Gøtzsche PC. Are benefits and harms given equal attention in scientific articles on mammography screening? A cross-sectional study. BMC Medicine 2007;5:12.
10. Rasmussen K, Jørgensen KJ, Gøtzsche PC. Citations of scientific results and conflicts of interest: the case of mammography screening. Evid Based Med 2013; doi 10.1136/eb-2012-101216.
Competing interests: No competing interests
Possible underdiagnosis in screening mammography for women aged 56-69 years
To the Editor
The excellent paper of Njor and colleagues (1) arrives at a very interesting conclusion: overdiagnosis of breast cancer accounts for only a small proportion of all breast cancers diagnosed. In particular, the authors found that overdiagnosis of breast cancer including ductal carcinoma in situ amounts to 2.3% in women aged 56-69 years when first targeted for screening. Even more interestingly, despite the considerable number of participants (57763 women), no statistically significant difference between the observed and the expected cumulative breast cancer incidence was achieved and there was a broad confidence interval of the estimated overdiagnosis proportion (95% confidence interval –3% to 8%), which included negative values. Thus, in the age group 56-69 years, breast cancer underdiagnosis cannot be excluded.
The authors rightly note that the risk of overdiagnosis is the most serious concern in screening mammography. Screening guidelines should adhere to the Hippocratic aphorism “ὠφελέειν, ἢ μὴ βλάπτειν”, (to help, or at least, to do no harm). Taking into account the importance of length of follow up in studies of overdiagnosis, this well-designed cohort study shows that the estimates of overdiagnosis reported by previous researchers are possibly exaggerated (2, 3), and, in 56-69 years age group, even an underdiagnosis could occur. Large properly designed prospective studies with adequate length of follow up are needed to clarify the magnitude of overdiagnosis in 40-55 years age group where the greatest debate exists (4)
Nikolaos Vlachadis M.D., D.M.D., M.P.H., M.Sc.
Eleni Kornarou, Ph.D.
National School of Public Health, 196 Alexandras Avenue, 11521, Athens, Greece
1)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;346:f1064.
2)Jørgensen KJ, Gøtzsche PC. Overdiagnosis in publicly organised mammography screening programmes: systematic review of incidence trends. BMJ. 2009;339:b2587.
3)Bleyer A, Welch HG. Effect of three decades of screening mammography on breast-cancer incidence. N Engl J Med. 2012;367(21):1998-2005.
4)Smith RA, Kerlikowske K, Miglioretti DL, Kalager M. Clinical decisions. Mammography screening for breast cancer. N Engl J Med. 2012;367(21):e31.
Competing interests: No competing interests