Intended for healthcare professionals

CCBYNC Open access

Rapid response to:

Research

Breast cancer mortality in organised mammography screening in Denmark: comparative study

BMJ 2010; 340 doi: https://doi.org/10.1136/bmj.c1241 (Published 24 March 2010) Cite this as: BMJ 2010;340:c1241

Rapid Response:

Re: Jørgensen et al. Breast cancer mortality in organised mammography screening in Denmark: comparative study

Jørgensen et al (1) claim that mammography screening in Denmark had
no impact on breast cancer mortality. This claim is unsubstantiated,
firstly because the authors used very crude data, and secondly because the
analysis was not geared to answer the question.

Firstly, breast cancer screening can only possibly have an effect on
women not already diagnosed with breast cancer prior to screening.
Therefore the so-called “refined mortality” should be used in evaluation
of screening. Jørgensen et al did not used refined mortality. Furthermore,
they merge data from three screening areas starting screening at different
points in time, and used age groups instead of cohorts. Together this gave
quite “polluted” data.

Secondly, they calculated “annual change in the relative risk of
breast cancer death” by time period and areas excluding 1992-1996. The
relevant outcome measure is, however, the change in breast cancer
mortality in the screening area controlled for the change in breast cancer
mortality in the non-screening area. Using Jørgensen et al.’s “polluted”
data this gives:

Age group	       (Death rateSS)/(Death rateSP)/
                   ((Death rateNS)/(Death rateNP)	
                                          Relative risk
                                         (95% confidence interval)
35-54 years	(29/39)/(27/36)	1.00 (0.86-1.17)
55-74 years	(102/121)/(106/109)	0.87 (0.79-0.95)
75-84 years	(195/196)/(179/175)	0.97 (0.87-1.09)



SS: Screening area, screening period

SP: Screening area, pre-screening period

NS: Non-screening area, screening period

NP: Non-screening area, pre-screening period

Thus even using these “polluted” data, the relative breast cancer
mortality decreased for women aged 55-74 covered by screening, while the
relative breast cancer mortality did not decrease for women aged 35-54
largely uncovered by screening, and the relative breast cancer mortality
was slightly but statistically non-significantly decreased for women aged
75-84 where the majority, but not all, of the person years were uncovered
by screening. Although this pattern in the data is actually visible in
Figure 1 in the paper by Jørgensen et al, it was missed in their analysis
among other things because they excluded data from the period 1992-1996.

As we have reported previously, the measured impact of mammography
screening on breast cancer mortality is highly dependent on the data set
used for the analysis. Use of “polluted” data leads to biased estimates
(2). Using cohort based refined mortality, we found a 25% decrease in
breast cancer mortality in the municipality of Copenhagen during the first
10 years following the introduction of mammography screening in April 1991
(3). We deliberately did not include data from Funen and Frederiksberg in
that analysis, as cause of death data were not available at that time for
the first 10 years of these two screening programmes.

Anne Helene Olsen, University of Tromsø

Sisse Njor, University of Copenhagen

Elsebeth Lynge, University of Copenhagen

References:

1. Jørgensen KJ, Zahl P-H, Gøtzsche PC. Breast cancer mortality in
organised mammography screening in Denmark: comparative study. BMJ 2010;
340:c1241 doi:10.1136/bmj.c1241

2. Olsen AH, Njor SH, Lynge E. Estimating the benefits of mammography
screening. The impact of study design. Epidemiology 2007;18:487-92.

3. Olsen AH, Njor SH, Vejborg I, Schwartz W, Dalgaard P, Jensen MB,
et al. Breast cancer mortality in Copenhagen after introduction of
mammography screening: cohort study. BMJ 2005;330:220-2.

Competing interests:
None declared

Competing interests: Age group (Death rateSS)/(Death rateSP)/ ((Death rateNS)/(Death rateNP) Relative risk (95% confidence interval)35-54 years (29/39)/(27/36) 1.00 (0.86-1.17)55-74 years (102/121)/(106/109) 0.87 (0.79-0.95)75-84 years (195/196)/(179/175) 0.97 (0.87-1.09)

25 March 2010
Elsebeth Lynge
Professor
Anne Helene Olsen, Sisse Njor
University of Copenhagen