Rapid responses are electronic comments to the editor. They enable our users
to debate issues raised in articles published on bmj.com. A rapid response
is first posted online. If you need the URL (web address) of an individual
response, simply click on the response headline and copy the URL from the
browser window. A proportion of responses will, after editing, be published
online and in the print journal as letters, which are indexed in PubMed.
Rapid responses are not indexed in PubMed and they are not journal articles.
The BMJ reserves the right to remove responses which are being
wilfully misrepresented as published articles or when it is brought to our
attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not
including references and author details. We will no longer post responses
that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
In a recent letter in the BMJ, Rosén and Rehnqvist have criticised
our analysis of the effects of mammography screening in Sweden (1,2,).We
have responded to their criticism in every aspect (3). The main reason is
the fact that the predicted number of breastcancer deaths from clinical
trials is significantly different from what is actually observed, a decade
after introduction of
mammography mass-screening in Sweden (no significant reduction). The
Swedish Board of Health and Welfare claims that mammography screening is
effective because this has been proven in clinical trials; ”few methods in
health care have ever been evaluated in clinical trials of such high
quality and been proven to be as effective as mammography screening” (our
translation). Because their reasoning is critically dependent on the
quality of these trials, they need to be scrutinized.
Three papers from these studies are frequently referred to. The meta-
analysis of five Swedish clinical trials was published by Nyström et al
1993 in The Lancet (4). In this paper, analysing the outcome 6-13 years
after randomisation into study and control groups showing a significant
24% reduction of breastcancer mortality, the authors reports that the
control groups in four of the five trials were invited to screening 4-5
years after randomisation! However the Malmö trial published by Andersson
et al 1988, showed an excess mortality the first 6 years of mammography
screening (5). This finding could be of great importance as screening in
the control group would lead to false “excess mortality”. If the results
in the 1993 paper are recalculated using
the Malmö results, subtracting the excess mortality contamination, no
significant reduction of breast cancer mortality remains in the study.
The ”Two County” trial, by Tabár et al, is considered by Rosén and
Rehnqvist to be one of the best clinical trials in the field of
mammography screening research. There are, however, several oddities
connected with the planning, analysis and publications of this study. In
both the paper 1985 in The Lancet and the nine year follow up 1989 it is
stated that ”the control group, (passive study population), was not
invited to screening” (6,7). The fact is that screening started in the
control group in 1982-83 (4)! Furthermore, it is reasonable to assume that
Tabár et al were aware of the results from the Malmö study at the
time they wrote up the 1989 paper. It is therefore strange that they chose
not to discuss their somewhat unusual “design”, in fact, they chose not to
reveal the screening in the control group until 1993.
With this in mind, can we stay assured that the positive findings in
the statistical analysis of the trial are not the result of a “fishing
expedition” where different ways of using the controls and to subgroup the
data where the means by which the magic p<0.05 finally appeared?
Interestingly, the Two County trial design was neither fully
presented in the paper 1985 nor in the preliminary report 1981 (8). One
needs to go back to 1979 to find the full design of that study (9). In the
1979 publications it is stated that the Two County Trial should be
- a cluster randomised study, based on geographic clusters
- a controlled trial with intervention by mammography screening in the
study group and no intervention in the control group
- the study should include two rounds of screening in the study group
- the observation time was set to be five years.
There is only one of the five Swedish clinical trials that meets the
criteria of a randomised and controlled clinical trial - the Malmö study.
This study showed a 4% non-significant reduction of mortality in breast
cancer after 9 years of mammography screening. The Two County trial has
seriously violated their original protocol and the descriptions of the
study. The 1985 and 1989 publications are remarkable distortions from what
was actually done according to the 1993 paper (4,6,7).
Thus because several assumptions, critical to the reliability of a
controlled clinical trial are not fulfilled, estimates of the efficacy of
mammography screening can only be based on the results from the Malmö
trial. The results from that trial are fully compatible with our analysis
of the effects of mammography screening in daily practice in 17 Swedish
Counties (1,5). As the
meta-analysis 1993 by Nyström et al includes the two County trial, and the
fact that they accept intervention in the control groups despite the
results from the Malmö study, means that their results are invalid and not
reliable.
Conclusion: There is no significant decrease in breastcancer
mortality in Sweden one decade after introduction of
mammography screening. Thus there is no scientific evidence, convincingly
showing that mass-screening by mammography is
effective in reducing breast cancer mortality. Neither is safety of the
method well documented and a re-evaluation is therefore urgently needed.
Göran Sjönell MD PhD FRCGP Family Physician
Kvartersakuten Matteus
Surbrunnsgatan 66 SE
11327 Stockholm
Sweden
Lars Ståhle MD PhD Assistant Professor
Department of Clinical Pharmacology
Karolinska Institute
Huddinge University Hospital
SE-14186 Huddinge
Sweden
References
1 Sjönell G, Ståhle L. Hälsokontroller med mammografi minskar inte
dödligheten i bröstcancer. (in Swedish) Läkartidningen 1999;96:904-13.
2 Rosén M, Rehnqvist N. No need to reconsider breast screening programme
on basis of results from defective study.BMJ 1999;318:809.
3 Sjönell G, Ståhle L. Hur länge skall vi behöva vänta på att se effekt på
dödligheten? Läkartidningen;96:1822-23.
4 Nyström L, Rutqvist LE, Wall S et al. Breast cancer screening with
mammography: overview of Swedish randomised trials. Lancet 1993;341:973-8
5 Andersson I, Aspegren U, Janzon L, Landberg T, Lindholm K et al.
Mammografic screening and mortality from breast
cancer: the Malmö mammografic screening trial. BMJ 1988;297:943-48.
6 Tabár L, Fagerberg CJG, Gad A, Baldertorp L, Holmberg LH et al.
Reduction in mortality from breast cancer after mass
screening with mammography. Lancet 1885;i:829-832.
7 Tabár L, Fagerberg G, Duffy SW, Day NE. The Swedish Two-County trial of
mammografic screening for breast cancer:
recent results and calculation of benefit. J Epidemiol. Common Health
1989;43:107-17.
8 Tabár L, Gad A. Screening for breast cancer; the Swedish trial.
Radiology 1981;138:219-22.
9 Tabár L, Gad A, Akerlund E et al. Screening for breast cancer in Sweden.
A randomized trial. (in) Logan WW, Muntz EP, eds Reduced dose Mammography.
Masson, 1979, pp 183-200.
Mammography screening in Sweden
Editor
In a recent letter in the BMJ, Rosén and Rehnqvist have criticised
our analysis of the effects of mammography screening in Sweden (1,2,).We
have responded to their criticism in every aspect (3). The main reason is
the fact that the predicted number of breastcancer deaths from clinical
trials is significantly different from what is actually observed, a decade
after introduction of
mammography mass-screening in Sweden (no significant reduction). The
Swedish Board of Health and Welfare claims that mammography screening is
effective because this has been proven in clinical trials; ”few methods in
health care have ever been evaluated in clinical trials of such high
quality and been proven to be as effective as mammography screening” (our
translation). Because their reasoning is critically dependent on the
quality of these trials, they need to be scrutinized.
Three papers from these studies are frequently referred to. The meta-
analysis of five Swedish clinical trials was published by Nyström et al
1993 in The Lancet (4). In this paper, analysing the outcome 6-13 years
after randomisation into study and control groups showing a significant
24% reduction of breastcancer mortality, the authors reports that the
control groups in four of the five trials were invited to screening 4-5
years after randomisation! However the Malmö trial published by Andersson
et al 1988, showed an excess mortality the first 6 years of mammography
screening (5). This finding could be of great importance as screening in
the control group would lead to false “excess mortality”. If the results
in the 1993 paper are recalculated using
the Malmö results, subtracting the excess mortality contamination, no
significant reduction of breast cancer mortality remains in the study.
The ”Two County” trial, by Tabár et al, is considered by Rosén and
Rehnqvist to be one of the best clinical trials in the field of
mammography screening research. There are, however, several oddities
connected with the planning, analysis and publications of this study. In
both the paper 1985 in The Lancet and the nine year follow up 1989 it is
stated that ”the control group, (passive study population), was not
invited to screening” (6,7). The fact is that screening started in the
control group in 1982-83 (4)! Furthermore, it is reasonable to assume that
Tabár et al were aware of the results from the Malmö study at the
time they wrote up the 1989 paper. It is therefore strange that they chose
not to discuss their somewhat unusual “design”, in fact, they chose not to
reveal the screening in the control group until 1993.
With this in mind, can we stay assured that the positive findings in
the statistical analysis of the trial are not the result of a “fishing
expedition” where different ways of using the controls and to subgroup the
data where the means by which the magic p<0.05 finally appeared?
Interestingly, the Two County trial design was neither fully
presented in the paper 1985 nor in the preliminary report 1981 (8). One
needs to go back to 1979 to find the full design of that study (9). In the
1979 publications it is stated that the Two County Trial should be
- a cluster randomised study, based on geographic clusters
- a controlled trial with intervention by mammography screening in the
study group and no intervention in the control group
- the study should include two rounds of screening in the study group
- the observation time was set to be five years.
There is only one of the five Swedish clinical trials that meets the
criteria of a randomised and controlled clinical trial - the Malmö study.
This study showed a 4% non-significant reduction of mortality in breast
cancer after 9 years of mammography screening. The Two County trial has
seriously violated their original protocol and the descriptions of the
study. The 1985 and 1989 publications are remarkable distortions from what
was actually done according to the 1993 paper (4,6,7).
Thus because several assumptions, critical to the reliability of a
controlled clinical trial are not fulfilled, estimates of the efficacy of
mammography screening can only be based on the results from the Malmö
trial. The results from that trial are fully compatible with our analysis
of the effects of mammography screening in daily practice in 17 Swedish
Counties (1,5). As the
meta-analysis 1993 by Nyström et al includes the two County trial, and the
fact that they accept intervention in the control groups despite the
results from the Malmö study, means that their results are invalid and not
reliable.
Conclusion: There is no significant decrease in breastcancer
mortality in Sweden one decade after introduction of
mammography screening. Thus there is no scientific evidence, convincingly
showing that mass-screening by mammography is
effective in reducing breast cancer mortality. Neither is safety of the
method well documented and a re-evaluation is therefore urgently needed.
Göran Sjönell MD PhD FRCGP Family Physician
Kvartersakuten Matteus
Surbrunnsgatan 66 SE
11327 Stockholm
Sweden
Lars Ståhle MD PhD Assistant Professor
Department of Clinical Pharmacology
Karolinska Institute
Huddinge University Hospital
SE-14186 Huddinge
Sweden
References
1 Sjönell G, Ståhle L. Hälsokontroller med mammografi minskar inte
dödligheten i bröstcancer. (in Swedish) Läkartidningen 1999;96:904-13.
2 Rosén M, Rehnqvist N. No need to reconsider breast screening programme
on basis of results from defective study.BMJ 1999;318:809.
3 Sjönell G, Ståhle L. Hur länge skall vi behöva vänta på att se effekt på
dödligheten? Läkartidningen;96:1822-23.
4 Nyström L, Rutqvist LE, Wall S et al. Breast cancer screening with
mammography: overview of Swedish randomised trials. Lancet 1993;341:973-8
5 Andersson I, Aspegren U, Janzon L, Landberg T, Lindholm K et al.
Mammografic screening and mortality from breast
cancer: the Malmö mammografic screening trial. BMJ 1988;297:943-48.
6 Tabár L, Fagerberg CJG, Gad A, Baldertorp L, Holmberg LH et al.
Reduction in mortality from breast cancer after mass
screening with mammography. Lancet 1885;i:829-832.
7 Tabár L, Fagerberg G, Duffy SW, Day NE. The Swedish Two-County trial of
mammografic screening for breast cancer:
recent results and calculation of benefit. J Epidemiol. Common Health
1989;43:107-17.
8 Tabár L, Gad A. Screening for breast cancer; the Swedish trial.
Radiology 1981;138:219-22.
9 Tabár L, Gad A, Akerlund E et al. Screening for breast cancer in Sweden.
A randomized trial. (in) Logan WW, Muntz EP, eds Reduced dose Mammography.
Masson, 1979, pp 183-200.
Competing interests: No competing interests