Should we use total mortality rather than cancer specific mortality to judge cancer screening programmes? YesBMJ 2011; 343 doi: https://doi.org/10.1136/bmj.d6395 (Published 13 October 2011) Cite this as: BMJ 2011;343:d6395
All rapid responses
Re: Should we use total mortality rather than cancer specific mortality to judge cancer screening programmes? Yes
Screening programs for cancer are thoughtless, unproved and dangerous.
Introducing screening for breast cancer before showing it works, typifies the Alice in Wonderland world of contemporary medicine; and so does the attempt to find an answer by debate (and fashionable online opinion poll – albeit without prizes for the “best” answer)1, rather than scientific proof.
The “debate” limits itself to the question of whether measurement of total or cancer specific mortality best gives the efficacy of breast screening, but the answer is neither: both are irrelevant and unusable, and for two reasons:
1. Patients diagnosed by presentation can never be matched for comparison of outcome with those discovered by screening. 2. Both “tests” use the outcome of mortality, but this isn’t what is being measured: screening nets disease before it becomes apparent; the mortality of disease found in this way is a separate matter.
Judgement about screening can only concern the efficacy the trawl; judgement about outcome of disease caught in this way must be established separately.
The poor efficacy of screening is already known; lesions filtered in this way have, at best, a 10% chance of being malignant. Therefore, screening for breast cancer is a very poor “test” and should not be offered; to present it to patients as a choice to be made by them, is a dereliction of our professional duty. There is no need for the further measures under debate.
However, apart from cost and patient anxiety, screening has the additional, serious defect of increasing the number of false diagnoses. The inevitability of this artefact, and its magnitude, arises arithmetically from the ratio of the high uncertainty of diagnosis of early disease, to the low population prevalence of the true disease, as I first showed for melanoma2. Screening will always produce spurious disease when the diagnostic error is high and the prevalence low, as is the case with melanoma and breast cancer.
The basis of screening is the not unreasonable belief that early detection, and therefore treatment, will decrease mortality; but the case has been made more by assumption than proof. And since there will be an inverse relationship between outcome and intrinsic tumour aggressivity, the benefit of early diagnosis, and therefore screening, will be similarly limited. Thus the magnitude of the advantage of early definition needs to be established for different tumours, instead of assuming it will inevitably merit the effort. In the case of malignant melanoma, the encouragement of early presentation of assumed early lesions, which is virtually a screening program, has served only to produce spurious diagnoses of malignancy, without any improvement in overall mortality3.
Screening procedures, such as those for breast cancer and melanoma, have been introduced prematurely, without prior definition of need, success and consequence, and without proper consideration of risk, in particular the creation of spurious disease and unnecessary patient anxiety.
We’ve had enough fashionable “debate”; it’s time for some unfashionable, but more reliable, science.
Emeritus Professor Sam Shuster
1. Penston J, Steele R, Brewster D. Should we use total mortality rather than specific mortality to judge cancer screening programs? BMJ;2011;343: 938-999.
2. Shuster S. Malignant melanoma: how error amplification by screening creates spurious disease. Brit J Derm; 2009; 161: 977-979.
3. Levell NJ, Beattie CC, Shuster S, Greenberg DC. Melanoma epidemic: a midsummer night’s dream? Br J Dermatol 2009; 161:630–4
Competing interests: none
James Penston is correct that "All cause mortality is a hard end
point that is free from bias, produces a robust estimate of the effect,
and answers the crucial question of whether cancer screening improves
overall survival (1)," and that it should therefore be the primary outcome
in cancer screening trials.
The mammography screening trials illustrate why we cannot rely on
cancer specific mortality. Not only has an effect on total mortality not
been demonstrated (2), it is also highly unlikely to exist (3,4).
The trials also failed to find an effect on all-cancer mortality (2).
This is one of the most important findings in our Cochrane review of
screening (2), but it has been completely ignored by screening advocates,
likely because it is so threatening to their beliefs.
We write in the Cochrane review that the major difficulty in
assessing cause of death in the trials might have occurred when the
patients were diagnosed with more than one malignant disease (2). The
importance of autopsy is illustrated by the fact that 21% of the women
with breast cancer who died in the Malm? trial had two or three types of
different cancers (2). Patients with cachexia and no signs of recurrence
of breast cancer would likely be assigned to another type of cancer.
Since cancer mortality is likely to be less subject to bias than
breast cancer mortality, we calculated what the expected cancer mortality
(including breast cancer mortality) would be if the reported reduction in
breast cancer mortality of 29% after seven years for the suboptimally
randomised trials were true. We found an expected relative risk of 0.95,
but cancer mortality in these trials was not reduced (RR 1.00, 95%CI 0.96
to 1.05), and this estimate was significantly higher than what was
expected (P = 0.02). This provides further evidence that assessment of
cause of death was biased in favour of screening. We also found that data
from the Two-County trial illustrated the misclassification directly (2).
The effect of mammography screening has never been anywhere near the
30% we have so often heard about (3,4). It was likely only 12% in the old
randomized trials, as judged by the difference in tumour size between the
screened and the control groups (4).
1 Penston J. Should we use total mortality rather than cancer
specific mortality to judge cancer screening programmes? Yes. BMJ
2 Gotzsche PC, Nielsen M. Screening for breast cancer with
mammography. Cochrane Database Syst Rev 2009;1:CD001877.
3 Jorgensen KJ, Keen JD, Gotzsche PC. Is mammographic screening
justifiable considering its substantial overdiagnosis rate and minor
effect on mortality? Radiology 2011;260:621-6.
4 Gotzsche PC, Jorgensen KJ, Zahl PH, et al. Why mammography
screening hasn't lived up to expectations from the randomised trials.
Cancer Causes & Control 2011 (in press).
Competing interests: No competing interests