Putting the risk of breast cancer in perspective

Full text

Education and debate
Putting the risk of breast cancer in perspective
John P Bunker, Joan Houghton,Michael Baum
Patients are frequently given the opportunity to
participate in making decisions about their care. To
assist them in making their decisions data that are as
accurate and as complete as appropriate will need to
be quickly available. These data will be needed not
only by patients but also by their doctors. A life table
constructed from regularly published statistics on
national morbidity and mortality can be used to
display the likelihood of developing or dying of a
disease (a cumulative probability) at any given age.
Such a life table may be a valuable tool in helping
patients to make decisions about their care. In this
paper we discuss the use of life tables to present to
patients the absolute and relative risks of breast cancer
and their use in comparing the risks from breast
cancer with those from other life threatening
conditions.
Cumulative incidence
Recently, notices in London’s underground warned
that women havea1in12riskofdeveloping breast
cancer. Women in America have been warned that
their risk is 1 in 8; consequently, “fear of breast cancer
is so pervasive among US women that it is causing
them to ignore far more serious health threats.”1For
women who manage to escape other, greater health
risks and survive to age 75 or 80, risks of 1 in 12 and 1
in 8 are approximately correct. (The greater risk in
America may reflect the earlier introduction of mass
screening, the fact that women are screened at younger
ages, and the overdiagnosis of tumours of low grade
malignancy or of ductal carcinoma in situ
—
conditions
that might never become clinically detectable or life
threatening.23
)
More relevant to the concerns of younger women is
the risk of developing breast cancer earlier in their
lives. A life table analysis of the cumulative incidence of
breast cancer in England and Wales shows that the risk
for women under age 35 is 1 in 625. This rises to 1 in
56 by age 50, to 1 in 18 by age 65, and to 1 in 13 by age
75.4This type of age specific data should offset dispro-
portionate fear of the disease among women but it falls
short of meeting the information needs of individuals.
Life table estimates based on the entire population are
averages; the risk for an individual varies depending on
the presence or absence of predisposing risk factors
such as a family history of the disease, age at menarche,
age at first pregnancy, age at menopause, whether oral
contraceptives have been used, and the amount of
alcohol consumed. In postmenopausal women risk
factors include the use of hormone replacement
therapy and obesity in women not receiving hormone
replacement therapy.
Coordinating absolute and relative risks
The quantitative significance of risk factors has been
poorly presented to both the public and the medical
profession. In 1996, many women were concerned
after publication of a review of the association between
breast cancer and hormonal contraceptives which
stated that the risk of breast cancer was increased by
20% among women who used oral contraceptives.5If
the risk had been presented as a cumulative increase in
absolute risk from 16 to 18.7 per 10 000 women for a
woman aged between 30 and 35, which was how it was
presented in the review (table 15), then different assess-
ments of personal risk might have occurred. This
increase in absolute risk represents one additional case
of breast cancer occurring annually (excess cumulative
incidence) among 3700 women taking oral contracep-
tives; this would increase to one additional case occur-
ring among 2100 women as a woman became five
years older. For a woman aged 30 to 35 this could be
Summary points
A life table constructed from national statistics on
morbidity and mortality can be a valuable
resource in helping patients to make decisions
about their care
Coordinating absolute and relative risks in a life
table offers patients and doctors the opportunity
to make more informed judgments
Life table analysis allows patients and doctors to
estimate the cumulative probability of dying of
breast cancer before a patient reaches a given age
The risk of developing or dying of breast cancer
should be seen in the context of other life
threatening conditions
For women who smoke the cumulative probability
of dying of lung cancer matches that of dying of
breast cancer when women reach their early 50s;
this probability doubles by age 65 and triples by
age 75
Cancer Research
Campaign and
UCL Cancer Trials
Centre, University
College London
Medical School,
London W1P 8AN
John P Bunker,
visiting professor
Joan Houghton,
senior lecturer,
department of
oncology
Department of
Surgery, University
College London
Middlesex Hospitals
Trust, London
W1P 7LD
Michael Baum,
professor of surgery
Correspondence to:
Dr Bunker
BMJ 1998;317:1307–9
1307BMJ VOLUME 317 7 NOVEMBER 1998 www.bmj.com

compared to a 40% increase in the risk of developing
breast cancer if she consumed the equivalent of “24 g
(1 oz) of absolute alcohol daily (about two drinks
daily).”6This risk would convert to an excess cumulative
incidence of one additional case of breast cancer
occurring among about 1500 women.
The benefits and risks of treatment are perceived
differently by health professionals78 and the public
depending on whether they are presented as absolute
or relative improvements.Members of a health author-
ity were presented with the results of a large Swedish
trial in which a relative risk reduction of 34% was
reported and asked if they would vote to fund
mammographic screening.89 The majority said that
they would favour funding the screening. When the
same results were disguised as a different trial and pre-
sented as an absolute risk reduction of 0.06%, the
majority indicated that they would vote against funding
screening. When presented with 1592 patients as the
number needed to treat to save one life, half of those
surveyed indicated that they would be in favour of
screening. Presenting these data in a life table similar to
table 1 would have offered a better opportunity for the
members of the health authority to make a more
informed judgment.
Cumulative mortality
A woman probably wants to know her risk of develop-
ing breast cancer at her age and with her risk factors;
she may have a better appreciation of the risk if the
numbers can be seen as absolute as well as relative
risks. But this is not enough. A woman will also want to
know her prognosis and, in particular, the probability
that she will live or die. She might also want to know
how such a probability compares with that of dying of
other life threatening conditions.
Analysing death rates may lead to some surprises
for doctors and their patients. The first surprise is that
there is an increasingly large discrepancy between the
incidence of and mortality from breast cancer in
America; since 1940 mortality has remained constant
while incidence has nearly doubled.2This discrepancy
can only partially be accounted for by the growing
success of treatment. Thus, the woman concerned
about the possibility of developing breast cancer in the
future should know that the likelihood of subse-
quently dying of breast cancer is relatively small and is
becoming smaller each year. A life table analysis based
on 1995 mortality statistics for England and Wales10
allows us to estimate the cumulative probability of a
patient dying of breast cancer before she reaches a
given age.
Mortality from breast cancer among women in dif-
ferent age groups was derived from data from the
Office for National Statistics (table 2).10 The data were
adjusted for the number of women surviving in each
age group. The number of women dying from all
causes in each age group was calculated from the
resulting life table, which was based on mortality in the
female population. The cumulative number of deaths
from breast cancer is found by adding the number of
deaths from the disease occurring at each age interval.
The per cent probability of dying from breast cancer
before the end of each age interval is found by dividing
the number in the cumulative risk column by 1000. In
England and Wales before the age of 50 only one
woman out of 136 dies of breast cancer. By the age of
60 this is one out of 65, by the age of 70 it is one out of
39, and by the age of 80 only one woman out of 26 dies
of breast cancer (table 2).
The risk of dying of breast cancer should be seen
in the context of other risks to a woman’s life. The risk
of dying of heart disease has been underestimated by
women and its threat is perceived to be low.11 Women
may believe that deaths from cardiac causes primarily
affect men and elderly women whereas, at least in
smokers, the number of deaths occurring in women as
a result of cardiac causes exceeds the number of
deaths from breast cancer at all ages (figure).11 Fearing
death from breast cancer more than death from heart
disease reflects not simply an ignorance of probabili-
ties. “Heart disease, which offers at least the possibility
of sudden death, may not frighten people quite as
much [as cancer]: for many, for perhaps irrational
reasons, it is preferred. That certainly is the conclusion
from responses to the question ‘How would you like
to die?’”1
Death from heart disease may be preferred by some
to death from breast cancer; death from lung cancer,
certainly as unpleasant as that from breast cancer, offers
a more apposite comparison. For women who smoke
Table 1 Estimated number of cases of breast cancer occurring in Europe or North
America in women who never used combined oral contraceptives and in women who
used them from age 25 to 29.5Used with permission from Elsevier Science
Age at
diagnosis
Breast cancers diagnosed in
women who never used
combined oral contraceptives
(n=10 000)
Breast cancers diagnosed in women who used
combined oral contraceptives from age 25 to 29
(n=10 000)
5 year
incidence*
Cumulative
incidence
Relative
risk
5 year
incidence
Cumulative
incidence
Excess
cumulative
incidence (SD)
Using estimates of relative risk for all users:
20-24 0.5 0.5 1 0.5 0.5 0
25-29 3.5 4 1.24 4.3 4.8 0.8 (0.1)
30-34 12 16 1.15 13.9 18.7 2.7 (0.5)
35-39 28 44 1.07 30.0 48.7 4.7 (1.0)
40-44 56 100 0.98 55.1 103.7 3.7 (2.0)
45-49 80 180 1.01 80.8 184.5 4.5 (3.6)
Using estimates of relative risk for users with total duration of use of oral contraceptives >1 year:
20-24 0.5 0.5 1 0.5 0.5 0
25-29 3.5 4 1.22 4.3 4.8 0.8 (0.1)
30-34 12 16 1.15 13.8 18.6 2.6 (0.5)
35-39 28 44 1.06 29.6 48.2 4.2 (1.0)
40-44 56 100 0.96 53.8 102.0 2.0 (2.0)
45-49 80 180 0.98 78.5 180.5 0.5 (3.9)
*Annual incidences per 100 000 never users were assumed to be 160 at ages 45-49 and 0.007(age−17) at
ages 20-44; these rates are intermediate between rates in the United Kingdom and the United States in the
mid-1980s.
Table 2 Cumulative probability of death from breast cancer in England and Wales in
1995 per 100 000 women. Mortality was adjusted for the number of women surviving
in each age group. Data derived from Office for National Statistics10
Age interval
No of women dying of
breast cancer in each age
interval/100 000 women
Cumulative No of deaths from breast
cancer occurring by the end of each
age interval/100 000 women
Probability of dying of
breast cancer by end
of each age interval
25-34 34.6 34.6 1/2873
35-44 176.3 210.9 1/474
45-54 522.5 733.4 1/136
55-64 794.7 1528.1 1/65
65-74 1060.4 2588.0 1/39
75-84 1189.5 3778.0 1/26
Education and debate
1308 BMJ VOLUME 317 7 NOVEMBER 1998 www.bmj.com

the cumulative probability of dying of lung cancer
matches that of dying of breast cancer when women
reach their early 50s; this probability doubles by age 65
and triples by age 75 (figure). Although there has been a
modest fall in the number of women who smoke
(mainly among older women), there is little evidence
that the fear of developing lung cancer matches the fear
of developing breast cancer. Ironically, lung cancer has a
cure rate of < 5% and can be almost entirely prevented
by avoiding tobacco but, on average, 70% of patients
treated for breast cancer can expect to survive for 10
years. In contrast to lung cancer there is comparatively
little that can be done to prevent breast cancer.
Conclusion
The statistic that 1 in 12 women will develop breast
cancer is thus correct only for women who have
escaped a number of equally serious but more likely
threats to life at an earlier age. For most women the
lifetime risk of dying of breast cancer is only 1 in 26;
the other 25 women will die of something else. Life
table analyses show that the incidence of breast cancer
and mortality from the disease are much lower among
younger women and these risks should be understood
in the context of other serious threats to life.
1 Assessing the odds [editor ial].Lancet 1997;350:1563.
2 Har ris JR, Lippman ME, Veronesi U,Willett W. Breast cancer. New Engl J
Med 1992;327:319-28.
3 Feuer EJ, Wun L-M. How much of the recent rise in breast cancer
incidence can be explained by increases in mammography utilization? A
dynamic population model approach. Am J Epidemiol 1992;136:1423-36.
4 Office for National Statistics. 1991 cancer statistics registrations. London:
Stationery Office, 1997. (Series MB1, No 24.)
5 Collaborative Group on Hormonal Factors in Breast Cancer. Breast can-
cer and hormonal contraceptives: further results. Contraception
1996;54(suppl 3):1-106S.
6 Longnecker MP, Berlin JA, Orza MJ, Chalmers TC. A meta-analysis of
alcohol consumption in relation to risk of breast cancer. JAMA
1988;260:652-6.
7 Fahey T, Griffiths S, Peters TJ. Evidence based purchasing: understanding
results of clinical trials and systematic reviews. BMJ 1995;311:1056-60.
8 McColl A, Smith H, White P,Field J. General practitioner s’ perceptions of
the route to evidence based medicine: a questionnaire survey. BMJ
1998;316:361-5.
9 Tabar L, Fagerberg CJG, Gad A, Baldetorp L, Holmberg LH, Grontoft O,
et al. Reduction in mortality from breast cancer after mass screening with
mammography: randomised trial from the Breast Cancer Screening
Working Group of the Swedish National Board of Health and Welfare.
Lancet 1985;i:829-32.
10 Office for National Statistics. 1995 mortality statistics: cause. London:
Stationery Office, 1997. (Series DH2, No 22.)
11 Pilote L, Hlatky MA. Attitudes of women toward hormone therapy and
prevention of heart disease. Am Heart J 1995;129:1237-8.
12 Peto R,Lopez AD, Boreham J, Thun M, Heath C. Mortality from tobacco
in developed countries: indirect estimation from national vital statistics.
Lancet 1992;339:1268-78.
(Accepted 26 June 1998)
Confidence intervals for the number needed to treat
Douglas G Altman
The number needed to treat (NNT) is a useful way of
reporting the results of randomised controlled trials.1
In a trial comparing a new treatment with a standard
one, the number needed to treat is the estimated
number of patients who need to be treated with the
new treatment rather than the standard treatment for
one additional patient to benefit. It can be obtained for
any trial that has reported a binary outcome.
Trials with binary end points yield a proportion of
patients in each group with the outcome of interest.
When the outcome event is an adverse one, the differ-
ence between the proportions with the outcome in the
new treatment (pN) and standard treatment (pS) groups
is called the absolute risk reduction (ARR = pN−pS).
The number needed to treat is simply the reciprocal of
the absolute risk difference, or 1/ARR (or 100/ARR if
percentages are used rather than proportions). A large
treatment effect, in the absolute scale, leads to a small
number needed to treat. A treatment that will lead to
one saved life for every 10 patients treated is clearly
better than a competing treatment that saves one life
for every 50 treated. Note that when there is no
treatment effect the absolute risk reduction is zero and
the number needed to treat is infinite. As we will see
below,this causes problems.
As with other estimates, it is important that the
uncertainty in the estimated number needed to treat is
accompanied by a confidence interval. A confidence
Age at death
No of deaths
706050403020 80
0
10 000
15 000
20 000
25 000
30 000
35 000
5000
Breast cancer
Lung cancer, smokers
Heart disease, smokers
Heart disease, non-smokers
Cumulative number of deaths in 1995 from breast cancer, lung
cancer in smokers, heart disease in smokers, and heart disease in
non-smokers per 100 000 women in England and Wales. A risk ratio
of 12.5:1 for lung cancer and 2.3:1 for heart disease for smokers v
non-smokers was assumed12
Summary points
The number needed to treat is a useful way of
reporting results of randomised clinical trials
When the difference between the two treatments
is not statistically significant, the confidence
interval for the number needed to treat is difficult
to describe
Sensible confidence intervals can always be
constructed for the number needed to treat
Confidence intervals should be quoted whenever
a number needed to treat value is given
Education and debate
Imperial Cancer
Research Fund
Medical Statistics
Group, Centre for
Statistics in
Medicine, Institute
of Health Sciences,
Oxford OX3 7LF
Douglas G Altman,
professor of statistics
in medicine
d.altman@icrf.
icnet.uk
BMJ 1998;317:1309–12
1309BMJ VOLUME 317 7 NOVEMBER 1998 www.bmj.com