The incidence of breast cancer is
increasing. There are several reasons
for actuaries to attend to this
The increase in incidence is considerable,
averaging over 80% across all ages since the
1970s when registration was set up in Great
Britain. Figure 1 shows how the increases are
small at ages 4044, large at ages over 50 (5054
and 5559), and intermediate at ages 4549.
The effect of the introduction of breast cancer
screening for women aged 5065 around 1989
is clearly apparent from the surge in rates at that
time. But increases after 1993 cannot be attributed
to screening. Though not many men get
breast cancer and quite a number of women get
lung cancer, breast cancer has overtaken lung
cancer to be our most common cancer. The
increase is concentrated in age groups over the
age of 50, where women are now claiming a
higher proportion of the higher paid posts and
professional opportunities. In these age groups,
5065, the incidence has more than doubled.
Of particular interest to actuaries in health
insurance and life insurance is the social gradient;
breast cancer is unlike other cancers in that
the incidence is greater among upper-class
women. This reverse or negative social gradient
has been found in all the European countries
where it has been investigated. Figure 2 shows
how this gradient has increased in England and
Wales, and is expected by the author to increase
further. Whereas the figure shows the mortality gradient, the incidence gradient is steeper.
Treatment has improved so increased morbidity
accompanies more stable mortality. The
traditional pattern of higher female morbidity
and lower female mortality is being accentuated
on account of breast cancer trends.
Risk factors driving the trends
Most of the known risk factors are reproductive,
pregnancy related or hormonal. There is hormonal
activity and breast cell development
during pregnancy. Induced abortion has a carcinogenic
effect that is greater when the woman
is nulliparous (no previous full-term pregnancy)
by leaving the breast cells in a state of interrupted
hormonal development where they are
more susceptible to cancer. Full-term pregnancies
leave breast cells more fully developed and
resistant to breast cancer. Breast-feeding confers
additional protection. Oestrogen, progestin and
other female hormones, whether naturally produced
or administered medically, fuel breast
Seven known reproductive risk factors could
be driving the trends:
u Abortion Most British abortions (53%) are
nulliparous and the subsequent breast cancer
risk is greater here.
u Age at first birth A low age is protective, as
made known by British epidemiologists.
u Childlessness increases the risk Nuns have
long been known to have a higher risk of
u Fertility More children increase protection.
u Breast-feeding This gives additional protection,
now estimated by British epidemiologists.
u Hormonal contraceptives These contain
oestrogen and progestin, and are conducive
to breast cancer.
u Hormone replacement therapy (HRT) This
contains female hormones and is likewise
conducive to breast cancer.
The challenge and opportunity
The technical challenge of assessing the risks of
breast cancer is perhaps attractive to actuaries
in that the risk factors take effect over a long
term. The average age is over 60 when breast
cancer is diagnosed, but under 30 for abortions
and live births. And there is the potential to use
the long time lags to make long-term forecasts
of cancer trends.
For England and Wales there are especially
useful national statistics. Cancer registration has
produced age-specific breast cancer incidence
rates for female malignant breast cancers and
carcinomas in situ since the 1970s. Cumulated
cohort rates for breast cancer incidence within
any age range can then be computed. Abortion
statistics since 1968 are comprehensive and age
specific. Demographic data on fertility provides
average age at first birth and childlessness for
successive birth cohorts of women besides completed
cohort fertility. Correlational analysis of
birth cohorts of English women shows that
completed cohort cumulated rates of induced
abortion and fertility are the best predictors of
breast cancer trends, and regression modelling
for birth cohorts of women with these two
explanatory variables has produced reasonable
forecasts. Figure 3 shows the parallel increases in completed cohort abortion rates and breast
cancer incidence within the ages 5054.
Some 1997-based forecasts published by the
International Congress of Actuaries (ICA) in
Cancun in 2002 have performed quite well in
the years 19982004. (Reference 1) The higher
forecast, allowing some weighting for the additionally
carcinogenic effects of nulliparous
abortions, was for increases of 2.2% pa in incidence
rates in England and Wales, and increases
of 1.37% in Scotland. Numbers of new malignant
breast cancers forecast are compared with
observed incidence in table 1. The combined
numbers of cases malignant and in situ are
shown in table 2. Table 3 shows the numbers of
new malignant breast cancer cases for Scotland.
New forecasts, incorporating data from 1998
to 2004 with 2004 as the base year, have now
been carried out to estimate new breast cancer
numbers in the next 25 years. Published in the
Journal of American Physicians and Surgeons
(JPANDS), these are summarised in table 4.
The paper in JPANDS also gives forecasts for
Northern Ireland, the Irish Republic, Sweden,
the Czech Republic, Finland and Denmark.
Details of modelling, model fitting and analysis
are given in the ICA and JPANDS papers.
The breast cancer epidemic is amenable to
epidemiological investigation: models can be
fitted to cohorts of women; forecast numbers
can be used to plan treatment facilities so that
women diagnosed can be treated without delay;
and the forecast rates can be used to calculate
insurance premiums and reserves.
(1) Carroll, P (2002) ‘Pregnancy related risk
factors in female breast cancer incidence’, Intl
Congress of Actuaries, Transactions, vol 4, 331375.
(2) Carroll Patrick S (2007) ‘The breast cancer
epidemic: modeling and forecasts based on
abortion and other risk factors’, Journal of
American Physicians and Surgeons, Vol 12, No 3,