The Abortion-Breast Cancer Hypothesis (again) – Understanding Risk.

Recently I published a series of five articles that looked in depth at the abortion-breast cancer (ABC) hypothesis and the complex relationship between women’s reproductive history and their short and long term risk of developing breast cancer.
Based on the research evidence I covered in those articles, we can summarise this relationship as follows:

– There is good evidence that a women’s risk of developing breast cancer is related to her exposure to ovarian hormones (endogenous oestrogen and progesterone). Increased exposure to these hormones, which stimulate cell growth, is associated with an increase risk of developing breast cancer.

– A number of reproductive factors can influence a woman’s duration and level of exposure to ovarian hormones, including; early onset of menstruation, late onset of menopause, later age at first pregnancy and never giving birth.

– Pregnancy and breastfeeding both reduce a woman’s lifetime number of menstrual cycles and, therefore, her cumulative exposure to these hormones. This may reduce a woman’s long-term risk of developing breast cancer, as long as they have no family history of breast cancer but for women who do have such a history, pregnancy may increase their risk of breast cancer.

– Pregnancy is also associated with a small, short-term increase in breast cancer risk, lasting up to ten years after the pregnancy, for all women, but the increase in risk is small and generally occurs at an age where the overall risk of developing breast cancer is also, thankfully, very small.

– Pregnancy and breastfeeding have a direct effect on breast cells, causing them to mature, so they can produce milk and some researchers have suggested that these mature breast cells are more resistant to the cellular changes that cause breast cancer than cells that have not matured.

– However, even allowing for the evidence that shows that pregnancy and breastfeeding may lower a woman’s risk of developing breast cancer, the best current evidence from prospectively designed epidemiological studies shows that women who have had an induced abortion are at no more risk of developing breast cancer than women who never had that pregnancy.

So, although the ABC hypothesis, which suggests that women may increase their risk of developing breast cancer by having abortions, because this deprive them of the protective effects of carrying a pregnancy to term and/or breastfeeding, is biologically plausible, it remains the case that the best current evidence from a meta-analysis of 13 prospectively designed studies and almost 40,000 women – Beral et al. (2004) – does not provide any evidence to support such claims.

It is nevertheless the case that a number of anti-abortion organisations continue to promote the ABC hypothesis as if it were a proven fact, prompting a number of organisations, including the British Humanist Association, to write to the Secretary of State for Education, Michael Gove, to complain that some organisations, notably the Society for the Prevention of the Unborn Child (SPUC), Lovewise and LIFE, have been actively promoting the ABC hypothesis, and other false claims about abortion, while giving school talks.

We are writing to express our concern about the false claims being made by groups invited to give lessons in schools on abortion and contraception, and to urge you to take action to prevent these claims being repeated. In particular, we are referring to the work of the Society for the Protection of Unborn Children (SPUC), Lovewise and Life.

To be clear, it is not the anti-abortion ideology of these groups that we are challenging. What prompts this letter are issues of fact: many of the claims these groups make are simply false. For example, there is no evidence that abortion can increase the risk of breast cancer or infertility; no evidence that hormonal contraception can cause an abortion; no evidence for a medical condition called ‘post abortion trauma’, or indeed that abortion causes more distress for women than carrying an unintended pregnancy to term.

SPUC and others have been repeatedly making these claims for a number of years. We believe it is time for the government to intervene. The Secretary of State for Education has the power to issue guidance on sex and relationships education to ensure that children are protected from inappropriate teaching materials and all state-funded schools must have regard to this guidance. This must surely include preventing materials that present false claims. We believe you should exercise your power accordingly and issue guidance to prevent lessons which can cause harm to young people – such harm would be caused if they are deterred from using contraception because they have been misinformed about its efficacy, or because they believe they will be infertile following abortion.

We therefore hope that you will make clear in future guidance on related matters that materials used in lessons which cover abortion and contraception must be based on fact insofar as they relate to medical and health matters. No group should be permitted to make claims for which there is no evidence.

One small quibble here – the issue here is not that these organisations have no evidence to support their claims, they do. The problem is not a lack of evidence but rather a lack of credible evidence. The evidence that these groups put forward in support of their claims is typically of pretty poor quality and has been carefully cherry-picked to support their ideological and religious beliefs while other, better quality evidence, is roundly ignored because it fails to support their arguments.

This might seem like a minor semantic point but it really is rather important. Amongst scientists and sceptics it’s generally understood that the phrase ‘no evidence’ is more or less synonymous with ‘no credible evidence’ or ‘no plausible evidence’, i.e. even if there is some evidence, I the form of published journal articles, the quality of that evidence is so poor that it can reasonably be disregarded. The wider public, most of whom are neither scientists nor sceptics, do not necessarily understand the phrase ‘no evidence’ in those same terms. If you say ‘no evidence’ then they tend to take you at you at your word and assume that there literally is no evidence, leaving you wide open to charges of bias and dishonesty when your opponents, be they anti-abortion organisation, homeopaths, chiropractors, anti-vaxxers or other quacks and conspiraloons come along, waving their piss-poor ‘evidence’ in the face of the public without any regard for the fact that the citations, testimonials and anecdotes they’re putting up as evidence are next to worthless.

And sure enough, that’s exactly what SPUC did in response to the BHA’s letter:

Dear Mr Copson (BHA) and Ms Hallgarten (EfC),

Recently a group of organisations, led by yourselves, wrote an open letter to Secretary of State for Education, Michael Gove, regarding the provision of information on abortion and contraception by organisations, including our own organisation, which give talks in schools. For your convenience we have reproduced the letter below:

[text of BHA letter]

It may be helpful for you and others to have a record of SPUC’s reply to these claims. We will begin with your claim that there is no evidence that abortion can increase the risk of breast cancer.

SPUC has indeed said in our schools talk that abortion “may” increase the risk of breast cancer. This claim is not even controversial with regard to the following:

a) Once pregnant, if a woman chooses to maintain her pregnancy and achieves a full term pregnancy, she will lower her risk of breast cancer. This is well known and undisputed in medical circles.

b) If she chooses to have an induced abortion she may remain childless, a condition which increases her risk of breast cancer

c) Or if she chooses abortion and then has another pregnancy, the abortion will have delayed this pregnancy, which delay also increases her risk of breast cancer.

d) If she already has a full term pregnancy and chooses to abort a subsequent pregnancy she loses the risk reduction that an additional full term pregnancy would have afforded her

e) The use of instruments such as dilators during an abortion increases the risk of having a premature delivery in future births. If that premature delivery is before 32 weeks, she will have an increased risk of breast cancer.

These uncontroversial facts, for which references are available, are not mentioned by the letter writers.

Each of these individual statements can be supported with published evidence drawn from credible journals, even allowing for the fact that there are some pretty big “if’s” involved, but does that mean that SPUC has a genuine argument, let alone the evidence to back it up?

The key issue here is risk and yet this is the one thing that SPUC doesn’t deign to define in detail. We’re told that each of these things may increase a woman’s risk of developing breast cancer but not by how much nor what the risk would be for someone who doesn’t have an abortion or, indeed, have any children at all, so we have no way of knowing whether this risk is genuine cause for concern or not.
When it comes to risk the figure that most people will have seen for breast cancer is the figure for lifetime risk, which is typically given as 1in 8 but what does that actually mean?

What many people think – incorrectly – is that this figure refers to the risk that a woman will develop breast cancer in her lifetime but this is only correct if that woman lives to at least 85 years of age, which many women don’t. Your actual risk of developing breast cancer, if you’re a woman, depends very much on your age both in terms of your annual risk, i.e. your risk of developing breast cancer in a particular year, and your cumulative risk, i.e. your risk of developing breast cancer by the time you reach a particular age, and the size of these risks is shown in the table below, together with the corresponding risks of actually dying of breast cancer, which is, of course, what some women really worry about.

AGE GROUP ANNUAL RISK OF BREAST CANCER CUMULATIVE RISK OF BREAST CANCER BREAST CANCER MORTALITY (ANNUAL) BREAST CANCER MORTALITY (CUMULATIVE)
1-4 Negligible Negligible Negligible Negligible
5-14 Negligible Negligible Negligible Negligible
15-24 1 in 2000 1 in 1000 Negligible Negligible
25-34 1 in 720 1 in 500 1 in 8000 Negligible
35-44 1 in 140 1 in 111 1 in 1000 1 in 1000
45-54 1 in 55 1 in 35 1 in 400 1 in 250
55-64 1 in 39 1 in 18 1 in 200 1 in 110
65-74 1 in 35 1 in 12 1 in 150 1 in 58
75-84 1 in 33 1 in 8 1 in 90 1 in 30

These figures give us a set of baseline risks to work with and we can use them gain a better understanding of the validity or otherwise of SPUC’s argument by carrying out a hypothetical experiment using these figures as a starting point.

The experiment we’re going to carry out is called a case control study in which we are going to take three groups of 1,000 women each of which is identical in every respect except for their reproductive histories.

One of our three groups, the control group, will consist of women who have carried two pregnancies to term before their 35th birthday but have never had an abortion and we’re going to assume that this group’s risk of developing breast cancer is the same as the baseline risks set out in the table above, so this group will gain the maximum protective effect from these pregnancies.

Our other two groups will consist of one group made up of women who have never carried a pregnancy to term and one consisting of women who terminated their first pregnancy but then went on to have two further pregnancies that they carried to term.

This leaves us only with the question of what is the increased risk of breast cancer associated with these two experimental groups and for that we’ll be using figures from a study which estimated that women who carry their first pregnancy to term have a 30% lower risk of developing breast cancer than women who never have children, with each subsequent pregnancy carried to term lowering the risk of developing breast cancer by a further 9% per pregnancy.

Based on this information, we’re going to follow the hypothetical women in each of these groups from their first to their 85th birthday, assuming they make it that far, and find out what happens each group in terms of the numbers of women developing and dying from breast cancer – and for added realism we’re also going to include columns for mortality from cardiovascular disease and from all other causes.

So, let’s start with our control group for which the figures look like this:

Age Group No. of Women in Group Breast Cancer Cases Deaths (Breast Cancer) Deaths (Heart Disease) Deaths (Other Causes) All Deaths
1-4 1000 0 0 0 1 1
5-14 999 0 0 0 1 1
15-24 998 1 0 0 4 4
25-34 994 1 0 1 5 6
35-44 988 7 1 3 9 13
45-54 975 19 3 10 18 31
55-64 944 27 5 21 37 63
65-74 881 33 8 43 92 143
75-84 738 42 16 95 207 318
At 85 420 130 33 173 374 580

So, by the time the surviving women in our control group reach the age of 85 we’ve seen 130 cases of breast cancer of which 33 (3.3 in 100) resulted in death, which means that breast cancer accounts for only 5.7% of all deaths amongst women in this group. By way of a comparison, 174 women didn’t make it all the way to 85 due to the biggest single killer, cardiovascular disease, while other contributed 374 women to our mortality figures. Women in this group are around eighteen time more likely to die of other causes before reaching the age of 85 than they are likely to die of breast cancer.

Our second group, the women who never had children are the group at greatest risk of developing breast cancer in our hypothetical study, and their figures look like this:

Age Group No. of Women in Group Breast Cancer Cases Deaths (Breast Cancer) Deaths (Heart Disease) Deaths (Other Causes) All Deaths
1-4 1000 0 0 0 1 1
5-14 999 0 0 0 1 1
15-24 998 1 0 0 4 4
25-34 994 2 0 1 5 6
35-44 988 10 1 3 9 13
45-54 975 26 4 10 18 32
55-64 943 38 7 21 37 65
65-74 878 46 11 43 92 146
75-84 732 58 22 95 207 324
At 85 408 181 45 173 374 592

Okay, so here we have what appears to be a fairly sizeable increase in the number of breast cancer cases, an additional 51 case over and above our control group, but this increase produces only an additional 12 deaths, raising the percentage of deaths due to breast cancer to 7.6%. These women are still around fourteen times more like to die of something else before they reach 85 than they are breast cancer.

Finally, we have our women who terminated their first pregnancy but then went to have two children, so they get some protective effect albeit not quite as much as the women who carried their first pregnancy to term and for this group the figures look like this:

Age Group No. of Women in Group Breast Cancer Cases Deaths (Breast Cancer) Deaths (Heart Disease) Deaths (Other Causes) All Deaths
1-4 1000 0 0 0 1 1
5-14 999 0 0 0 1 1
15-24 998 1 0 0 4 4
25-34 994 2 0 1 5 6
35-44 988 9 1 3 9 13
45-54 975 22 3 10 18 31
55-64 944 32 6 21 37 64
65-74 880 39 9 43 92 144
75-84 736 49 19 95 207 321
At 85 415 154 38 173 374 585

As you might well expect, this groups falls somewhere in between the other two with 24 additional cases of breast cancer over 85 years and five additional deaths, an increase in women’s annual risk of developing breast cancer of a little under 0.3% per year.

If we translate this increase back into our original table of baseline risks then our figures for women who terminate their first pregnancy but then have two children, which is not going to be that uncommon a scenario, will look like this.

AGE GROUP ANNUAL RISK OF BREAST CANCER CUMULATIVE RISK OF BREAST CANCER BREAST CANCER MORTALITY (ANNUAL) BREAST CANCER MORTALITY (CUMULATIVE)
1-4 Negligible Negligible Negligible Negligible
5-14 Negligible Negligible Negligible Negligible
15-24 1 in 1950 1 in 1000 Negligible Negligible
25-34 1 in 700 1 in 333 1 in 8000 Negligible
35-44 1 in 136 1 in 83 1 in 1000 1 in 1000
45-54 1 in 53 1 in 29 1 in 333 1 in 250
55-64 1 in 38 1 in 15 1 in 166 1 in 100
65-74 1 in 34 1 in 9 1 in 111 1 in 52
75-84 1 in 32 1 in 7 1 in 51 1 in 26

So, yes, the risk increases… but not by very much.

This is all, of course, hypothetical and based on treatment women as statistical abstractions. Real life, and real women, are nothing like so straightforward and yet the meta-analysis which provides the best current evidence relating to abortion and breast cancer – Beral et al (2004) – did something very similar to this when they analysed the data from more than 39,000 real women. The researchers did, as far as they possibly could, control for factors such as age, number of pregnancies carried to term and a range of other potential confounding factors in order to ensure that when they compared the prevalence of breast cancer in women who had terminated pregnancy with women who hadn’t they were making as close to a like-for-like for comparison as they possibly could, and that study found no evidence for an increased risk of breast cancer in women who had had an abortion at some point in their lives.

So why did the apparent preventative effect of pregnancy and breastfeeding we explored in our hypothetical case control study not show up when Beral et al. looked at the actual data for real women?

Well, for the simple reason that pregnancy is just one of several factors which can affect and alter an individual woman’s risk of developing breast cancer. Age is by far the most significant risk factor, if you discount women who have already had an earlier cancer that has been successfully treated, but other factors can, and do, come into plat including alcohol intake, obesity, hormone replacement therapy, radiation exposure (from medical X-rays, etc.) and other clinical and genetic factors. The increased risk associated with not carrying a first pregnancy to term, whether that’s due to an induced or spontaneous abortion or because a woman never falls pregnant, is about the same as the increased risk associated with drinking two alcoholic drinks a day when compared to a teetotaller. It’s also around the same as risk associated with obesity, with taking HRT for five years, with a woman having her menstrual period before the age of 12, which is becoming increasingly common in Western Europe and North America, and with the late onset of the menopause.

It is not a simple relationship and the effect sizes for all these risks are small, so it should come as no surprise at all that when we look at the data for real women living real lives we discover that that data does not support the ABC hypothesis.

If you are going to go into schools and talk to young women about the risks associated sexual intercourse, teenage pregnancy, abortion and everything else that fall under the general heading of sex and relationships education then you have clear moral duty to tell them the truth, based on the best available evidence, and give them the facts – all the facts, not a carefully cherry-picked selection made up of only the evidence which appears to support your own preconceived religious and ideological beliefs. When it comes to dealing with risk, women have the right to be told not only whether or not a particular risk exists but also to be given clear information that both quantifies the size of that risk and also places it in its proper context. When we’re talking about the relationship between a woman’s reproductive history and her long-term risk of breast cancer we are talking about a small increase in long-term risk which, for teenagers, is not going to kick-in to any significant extent for at least 20-30 years, and maybe even more, and that risk may easily be mediated by a range of other factors in the meantime to the extent that, for many women, it will never materialise.

Against that, there are much more immediate risks that women also need to be aware of and not clinical risks (STDs, cervical cancer, miscarriage and complications during and after pregnancy) but also a range of other risks. Numerous studies have shown that early motherhood, which groups like SPUC advocate for women who do fall pregnant early in their adult life, is associated with poorer mental health outcomes, lower levels of educational attainment, lower rates of workforce participation, lower incomes and higher rates of welfare dependency and although this is something of a ‘chicken and egg’ association – because women who are already at a higher risk of these factors are also more likely to fall pregnant and have children in their teens anyway – what is clear from this evidence is that early motherhood can, and does, lock women into a cycle of long-term socio-economic deprivation from which they find it very difficult to break free, more so than women who start out with the same disadvantages in life but who avoid early motherhood. In these studies, controlling for key sources of confounding reduces the association with poorer mental health outcomes to the point where it ceases to be statistically significant but the associations with poverty and lower educational attainment remain significant after introducing the same controls.

SPUC’s preferred alternative for young women who do fall pregnant but either cannot or do not want to bring up a child is, of course, adoption and although its not been reported quite what, if anything, the organisation has to say about adoption in its school talks it’s a fair bet that it doesn’t include this information, which comes from a 2004 literature review by Elspeth Neil of the University of East Anglia:

What do we know about birth relative’s experiences?

Most of what is known about the birth relatives of adopted children relates to a very different population of people to those whose children are adopted today. Almost all research has been carried out on mothers who relinquished their children for adoption. The research on relatives other than mothers, and on the experiences of people who did not choose to have their child adopted, is very limited. Below, the relevant research is reviewed and summarised.

Relinquishing a child for adoption is an experience of loss that can have long term negative psychological consequences

A number of studies of relinquishing birthmothers have found that having a child adopted is an experience of loss and grief that persists beyond the immediate aftermath of the parting, and in many cases is long term. Winkler and Van Keppel (1984) studied 213 women who had all relinquished a child for adoption when they were young and single. A great sense of loss was a key feature of many women’s stories and the greater the sense of loss reported by the women, the worse was their adjustment. For many women this sense of loss did not diminish with time, in fact 48% of the sample reported that it had intensified and was worse at particular times such as birthdays and Mother’s Day. For some women a strong sense of loss had persisted for up to 30 years. Well over half of respondents rated the adoption of their child as the most stressful experience of their life. The psychological functioning of the birthmothers was also measured and was found to be significantly worse than a matched sample of women who had not had a child adopted. This research clearly shows that it is unrealistic to make the assumption that women whose children are adopted will quickly ‘get over’ this experience. In many cases the negative consequences are serious and long lasting. Many similar findings have been outlined by other researchers both in this country (e.g. Bouchier et al, 1991; Howe et al, 1992; Hughes and Logan, 1993; Logan, 1996; Wells, 1994) and abroad (e.g. Condon, 1986; Deykin et al, 1984; Rockel and Ryburn, 1988) and key themes are obvious in biographical accounts (e.g. Powell and Warren, 1997).
As Neil notes, the evidence on adoption shows that it is “unrealistic to make the assumption that women whose children are adopted will quickly ‘get over’ this experience” and yet this appears to be prevailing assumption amongst anti-abortion organisations that promote adoption as an alternative to abortion.

The inclusion of false claims is only part of the problem with allowing anti-abortion organisations to go into schools and peddle their biased wares to young women, their misleading use of cherry-picked evidence, failure to an accurate and honest explanation of the scale and context of the risk they are talking about and the omission of other key informatio inclusing, one suspects, anything at all about the long-term risks associated with giving a child up for adoption are equally serious flaws in their wholly biased and misleading presentation.

  • Pingback: Unity()

  • Pingback: Tim Ireland()

  • Pingback: Education For Choice()

  • Pingback: Education For Choice()

  • Pingback: British Humanists()

  • Pingback: AbortionSupportNtwk()

  • Pingback: Abortion Rights()

  • Pingback: Jess()

  • Anthonymccarthy

    Your link to the SPUC letter doesn’t work, which is unfortunate as you do not reproduce the letter in full. Those who follow the link here http://www.spuc.org.uk/documents/papers/humanists20120525 will see that you do not address most of what is said in the letter (although you do give an airing to the false allegations of the BHA).

    On the question of risk – the papers cited in the letter go into a lot of detail on this and contextualise risks. And of course, women who are already at higher risk of a certain condition need to know about avoidable additional risks (so what you claim is ‘negligible’ ain’t necessarily so). As for ‘wholly biased and misleading presentation’ : you have never actually heard/seen the SPUC presentation. Nor have you shown that it misleads anyone about anything. Indeed, you have failed to do so.

    I look forward to your giving your readers ‘all the facts’ with regard to the abortion industry: an industry you appear to defend unflinchingly no matter what it does. On the philosophical arguments against abortion, you might want to engage with this: http://www9.georgetown.edu/faculty/ap85/papers/IWasAFetus-short.html

    As for your defence of infanticide, I refer you to the work of David Oderberg and others on the philosophical incompetence and dishonesty of Peter Singer – a writer who, as a cheerleader for both abortion and infanticide, shares many of your vices.Anthony McCarthy