What are the major studies?
Does abortion improve mental health?
Does abortion harm mental health?
Is it because of the stigma?
“Women having abortions have worse mental health to begin with”
Is the link causal?
“95% of women having abortions feel they made the right decision”
“Women suffer emotionally if they don’t get abortions”
What about the APA review?
What about the NCCMH review?
What about Gilchrist’s study?
What did Fergusson’s meta-analysis show?
What about the Turnaway study?
What about Steinberg’s Lancet study on suicide?
How common is ambivalence before an abortion?
Summary of abortion and mental health
What are the policy implications?
What are the major studies?
On my academic papers page, you can read my book chapter on ‘The mental health complications of abortion’. This covers the questions I address in this section in greater detail. I also have another forthcoming book chapter entitled ‘Abortion’s causal role in trauma and suicide’, available on request, covering similar material in a bit more depth.
This section will include discussion of a variety of key papers and reviews:
- The Turnaway study is a study conducted by pro-choice researchers allegedly demonstrating many negative effects of denying women access to abortion. It compares the outcomes of women who had first-trimester abortions, women who had later abortions just under the clinic’s gestation limit, and women who were denied later abortions just over the clinic’s gestation limit, splitting the latter group into women who went to have an abortion elsewhere (or miscarried) and women who gave birth. It is spread across a wide variety of papers and claims that denying women abortion is associated with a wide range of negative outcomes for that women and her other children.
- The APA review (2008) was a review of abortion and mental health by the American Psychological Association, another group openly committed politically to abortion rights. It concluded that ‘among adult women who have an unplanned pregnancy the relative risk of mental health problems is no greater than if they have a single first-trimester abortion than if they deliver that pregnancy’. It thus limited its conclusions significantly, and even conceded that the evidence was more suggestive in the case of multiple abortions (bearing in mind most abortions are one of multiple). It ultimately based its conclusions on a single paper, Gilchrist (1995), noting that Fergusson (2006) and Gilchrist (1995) were the two best studies, but claiming that Gilchrist (1995) was slightly better. Fergusson (2008) was not published at the time.
- The NCCMH review (2011) was a more comprehensive review conducted by the National Collaborating Centre for Mental Health. By contrast, it performed meta-analyses using the best studies available for a range of psychiatric outcomes. It claimed that Fergusson (2008) was the single best paper available on the topic – a study which concluded that there were indeed links between abortion and a variety of mental health disorders. It also agreed that women with an unplanned pregnancy were more likely to be suicidal if they had an abortion than if they continued the pregnancy, but said that when women explicitly with unwanted pregnancies were investigated, there was no significant difference either way. It said that there was insufficient evidence of an increased risk of other conditions. They said that women having abortions had a lower risk of psychosis than women continuing an unplanned pregnancy, seemingly arguing that this cancelled out the increased risk of suicide so that the overall risk of mental disorders was about the same either way. It appears to concede that multiple abortions are associated with an increased risk of anxiety.
- David Fergusson has a body of work, having published a major study on abortion and mental health in 2006, which was considered one of the two best studies at the time. This was improved in his 2008 study, which was (and still is) considered as the best study on the topic. In 2009 he published a paper on a slightly different theme, more focussed on the prevalence of specific negative emotions after abortion and subsequent psychiatric outcomes. In 2013, he published a meta-analysis, improving upon the NCCMH review. This remains the most recent major review of the evidence. Fergusson’s primary studies and meta-analysis find that abortion is causally associated with a wide variety of psychiatric disorders including anxiety, suicidality, alcohol misuse and drug misuse.
Most contemporary guidance on this topic cites the conclusions of the NCCMH review, and sometimes the APA review as well. No further reason is typically given for omitting Fergusson’s analysis, or addressing the problems with the APA and NCCMH reviews. In the following questions, I will comment more on each of these studies and reviews.
Does abortion improve mental health?
Abortion is frequently justified on mental health grounds – indeed, 98% of abortions are officially justified under this reason in the UK, where ‘abortion for mental health’ functions as abortion on demand.
In fact, it is fairly widely agreed that there is no evidence that abortion improves the mental health of a woman experiencing an unwanted pregnancy. The leading researcher on abortion and mental health in the world – himself pro-choice – concluded in the latest major review on this topic: ‘at the present time there is no credible scientific evidence demonstrating that abortion has mental health benefits.’ The Royal College of Obstetricians and Gynaecologists – a professional body vehemently in favour of abortion access – says that ‘Women with an unintended pregnancy should be informed that the evidence suggests that they are no more or less likely to suffer adverse psychological sequelae whether they have an abortion or continue with the pregnancy’. The most high-profile study on abortion and mental health in recent years, led by leading pro-choice advocates, and which generally claims that denial of abortion is harmful for women, nevertheless concludes that ‘carrying an unwanted pregnancy to term was not associated with mental health harm.’ The primary researcher behind the study concluded: ‘I expected that raising a child one wasn’t planning to have might be associated with depression or anxiety. But this is not what we found over the long run… Women are resilient to the experience… at least in terms of their mental health… women’s symptoms of depression and anxiety are slowly relieved following an unwanted pregnancy, regardless of how that pregnancy ends.’
Does abortion harm mental health?
Because there is such a broad consensus that abortion does not improve mental health (relative to continuing an unwanted pregnancy), most of the research has focussed on whether abortion harms a woman’s mental health, once she has an unwanted pregnancy. In 2008, this was judged by the editor of the world-leading British Journal of Psychiatry to be on its way to becoming an ‘established fact’. The suggestion is now dismissed by some as being close to pseudoscience. So what is the truth?
The editor of the BJPsych said this in response to a study by pro-choice Professor of Psychology David Fergusson. This study found that abortion was associated with a 30% increase in the rate of mental disorder, and that 1.5-5.5% of all mental disorders may be attributable to abortion. Fergusson and his colleagues gave six further arguments that the link was causal. Their work is widely agreed to be top quality, even by the two major reviews some people use to deny the link between abortion and mental health problems (the APA review and the NCCMH review – see below). The NCCMH review explicitly said that Fergusson’s 2008 study was the single best quality study on the topic. Yet Fergusson was asked by the New Zealand government not to publish his results because of the potential political implications. This goes a long way towards explaining why so many continue to deny an association.
We will start with what the latest good quality evidence shows. In 2013, Fergusson himself undertook a meta-analysis of the best quality studies on abortion and mental health, and found a clear trend. Abortion was associated with increases of 28% for anxiety, 13% for depression, 69% for suicidal behaviour, 134% for alcohol misuse and 291% for drug misuse. The results generally persisted even when only the widely agreed best studies were used. Notably, the Turnaway study also found that alcohol misuse was massively reduced as soon as women were denied an abortion.
The link between abortion and suicidal behaviour is particularly important, not just because of the severity of another tragic loss of life, but because suicide has for a long time been the leading cause of maternal deaths in many developed countries, such as the UK. Pregnancy and childbearing are known to be major protective factors against suicide, so the causal explanation is perfectly plausible.
This increased risk of suicide is partly responsible for the fact that the mortality rate a year after abortion is triple the mortality rate a year after continuing a pregnancy (see ‘Abortion is 14 times safer than childbirth’).
The evidence is also uncontroversial that some women suffer post-traumatic stress disorder symptoms after abortion, with abortion-specific symptoms. Studies differ widely on the frequency of this (between 1.4% and 14.3%) and on whether the rates after abortion are higher than after pregnancy, but it is uncontroversial that abortion-specific post-traumatic stress disorder is a real phenomenon occurring commonly or very commonly (according to the standard definitions – page 38 – of ‘common’ and ‘very common’ used for intervention side effects). Even the minimum estimate of 1.4% would mean well over 10,000 women a year in the US each year getting PTSD as a result of the abortion.
Likewise, studies show that most women feel a range of negative emotions following abortion. Fergusson’s work in New Zealand – where abortion stigma is minimal – found that 71.1% of women felt sadness, 34.6% very much so. 63.5% felt guilt, 33.7% very much so. 60.6% felt no satisfaction at all. All of these would easily count – by a long way – as ‘very common’ according to the standard definition used for intervention side effects. Even in the Turnaway study, whose low participation rate and high attrition rate almost certainly selected out those who responded worst to the abortion, found that 17% still felt significant guilt five years later.
There is also evidence showing that women feel worse about their abortions as time goes on – so some studies may not capture the full psychiatric consequences of abortion if they have only a short follow-up period. APA review author Brenda Major found that as time went on, negative emotions about the abortion increased, while positive emotions and decision satisfaction decreased.
Is it because of the stigma?
It might be argued that the poorer mental health outcomes from abortion are primarily a result of the stigma attached to abortion rather than the abortion itself. It is not clear what the conclusion is supposed to be: even if this is so, for example, women should be warned of the poorer mental health consequences regardless of the exact mechanism. Arguably this is not an objection at all: negative attitudes towards abortion are the primary proposed intermediary mechanism for negative mental health outcomes. Though some have suggested that a sudden interruption of the hormonal changes in pregnancy may contribute, the primary supposed mechanism is from women recognising something troubling about the abortion itself. So perhaps the objection is supposed to be that external stigma from other sources is the cause.
It is, of course, likely that external stigma contributes to the negative psychological consequences of abortion. But there are reasons to think that this an incomplete explanation – and that even if it were a complete explanation, this would not necessarily be relevant to policy.
The first reason is that many of the studies showing poorer mental health after abortion are from countries with minimal abortion stigma; the most careful work from Fergusson comes from New Zealand, a strongly pro-choice country which recently managed to legalise abortion up until birth with minimal restrictions. This work shows not only that abortion is associated with poorer outcomes, but specifically that the large majority of women obtaining abortions feel guilt and other negative feelings about their abortion (see above). Some of the work linking abortion and suicide comes from Finland, which legalised abortion on mental health grounds in 1950 and more explicitly on socioeconomic grounds in 1970, and which has seen no significant attempt to restrict the law since.
Second, some stigmatising activity which has long been claimed to cause intense mental harm and trauma to women obtaining abortions has been found to have no significant impact on women’s feelings about their abortion. For example, the only studies examining women’s mental health in response to anti-abortion protestors outside clinics found that these protests made no substantive long-term difference at all—in the most recent study, by just a week after the abortion, no psychological impact from protestors was discernible.
Third, in the Turnaway study, 33 women (4.7%) with post-traumatic stress symptoms cited ‘the abortion experience or decision’ as the reason for their post-traumatic stress symptoms, with a further 3 citing ‘being reminded of abortion’ – for example, seeing small children. By contrast, only 4 said that ‘others’ reaction to abortion’ was the cause. The overwhelming majority of cases therefore were attributable to the abortion decision or experience, rather than the reactions of others.
Fourth, external stigma is likely never to be eliminated. Many people will always oppose abortion and will be vocal about it. Fifty years after Roe v Wade, opinion polling in the US has held steady, and enormous political advances have been made – Roe v Wade now looks closer than ever to being overturned with many states lined up to criminalise, or mostly criminalise, abortion as soon as the Supreme Court permits them. Even if stigma were the sole explanation for poor mental health outcomes, it is highly doubtful whether it is preventable – if not, then stigma is a given which must be factored into informed consent and other policy-making.
Finally, internal stigma is likewise ineliminable because of the reality of abortion. The testimony of women undergoing medical abortions demonstrates that women who did not at all expect to deliver anything resembling a human (and for whom internal stigma was likely low) nevertheless are not uncommonly alarmed by the sight of the embryo even at an early stage in pregnancy. An evidence review from the National Institute for Health and Care Excellence addressing informational needs for women undergoing abortion found that women were unprepared for viewing the baby even under 9 weeks:
‘was totally unprepared for seeing the embryo… became very sad… I could clearly see that it would be a human being.’
‘[I] just felt compelled, that I had to look… In hindsight I wish I hadn’t looked but I did, and that was probably the most traumatic thing I’ve ever seen or done. I thought ‘what on earth…?’
Later abortions, with a more recognisable baby, though less common, are still common—in England and Wales alone, there are around 40,000 abortions after 9 weeks, and around 13,000 abortions after 13 weeks. NICE highlighted evidence on second trimester abortion from Sweden, one of the most pro-choice countries in the world:
‘The women’s stories revealed how ignorant they were about the abortion procedure… The actual abortion of the fetus was an emotional experience, which they were not prepared for…some of them had decided in advance not to see but saw it anyway. One woman described that something hung in a string between her legs and realized after a while what had happened. The abortion of the fetus was not a pleasant sight and some described how they ‘broke down.’ The picture of the fetus was something they would never forget.
‘You could see the fetus, where the ears were, the arms, I was really frightened’…
Some of the women thought that they had killed a life and could never forget the pregnancy that was ended. They found it unnatural to have an abortion and experienced feelings of guilt and shame. One woman, although being certain of the decision, experienced serious regrets afterwards.
‘I have killed a life, a person that cannot live because of my decision and I will never ever forget’…
Furthermore, women also experienced a psychological pain, as anxiety, depression, and worry, which remained after the abortion and was described as varying from emotional disorder and reaction to emotional trauma…
‘the pain dissipated the thoughts a little, the bad pain pushed the other bad pain away and the other is more of a psychological pain’…
The women regretted the abortion to and fro and some of them expressed thoughts that they would never become normal again.
‘I will never forget the actual aborting… it was… it came like a shock to me’…
The women thought that it was difficult to foresee the grief before being in the actual situation. They expressed that the time period after the abortion was hard even if the decision to have an abortion had been their own. Some women still struggled with thoughts of whether they had made the right decision and talked about an understanding towards other women in the same situation. The women struggled also with the memory/picture of the fetus. They were astonished that some women could expose themselves to several abortions. The women’s emotionally difficult experience was brought up when they came home and some time afterwards. They experienced that the body recovered faster than their mind.
‘You can never forget this… I just want to wipe it out from my life’…
The women experienced strong negative feelings after the abortion, which were permanent, and the look of the fetus was disturbing.’
A significant proportion of women undergoing abortions likewise are morally opposed to abortion themselves. Studies from Sweden and Norway found that significant proportions of women had negative ethical attitudes towards abortion, while a study in Russia and the US found half of all women undergoing abortion thought abortion was morally wrong, with a further percentage unsure – in Russia’s case, after a century of legal abortion and one of the highest abortion rates in the world for the entire period. Though dated, an LA Times poll from 1989 – when pro-abortion sentiment was somewhat higher – found that 37% of those who had had an abortion said abortion was morally wrong, and only 39% felt it was morally right. One third of those who had had an abortion considered it murder. A 1996 University of Virginia poll found that 74% of all respondents considered abortion to be either murder or the taking of human life; if one third of all women obtain at least one abortion in their lifetime in the US, a significant proportion will fall into this category.
More recently, a study of 5,000 women obtaining abortions found that 17.5% felt abortion was the same, or “kind of” the same as killing a baby that is already born. Presumably a further – perhaps large – percentage feel that abortion is morally problematic even if not akin to infanticide: the same study found that 45% had spiritual concerns about their abortion. Again, presumably there are more who would have moral concerns about abortion without spiritual terminology. In the Turnaway study, 20% felt that abortion was morally wrong, with a further 15% thinking it depended on the circumstances. This is very likely an underestimate given the low participation rate and high attrition rate, both of which would select out those worst affected by abortion.
Abortion providers – perhaps those least likely to stigmatise abortion – are likewise confronted with considerable distress at the basic reality of abortion, at least at later stages. A study of Japanese nurses found that abortion after six months was the single most traumatic event nurses experienced, with every nurse experiencing it rating the trauma of it as 10/10. Feticide – an injection to end the baby’s life before inducing an abortion in order to prevent it being born alive – is widely described in dramatic terms: ‘Over half of the [fetal medicine specialists performing feticide] expressed internal conflict about the provision of feticide and the need to ‘separate yourself from it completely’. They described it as ‘brutal’, ‘awful’, and ‘emotionally difficult’, referring to it as ‘stabbing the baby in the heart’.’. Almost all fetal medicine specialists, the study found, have a line they do not cross – abortion on demand at any point in pregnancy is emotionally and ethically anathema even to most abortion providers.
Earlier in pregnancy, there is widespread unease among abortion providers even for second trimester abortions. This was described by abortion doctor Lisa Harris in a leading journal in favour of abortion access:
‘Kaltreider et al found that some doctors who provided D&E had “disquieting” dreams and strong emotional reactions. Hern found that D&E was “qualitatively a different procedure – both medically and emotionally – than early abortion”. Many of his staff members reported:
“…serious emotional reactions that produced physiological symptoms, sleep disturbances (including disturbing dreams), effects on interpersonal relationships and moral anguish.”’
Harris goes on to describe another abortion doctor who vowed to stick to 14 weeks as a limit, after seeing an arm pulled through the vaginal canal; ‘the reality is, this cannot be called ‘tissue’. It was not something I could be comfortable with.’ Harris then relays her own experiences as someone who still supports second trimester abortion access, and presumably still performs them: how, while performing a second trimester abortion, she had an uncontrollable visceral reaction which travelled ‘from my hand and my uterus to my tear ducts’ – since at the exact same time as she ‘separated the leg’ with a ‘quick tug’ and held the leg and foot in her forceps, she felt her baby of the same gestation kicking in her abdomen. She described another episode where she performed a 23 week surgical abortion (dilation and evacuation, or dismemberment abortion) and shortly afterwards visited a 23-24 week premature delivery, the baby ending up in neonatal ICU: ‘I thought to myself how bizarre it was that I could have legally dismembered this fetus-now-newborn if it were inside its mother’s uterus – but that the same kind of violence against it now would be illegal, and unspeakable.’
The literature on abortion providers’ attitude to surgical abortion is replete with similar sentiments. It frequently draws attention to the ‘coping strategies’ needed by nurses to ‘conceal emotions’, and the need for official psychological and emotional support programmes for people providing abortions. One study in the NICE evidence review reported a woman and medical trainee so shocked at the ‘little human being’ delivered that the woman’s partner had to calm down the ‘abandoned’ trainee.
For all these reasons, the objection that negative psychological outcomes from abortion are only the result of external stigma are unpersuasive. There will always be a level of ineliminable stigma attached to abortion as a result of its basic reality as killing human life; if not for very early abortion, at least for later abortion – and these are unlikely to become separated in the public mind for the foreseeable future.
“Women having abortions have worse mental health to begin with”
This is true, but it does not explain the evidence. The reason is that the best studies – cited above – control for confounding factors such as prior mental health. They found that even after comprehensive confounder control, having an abortion was still associated with poorer mental health than continuing an unwanted pregnancy.
Is the link causal?
Clearly, there is unassailable evidence of a causal link in some cases (e.g. in abortion-specific PTSD). There is also very strong evidence of a causal link in general: the explicit indexing of mental health problems to the abortion among many women, the obvious causal mechanisms (the frequency of women feeling morally or spiritually uneasy about abortion, the frequency of guilt, the protective effects of pregnancy and childrearing against suicide and alcohol/substance misuse), the persistence of association after thorough confounder adjustment, the dose-dependency, and so on. Fergusson offers a more complete discussion.
“95% of women having abortions feel they made the right decision”
This statistic comes from the Turnaway study, which I discuss under ‘What about the Turnaway study?’ In summary, there are a huge number of problems with the Turnaway study which make this an unreliable conclusion: in particular, the enormous attrition rate, meaning that a huge number of those invited to the study either declined or later dropped out of the study. Since women who found the abortion a more traumatic experience are more likely to drop out (see ‘What about the Turnaway study?’), 95% is probably a significant overestimate. This is confirmed by the fact that other studies of much higher quality have found that most women feel they made the right decision, though with significant levels of doubt. APA review author Brenda Major found that just 2 years later, only 72% of women were satisfied with their decision, and this decreased over time (and only 69% would make the same decision again). Fergusson found at age 30, 89% of women felt they made the right decision.
There is an often overlooked fact from the Turnaway, however, which mirrors (and even outweighs) the 95% statistic. This is that 98% of women refused abortions who kept the baby and raised it were glad they were refused an abortion by the child’s 5th birthday (and the 2% included ‘unsure’). 85% of those who kept the baby and placed the child for adoption were glad by the child’s 5th birthday – lower, but still high. On average, 96% of women who kept the baby until birth were glad they did so (or, one might say, were forced to do so). The remaining 4% (or 2% for women who raised the child) included women who answered ‘don’t know’.
Nor does this prove that abortion regret is rare. A substantial minority of women feel they made the wrong decision, and most women even in very liberal Sweden say they would never have an abortion again. Around a third of women feel the emotion of regret, even though many of these feel they made the right decision. Feelings of regret regarding abortion are complex phenomena and cannot simply be reduced to feeling one might the right decision.
It is true, therefore, that most women feel they made the right decision in the years following an abortion (to my knowledge, there are no very long-term studies). But this number is significantly lower than 95%, and is actually lower than the proportion of women who were eventually be glad to be denied an abortion. Moreover, this is consistent with a significant proportion of women feeling emotions of regret and most women wishing never to have an abortion again.
Women suffer emotionally if they don’t get abortions
By now, we are in a position to respond to this claim. It is clear that abortion is not necessary to prevent clinical mental disorders, and in fact makes them worse. Recall that the RCOG say that ‘Women with an unintended pregnancy should be informed that the evidence suggests that they are no more or less likely to suffer adverse psychological sequelae whether they have an abortion or continue with the pregnancy’, while Fergusson concluded that ‘there is no credible scientific evidence demonstrating that abortion has mental health benefits.’ The lead author of the Turnaway study themselves admitted: ‘carrying an unwanted pregnancy to term was not associated with mental health harm… I expected that raising a child one wasn’t planning to have might be associated with depression or anxiety. But this is not what we found over the long run… Women are resilient to the experience… at least in terms of their mental health… women’s symptoms of depression and anxiety are slowly relieved following an unwanted pregnancy, regardless of how that pregnancy ends.’
But is abortion necessary to avert serious emotional distress short of mental disorder? Again, the evidence is pretty clear that the answer is no, not in general. Although it can easily be seen how a crisis pregnancy causes an initial panic, and abortion is the ‘great reliever’, the evidence that unwanted pregnancy causes lasting emotional harm is virtually non-existent (except in rare cases; but the same is true – more commonly – of abortion), and eventually almost all women are glad they continued the pregnancy. As we have just seen, a higher proportion of women denied abortions were glad to have been denied than the proportion of women who feel they made the right decision to abort. The Turnaway study also showed that life satisfaction was virtually identical between women who aborted and women who were refused abortions. The only significant difference was that women denied an abortion but who went on to have an abortion elsewhere (or a miscarriage) had marginally worse life satisfaction on average, a week after originally being denied an abortion. This did not last longer than a week in the study, however. Otherwise, life satisfaction was virtually identical between all the groups. Only 9% of women had poor maternal bonding, and even the overwhelming majority of these were still happy they had the baby.
Foster, the lead author of the Turnaway study, writes: ‘How do women feel about having been denied an abortion? Initially, bad. But over time, most of the women who ended up carrying the unwanted pregnancy to term reconciled themselves to their new reality, especially after their babies were born… Women don’t often say they want an abortion for fear of what an unwanted pregnancy would do to their mental health. And mental health rarely seems to suffer, even when abortion is denied… Most of the women turned away, over time, reported that they were happy they had the baby.’ She then goes on to describe the testimonies of women denied abortions and their feelings about their children: one started crying at the thought of not having her child: ‘She is just everything to me’; for another, the child was ‘terrifyingly bad timing’ but the mother was still ‘relieved that what happened happened… it breaks my heart that I actually thought about [having an abortion].’ Another described her life as crazy prior to becoming pregnant, with very little money, and she became angry at being denied an abortion. But she eventually said that having her child was ‘the best experience of my life’, making her a ‘better person’ and providing unparalleled intimacy. She said that she thought she would have died from drugs or recklessness if not for her baby.’
Although in some cases this coming to terms with the pregnancy can take some time, it often happens remarkably quickly. Even a week after abortion denial, only 65% of women still wished they could have an abortion, while 49% felt relieved. Only 12% still would have preferred an abortion by the time of the child’s birth. 27% of women who received an abortion felt happy about the pregnancy a week later, but 60% of those denied an abortion felt happy about it at the same time. A similar rare study on victims of rape found that over the course of the pregnancy, the average psychological state of rape victims improved dramatically across a wide range of outcomes. One recent study found that even just a pregnancy scare considerably increases the desire for pregnancy on average: women with a pregnancy scare were up to five times more likely to desire a pregnancy compared to their baseline. The authors suggested that the mere-exposure effect may be responsible for this finding. Another reason feelings about pregnancy may change so quickly is that pregnancy intention itself is not as binary as often thought: what are labelled unplanned or unwanted pregnancies often turn out to be highly ambivalent pregnancies.
In summary, there is no evidence that abortion is needed for either the mental health or ‘life satisfaction’ of women. There is evidence that abortion is worse for women’s mental health, however, and evidence that it has made women unhappier in other ways (see ‘How abortion harms women’).
What about the APA review?
The APA review claims that ‘among adult women who have an unplanned pregnancy the relative risk of mental health problems is no greater than if they have a single first-trimester abortion than if they deliver that pregnancy’, conceding that ‘The evidence regarding the relative mental health risks associated with multiple abortions is more equivocal’.
The review initially claims that the question of whether abortion harms mental health is ‘not scientifically testable’. This is a strange claim which appears to render the conclusion of no association inevitable. The reasoning is that randomised experiments are not ethically possible. This is true, but does not mean that the question is untestable. It simply means that the very best kind of evidence is not available for this question. We can still have good evidence.
The review is also severely limited in scope; it is limited to single, first-trimester abortions in adults with unplanned pregnancies – thereby excluding the (overall) majority of abortions which occur either as one of many, in the second or third trimesters, in children, or in planned/wanted pregnancies. Indeed, these are widely acknowledged risk factors for mental health sequelae after abortion – but these women are entirely excluded from the study. The review’s conclusions, even if true, would be virtually irrelevant to the large majority of women seeking abortion. It even appears to concede a possible detrimental effect of multiple abortions.
Still, the review does provide a basis for its limited conclusion: two primary studies are highlighted as being of standout quality, Gilchrist et al. (1995) and Fergusson et al. (2006). Although Fergusson found that women having abortions had higher rates of mental disorder after adjusting for confounding factors, his conclusions were roundly dismissed owing to a few methodological concerns. Hence the conclusions were based on a single paper by Gilchrist. Gilchrist did find that deliberate self-harm was higher in women who had abortions and women who were denied abortions compared to women who chose to keep an unplanned pregnancy. However, this was only among women with no prior psychiatric history. Among those with previous psychiatric history, women denied an abortion had a far lower rate of deliberate self-harm than women who aborted, though the small sample size prevented any firm conclusions being drawn. Among women with prior non-psychotic mental illness, deliberate self-harm was much higher among women who aborted compared to both women who had an unplanned pregnancy and chose to keep the baby, and women who were denied abortions. Other than psychosis, for which no significant results were found, Gilchrist did not address other specific mental disorders, giving the study limited value for studying specific conditions.
Hence even conclusions based solely on Gilchrist’s study should have offered these more nuanced conclusions, rather than a blanket assertion of no association between abortion and mental health. Likewise, it should have said that there is no top quality evidence on anxiety, drug misuse, and alcohol misuse specifically, and hence remained agnostic on their relationship with abortion. I will comment more on the Gilchrist study in the NCCMH question below.
In fact, however, Fergusson published an updated and improved paper in 2008 which resolved the methodological concerns of the APA review, and found the same results: increased rates of a wide variety of mental disorders. For this reason, the 2011 NCCMH review rated Fergusson et al. (2008) as superior to that of Gilchrist, and the best study in existence. The APA review is therefore critically outdated—but by its own standards, a revised review would presumably have to follow the conclusions of Fergusson (2008)—that abortion is causally* associated with a diverse range of mental disorders. Fergusson estimated that abortion led to a 30% increase in mental disorders overall, and that 1.5-5.5% of all mental disorders among the cohort were attributable to abortion.
*Fergusson et al. give a number of arguments that the link is causal.
What about the NCCMH review?
The 2011 NCCMH review was considerably more thorough than the APA review and, crucially, included Fergusson’s 2008 study. It also refused to base conclusions on one paper: it performed a meta-analysis across the few best quality studies on each mental health outcome. These studies all compared women having abortions to women continuing unwanted or unplanned pregnancies – failure to include this comparison group ruled the study out of consideration.
It claimed, firstly, that Fergusson (2008) was indeed the best quality study in the whole field. At the time of writing, no study has surpassed it in quality (see below for comments on the Turnaway study and a more recent study by Steinberg et al.). So it is worth examining the precise conclusions of the NCCMH review, since they appear to differ from Fergusson’s.
Regarding anxiety, the meta-analysis found a 28% increase in anxiety for women undergoing an abortion, but since the confidence interval of the odds ratio marginally overlapped 1 (0.96-1.71), they concluded that there was insufficient evidence of an elevated risk – importantly, not that there was evidence of no elevated risk. The insufficient evidence was likely due to the small overall sample size, not a lack of association. This is confirmed by a fact unchallenged in the NCCMH report – that the only good quality study to address the risk of anxiety from multiple abortions found a statistically significant increased risk of 69%. This dose dependence argues in favour of a causal link, and of a genuinely significant link between single abortions and anxiety.
Fergusson (2008) was the only good quality study available on alcohol dependence. Although it found that the risk of alcohol dependence was seven times higher among women having abortions, the small sample size again prevented a statistically significant conclusion. Again, the evidence was deemed insufficient, not deemed to show a lack of association. The same is true of substance misuse, though in this case the risk was thirteen times higher in women having abortions. This prima facie association is entirely to be expected given the well-documented causal decrease in alcohol and drug use among pregnant women (the Turnaway study itself found that within a week of women being denied an abortion, drinking alcohol and binge drinking were massively reduced compared to women receiving an abortion, despite similar levels prior to seeking an abortion).
The results on suicidal behaviour and self-harm were more nuanced. In the meta-analysis, there was a clear association – explicitly concluded by the NCCMH – between abortion and suicidal behaviour, compared to continuing an unplanned pregnancy. However, the NCCMH claimed that when unwanted pregnancies were used from Gilchrist’s study – as defined by women who sought an abortion but were denied – the association disappeared on average between studies. Notably, the review only included women with no prior psychiatric history from the Gilchrist study – neglecting the fact that among women with a prior psychiatric history, rates of self-harm were much higher among women aborting than those with ‘unwanted’ pregnancies (i.e. denied an abortion). This skewed the NCCMH’s summary of Gilchrist’s study towards the null hypothesis. Again, the result on suicidal behaviour should be unsurprising, since pregnancy and motherhood are well-established strong protective factors against suicide (see ‘Does abortion harm mental health?’).
The review appeared to suggest that the lower rate of psychosis following abortion (from Gilchrist’s study) cancelled out the increased rate of suicidal behaviour, leading to no net difference. But Gilchrist’s study clearly stated that the findings on psychosis were spurious and that a better measure of psychosis found no disparity between the two groups. The NCCMH review appears to have misinterpreted the evidence on psychosis, leading to a misleading ‘balancing out’ of the risks of suicidality and psychosis.
The review found that overall levels of mental disorder were slightly higher in women aborting than women continuing an unplanned and unwanted pregnancy—but this did not meet statistical significance.
Hence, in summary, the NCCMH study should be interpreted not as concluding that abortion does not cause psychiatric disorder, but as concluding that the best quality evidence does suggest various associations, though these need to be confirmed with larger sample sizes. It also concludes clearly that multiple abortions are significantly associated with anxiety. Likewise, it demonstrates that abortion is worse for one’s risk of suicidal behaviour than choosing to keep the pregnancy, but it erroneously balanced this off against an illusory reduction in psychosis from abortion.
What about Gilchrist’s study?
The more controversial question is the NCCMH’s interpretation of the evidence on suicide. While it agrees that keeping an unplanned pregnancy reduces one’s risk of suicide compared to abortion, it suggests that when the pregnancy is unwanted (i.e. an abortion is sought), the evidence is more mixed, since Fergusson’s study is cancelled out by Gilchrist’s study. We have already seen that this conclusion involved erroneously omitting anyone with prior psychiatric disorder from Gilchrist’s data, but there are a number of other problems with treating Gilchrist’s study as equally informative to Fergusson’s:
- In most Western countries, since abortion is generally legally available, the most relevant comparison is women choosing abortion with women with unplanned pregnancies choosing to continue pregnancy. Hence it makes no sense to dismiss the evidence regarding unplanned pregnancies as irrelevant. Women should be informed that, in the event of an unplanned pregnancy regarding which she may be ambivalent, the evidence supports her continuing with the pregnancy.
- It is universally acknowledged that Fergusson’s study was better quality than Gilchrist’s, and the former showed an association between abortion and suicidality. To prioritise Gilchrist’s study, or even give it the same weight, would not do justice to the evidence. Gilchrist’s study also fails to cohere with the findings of the Turnaway study, which showed that women refused abortions who continued their pregnancies had no substantive negative mental health consequences, despite the study being biased towards this outcome (for reasons explained in ‘What about the Turnaway study?’).
- There are a variety of concerns regarding Gilchrist’s methodology:
- Unrepresentative sampling: the sample was selected from general practice, leading to a potentially unrepresentative sample (APA 2008).
- High attrition rate: by the end of the study, only 34.4% of the abortion group remained, and 43.4% of the continued pregnancy group.
- Follow up: it is unclear how reliable follow up was, since this relied on GPs having access to all diagnoses for women involved – which may not have been the case for certain hospital admissions.
- Measurement of outcomes: likewise, the study has been criticised for not using any standardised measure of mental illness, relying on highly variable assessments by GPs (APA 2008).
- Small sample size: these considerations are particularly important given the small sample size for the self-harm results: just 8 women who were denied abortions had self-harming episodes. As a result, the confidence intervals for women denied abortions and those obtaining abortions mostly overlapped – the increased rate was not statistically significant.
- Miscarriages: it is likely that some women who were refused abortion ended up aborting the pregnancy elsewhere – those refused abortion had miscarriage rates about twelve times higher than those continuing the pregnancy. If so, then the group of women refused abortion will include some women who had abortions, whose negative outcomes would give a misleading impression of women refused abortion who kept the baby as a result. Women who keep their babies after being refused an abortion generally have significantly better outcomes than those who have an abortion elsewhere, even when their pregnancy intentions and ease of decision to abort are similar (note that this is another argument in favour of the causal effect of abortion, since the discrepancy cannot easily be attributed to women seeking abortions elsewhere being more desperate for abortions as a result of worse circumstances); even if not, however, women aborting elsewhere will artificially inflate the rates of mental disorders in the overall cohort unless separated.
- Confounder control: the study has limited confounder control, in particular failing to note the particularly important confounder of social support. In the early stages of legal abortion in the UK, evidence suggests women were more likely to be denied an abortion if they had poor social support – and hence this group of women with poorer social support would be expected to have worse mental health outcomes regardless.
- Unrepresentative: by the late 1970s, abortion was widespread in the UK and access to it may have been expected by women. Some of the negative outcomes from being refused an abortion may have instead been due to disappointed expectations and would not apply in a country or state where abortion is not expected to be available. Another sense in which it is unrepresentative is that it is unclear what pregnancy and childcare support was available to women denied abortion in the UK in the 1970s, compared with the significant support available publicly and privately in many countries, especially the US, in the contemporary setting.
- As noted above, the analysis from the NCCMH excluded women with prior mental health difficulties. This group had lower rates of self-harm when denied abortion than women obtaining abortions, and is the group at highest risk for post-abortive mental health problems.
For all these reasons, it is implausible to suppose that Gilchrist’s results undermine Fergusson’s findings.
What did Fergusson’s meta-analysis show?
Following the NCCMH review, David Fergusson himself – by this time widely agreed to have led the best study on this subject – performed a meta-analysis, including a few more of the higher quality studies to increase the sample size and therefore probe whether the non-statistically significant associations identified in the NCCMH review were genuine associations or probabilistic artefacts. He found that when other good quality studies were included, abortion was associated with statistically significant increases in anxiety, alcohol misuse, illicit drug misuse and suicidal behaviour, compared to continuing an unplanned or unwanted pregnancy. These links persisted after a sensitivity analysis to eliminate the lowest quality studies.
Table 1. Odds ratios and significance levels from Fergusson’s (2013) meta-analysis. Significance levels are for two-tailed tests. Adapted from Fergusson et al. (2013).
|Outcome||Odds Ratio||Confidence Interval||Significance|
|Illicit drug use/misuse||3.91||1.13-13.55||0.03|
Fergusson was himself pro-choice, and did not anticipate these findings. But he was committed to clinical accuracy regardless of politics, and said that while he thought abortion should be legal, it should not be performed on a false pretence of mental health, for which there is ‘no credible scientific evidence’. The latest meta-analysis, therefore, confirms that abortion is associated with anxiety, alcohol misuse, illicit drug misuse, and suicidal behaviour, even after adjustment for possible confounding factors.
What about the Turnaway study?
The Turnaway study compared women who received abortions with women who were denied abortions because they were over the gestation limit. While it has received a lot of media attention in recent years, has fatal methodological weaknesses:
- It had a very low response rate and a very high attrition rate. By the end of the study, only 17% of those invited to the study remained (and the large majority refused to even join the study). Since women who have negative reactions to abortion are more likely to drop out (as the Turnaway study themselves found), this very depleted sample cannot be relied upon for conclusions about mental health.
- The sample was not chosen randomly but seemingly chosen at the discretion of researchers. Only 58% of eligible women were approached for the study, with no explanation of why the others were not included.
- The worse mental health outcomes among those denied an abortion were short term, and most importantly, only occurred among those who went to obtain an abortion elsewhere. Those women denied abortions who kept their babies had no worse mental health outcomes at any point. In fact, this is suggestive evidence that women who have abortions have worse outcomes than those who continue the pregnancy.
- There is likely a selection bias, since women who are unaware of the gestation limit at their local abortion clinic and who haven’t checked eligibility several months into the pregnancy are more likely to be more disorganised, or perhaps more ambivalent, than women who present within the time limit. These sorts of factors are likely (or indeed proven) to be associated with worse mental health outcomes after abortion.
- It has been reported that the authors of the Turnaway studies have refused to publish the questionnaire used, raising concerns about the selective publication and omission of important results.
- The study provides minimal information about the support given to women who were refused abortion, so it is unclear whether these women were given the kind of support endorsed and offered by pro-lifers.
- The measures of anxiety and depression used were extremely limited, consisting of only 6 questions each.
- Finally, of course, as mentioned above, the same study, even if methodologically sound, still found that at least 96% of women denied an abortion who kept the baby were glad about it in retrospect – and 98% who didn’t place the baby for adoption.
In sum, the Turnaway study is of significantly lower quality than other studies on this topic and does not provide any reason at all to doubt our conclusions. Indeed, it even offers some limited data supporting our conclusions. See also Reardon’s response to the study.
What about Steinberg’s Lancet study on suicide?
A recent paper by Steinberg et al. is frequently cited to undermine what is now a relatively clear link between abortion and suicidality specifically. The study, using a large sample in Denmark, found that ‘women who had abortions had a higher risk of non-fatal suicide attempts compared with women who did not have an abortion. However, because the increased risk was the same both the year before and after the abortion, it is not attributable to the abortion’. There are a few problems with this study:
- The study is not representative of women having abortions. It excludes women who have had previous abortions, women who have had previous suicide attempts, women having abortions after the first trimester, women having abortions at private clinics, and children. Thus it excludes many women with the highest risk of mental disorders following abortion.
- Probably the most fatal problem with the study is that it does not include a comparison group of women with an unplanned or unwanted pregnancy. Thus, by the inclusion criteria for high quality studies in NCCMH, it would be automatically excluded on these grounds. This is reason enough to discard the study in favour of better quality studies. Ironically, Steinberg herself had previously criticised Fergusson’s earlier paper for not having a comparison group of women with an unintended pregnancy. But her new study does not even have a comparison group of women with any pregnancy.
- The authors claim that because the suicide risk was the same before and after the abortion, abortion was not responsible for the (conceded) higher risk of suicide. This logic simply does not follow. Since pregnancy and childbearing are protective against suicide, the relevant comparison is with women continuing a pregnancy, not comparison with prior to an abortion. The comparison with women continuing a pregnancy is never made.
In fact, the study gives subtle evidence that abortion does increase the risk of suicide relative to pregnancy. In the month before the abortion, during which many women were aware they were pregnant, the suicide rate falls dramatically and seemingly anomalously. By contrast, after the abortion, the suicide risk goes immediately back up to baseline. This confirms the now well-established fact that abortion eliminates the protection that pregnancy offers against suicide. The study also notes that, in their sample (as in other research), women having given birth to a child had a significantly lower suicide risk than women who had not.
4. Finally, the discussion in the paper makes straightforwardly false claims about previous studies, claiming that ‘Other research has not found significant associations between abortion and suicidal ideation.’ As we have seen, this is simply untrue. The study entirely whitewashes the history of study on this question, not even citing Fergusson’s seminal 2008 paper, widely agreed to be the highest quality study on this topic.
For these reasons, the study is profoundly discreditable. Reputable journals can sometimes publish very poor or discreditable research. The Lancet itself has done so on numerous occasions. This is certainly not a reason to dismiss everything they publish – but it does warrant a healthy scepticism and critical attitude towards some literature published therein.
How common is ambivalence before an abortion?
Ambivalence before abortion is very common. As explained in ‘95% of women having abortions feel they made the right decision’, research from Major and Fergusson shows that between 11-30% of women did not feel they made the right decision after an abortion – and this number may increase over time.
In the Turnaway study, 27% of participants found the decision ‘very difficult’, and a further 27% ‘somewhat difficult,’ even despite a possible selection effect whereby women finding the decision difficult refused to participate; the authors note that women who felt less relief and happiness at baseline were more likely to be lost to follow up.
Husfeldt et al. found that 30% of women were still in doubt about their abortion 2 days beforehand. Rocca et al. found that within a week of being denied an abortion, 6% already said they didn’t know whether they wished they could still have an abortion, while 35% said they did not wish they could still have an abortion. In Sweden, Soderberg et al. found that 12% of women seeking abortion decided to continue the pregnancy in the end, with Holmgren and Uddenberg arguing that the rise of medical abortion in Sweden reduced decision-making time and exacerbated the problem of hasty ambivalent abortions. BPAS, the leading abortion provider in the UK, reports that 16% of women seeking abortion there end up having their baby, even though at other abortion providers women are encouraged to proceed with abortion regardless of their ambivalence. Likewise in the UK, 3% of women said they considered continuing the pregnancy right up until the night before the abortion. In Norway, 15% had not decided to have an abortion before seeking out abortion services. Early literature in the UK showed that of 7,110 women seeking abortion, 4.5% decided against abortion in the end, and a further 5.3% were denied. Among those with a failed medical abortion and continuing pregnancy in the UK, 10% decided to continue with the pregnancy in the end. In the major study on abortion pill reversal, 754 patients sought abortion pill reversal, and 8% of those seeking abortion reversal ended up having an abortion again, along with up to 15% lost to follow up. Practitioners in the UK claim that 90 women have sought abortion pill reversal in a one year period.
As pro-choice reproductive health scholar Sam Rowlands puts it:
“It is well known that women requesting abortion do change their mind. Change of mind is correlated with the degree of ambivalence. All abortion providers see occasional cases of women backing out at the last moment, even in the anaesthetic room.”
See also ‘How common is coercion?’
Summary of abortion and mental health
In short, we know the following with considerable confidence:
- Abortion is causally associated with increased rates of suicidality, drug and alcohol misuse, and anxiety. Multiple abortions appear to increase the risk substantially more.
- Abortions-specific post-traumatic stress disorder is a common to very common complication of abortion, affecting somewhere between 1.4-14.3% of women obtaining abortions – equating to at minimum 10,000 women in the US every year, for example.
- The large majority of women obtaining abortions feel guilt, sorrow, sadness and disappointment following an abortion, even though most also feel relief. A substantial minority of women feel guilt even many years later.
- Most women feel they made the right decision to have an abortion. However, a substantial minority (up to 30%) did not feel they made the right decision, and most say they would never want to go through an abortion again. There is some evidence that decision satisfaction decreases over time.
- When denied abortion, women quickly come to terms with the pregnancy and almost always are glad about it by the child’s 5th birthday. 98% of women denied abortions who raise the baby are glad they were refused an abortion by the child’s 5th birthday, and 85% of women denied abortions who placed the child for adoption were glad (even just one week after being denied an abortion, only 65% of women still wished they could have an abortion). Only 9% of women denied abortions had poor maternal bonding, and the overwhelming majority of these were still happy they had the baby.
- Ambivalence is very common among women obtaining abortions, with only 65% of women denied abortions wishing they could still have one just a week later.
- While external stigma contributes to these negative mental health outcomes, it cannot fully explain the association between them, and even if it did, it may be ineliminable.
- There is no evidence that women denied abortions, on the whole, experience negative emotions or mental health as a consequence.
- Abortion may be responsible for around 1.5-5.5% of mental disorders in a given cohort of reproductive age women.
All this is particularly important since suicide is the leading cause of maternal death in the developed world; it is partly for this reason that mortality following abortion vastly exceeds mortality following completion of a pregnancy. Typically, while maternal mortality statistics include deaths from suicide after pregnancy, abortion mortality statistics do not; this is one of the numerous reasons to be sceptical of claims that abortion is safer than childbirth.
What are the policy implications?
In light of this evidence, there is a good case for informed consent laws ensuring that women are properly informed about potential mental health consequences of having an abortion, prior to doing so. Women should also be screened for risk factors of negative psychological outcomes (this is already widely recommended, for example by the Royal College of Psychiatrists and the NCCMH review, and yet still not routinely implemented), with the risk factors being well-known. Moreover, doctors should generally refuse abortions on clinical grounds, even if abortion on demand should be available as a non-clinical intervention. Routine psychological follow-up could be justified, and psychological support after abortion should at least be expanded, given widespread concerns that, while it is easy to obtain an abortion in the UK, post-abortion support is severely lacking, especially in the case of disability-selective abortion. There is a case for limiting abortion to protect women’s mental health, either by limiting it to exceptional circumstances or more limited interventions, like mandatory waiting periods. Elective abortion should not be considered in any way healthcare (whether in terms of public funding or otherwise).
More research would also be most welcome. Contrary to popular representation, which cites the NCCMH review as the decisive evidence showing abortion is not associated with poorer mental health outcomes, the most claimed by the NCCMH review is simply that the evidence was very uncertain for every outcome. It is puzzling, therefore, that the review recommended diverting research away from this question. It can only be reasonably concluded that some working in this field do not want an answer – or perhaps are scared of the answer. The censorship experienced by David Fergusson – himself pro-choice – is perhaps an indicator of why.