“Abortion is 14 times safer than childbirth”: abortion and mortality risk

Do women die from not receiving abortions?: Savita and Izabela

Does legalising abortion prevent women dying from backstreet abortions?

“Abortion is 14 times safer than childbirth”: abortion and mortality risk

It is sometimes claimed that abortion is 14 times safer than childbirth – meaning that the mortality rate is just one fourteenth of the mortality rate from continuing a pregnancy. This is based on a 2012 study by Elizabeth Raymond, a doctor at Gynuity, an abortion advocacy research organisation.

When considering mortality rates after pregnancy and abortion, it is of central importance to discern exactly which deaths are being included and which are not. Deaths from suicide, for example, are often included under maternal deaths (relating to pregnancy), but not under abortion-related deaths. This is important since suicide is, in fact, the leading cause of maternal deaths in many developed countries. Likewise, ectopic pregnancies often count as maternal deaths, though they are clearly irrelevant to the relative safety of childbirth vs abortion, since an ectopic will generally have been discovered and treated by the time a woman has to decide whether or not to have an abortion. Indeed, deaths from abortion are often included as maternal deaths and are cited by abortion advocates as one of the leading causes of maternal mortality in developing countries. This is a strange categorisation if women having abortions are not already mothers, but we will set that concern aside for the moment.

The Raymond and Grimes study at least seems to avoid these typical errors, by specifically looking at deaths from live births, and therefore excluding deaths from ectopic pregnancies and abortions. Nevertheless, a variety of problems remain.

For example, Raymond and Grimes dismiss the long-term evidence on mental health by reference to a study by Charles et al. This review is now long-outdated, and was outdated even by the time of Raymond and Grimes’ paper (the Charlies review predated the APA report, NCCMH report, and Fergusson’s meta-analysis; for specifics on problems with the Charles review please email me); as I show in my section on mental health, the literature on abortion and mental health now clearly shows causal associations between abortion and mental health.

Another problem is that, of course, most women who find themselves pregnant want to continue their pregnancy. This statistic will not help them. What about those women who don’t want to be pregnant? Presumably the argument being made by abortion advocates here is that by limiting access to abortion, you will be subjecting such women to a higher risk of death. Some women will die as a result. One of the problems with this argument (and all arguments based on negative outcomes for women refused abortion) is that studies show that, generally, limiting abortion access has a very strong effect in preventing women from becoming pregnant in the first place: women who really do not want to be pregnant are typically much more careful not to become pregnant when there are limitations on abortion, just as one would expect. Hence, with moderate limitations on abortion, the abortion rate goes down, but the birth rate stays the same. With stronger restrictions on abortion (i.e. a complete ban), this can temporarily increase the birth rate, but often only temporarily and by a surprisingly small amount. So in fact, for most women who desperately do not want to be pregnant, limiting abortion access will not subject them to dangerous pregnancies, but rather will prevent them from the distress of an unwanted pregnancy in the first place. Hence it is questionable whether any extra women would die from pregnancy-related complications as a result of restricting abortion, even assuming pregnancy was risky and abortion was entirely risk-free (which, of course, is not true). For more on this, see ‘Does restricting abortion lead to more unwanted pregnancies?’

Another problem, which the authors concede themselves, is that maternal deaths are often unreported, and differentially by outcome, referencing a study finding that neither of the women dying from abortion in the sample were identified as having had an abortion from death records alone. In fact, this problem is far more widespread than even they imply, as noted by Reardon and Thorp in their systematic review on this topic. Among others, Calhoun and Reardon have detailed the problems with accurately identifying maternal deaths and abortion-related deaths. In fact, Finnish data have shown that pregnancy is far less likely to be mentioned in death records in the case of abortion than in the case of a live birth. This means that deaths from abortion are especially likely to be missed – by a large margin. The references given above contain more detail on this issue. This is not a problem limited to the US: see my chapter on telemedicine abortion for further examples of how abortion complications are massively underreported.

This is obviously a critical problem: if a large proportion of deaths from abortion go unreported – as seems to be widely agreed – there simply isn’t a basis for the abortion mortality rate offered by Raymond and Grimes. While their estimate of pregnancy-related deaths is probably fairly accurate – it is not far off the maternal mortality rate in other developed countries – their estimate of abortion mortality is wholly unreliable.

For another critical problem, consider the following quote:

‘Chronically high levels of stress can wreak havoc on bodily systems from the brain to the heart, and accelerate the pace of biological aging, according to experts. For people of color, race-based stressors can take a heavy toll on health for a life-time – and even across generations.’

So reported a news release from Harvard’s TH Chan School of Public Health, in the context of the impact of racism on the physical health of people of colour. I have no doubt that this is true. Medical sociology has known for a long time that psychosocial factors have an enormous effect on physical health – so much so that they simply cannot be excluded. The problem is that they are rarely reported as causing deaths. Cardiovascular disease certainly is – and is hugely affected by stress. But the psychosocial issues themselves are never mentioned.

Why is this important? It is important because there is inevitably a bias towards counting the riskiness of childbirth and abortion simply by their immediate obstetric complications. Although the data on these are unreliable in the case of abortion (and to some extent for live birth), it is possible that childbirth presents greater immediate obstetric risks than early abortions (for later abortions, things are less clear). But this entirely ignores the myriad ways in which pregnancy outcomes – and legislation relating to them – can affect one’s mental and physical health – indeed, across generations.

Another reason this is important is because most of the leading causes of death – if not all of them – among younger women are significantly affected by pregnancy outcomes. For example, the 5 leading causes of death of women aged 20-44 are unintentional injuries, cancer, heart disease, suicide, and homicide. All of these are strongly affected by pregnancy experiences, in ways which would very rarely show up on death certificates. It is known that women having an abortion are far more likely to die from unintentional injuries, suicide and homicide than women completing a pregnancy, other things being equal (see evidence below). A significant proportion of this disparity is thought to be causally related to the experiences of pregnancy, childbirth or the abortion. Having a completed pregnancy at a younger age is known to be protective against breast cancer, and most cervical cancer is caused by sexually transmitted disease, whose prevalence is known to increase in countries legalising abortion. Heart disease is hugely affected by psychosocial stressors (see the section on abortion and mental health), smoking and alcohol intake, which are all associated with abortion. As we will see below, death from heart disease is significantly associated with pregnancy outcomes. As I have suggested in ‘How abortion harms women’, there are myriad ways in which the legalisation of abortion contributes to poor psychosocial health through facilitating family breakdown, poverty, and so on. Some of these deaths can be quantified fairly easily; others are very difficult. They show how even women who do not have abortions can be adversely affected (in some cases dying) as a result of the societal and cultural changes resulting from the legalisation of abortion. Moreover, deaths where abortion is a contributing factor can occur many years later, and even across multiple generations (since e.g. poverty, family breakdown, and mental health difficulties and other conditions are often transmitted down multiple generations).

As I describe under ‘How abortion harms women’, the legalisation of abortion has been strongly linked with increased family breakdown and poverty, which both affect physical health in a wide variety of ways. It is also associated with preterm birth, the leading cause of death for under 5s worldwide, and increased sexually transmitted diseases, which can cause significant morbidity and even mortality in certain cases (e.g., HIV/AIDS). Of course, legalising abortion also increases the number of pregnancies dramatically, thus subjecting more women to the risks of both pregnancy and abortion.

This is only a brief summary, but it is clear that there a huge number of ways in which both abortion and its legalisation can significantly cause increased mortality, probably by a long way exceeding the very low number of women who would die from unwanted pregnancies if abortion were to be restricted. Most of these mechanisms will never show up in death records, of course.

It is therefore helpful to look briefly at the studies which look at overall mortality after abortion and pregnancy, which offer a more holistic picture (I address the question of maternal mortality rates and how they relate to legalisation under ‘Does legalising abortion prevent women dying from backstreet abortions?’). There are a number of record-linkage studies – where death records are matched to pregnancy outcomes throughout a country or state, giving a full dataset – in the US, Denmark and Finland. These show repeatedly that, for various reasons, the mortality rate after continued pregnancy is significantly lower than the mortality rate for both non-pregnant women and women having abortions, and this trend becomes stronger for women having multiple births or abortions. For example, in the most recent Finnish study of this kind,  the risk of death after abortion was triple the risk of death after giving birth, mainly because of a dramatic increase in the risk of accidents, suicide, and homicide. Even the risk of dying from medical conditions was marginally higher after an abortion than after giving birth, however – for reasons such as lower alcohol and drug abuse, increased health monitoring, and increased healthier behaviours more generally. The risk of a non-pregnant woman dying was double the risk of a woman giving birth.

These studies typically are not able to control for prior physical health, which is a major limiting factor (though many control for age, and some for prior psychiatric history). However, they are at the very least consistent with the idea that abortion causes a higher rate of deaths from suicide and accidents (often from drug use). Given that the studies claiming that birth is much more dangerous than abortion are so problematic in the ways described above, there is no convincing basis for making this sort of claim. At best, we do not know which is overall riskier. But in fact there is powerful evidence that abortion is riskier once indirect mechanisms are taken into account.

Reardon and Thorp have a helpful systematic review of the available record-linkage studies, and a fairly comprehensive overview of the ways in which abortion can increase the various causes of death. Calhoun has a good summary of some of the specific measurement problems with Raymond and Grimes’ initial study.

Do women die from not receiving abortions?: Savita and Izabela

In recent years, there have been major headlines – and protests, and even legislative changes – regarding the deaths of women who did not receive abortions despite their life begin at risk. In Ireland, Savita Halappanavar died of chorioamnionitis (effectively infection of the pregnancy itself – membranes and amniotic fluid; hence delivery of the baby is required to save the woman’s life) after not receiving an abortion/having the baby delivered. In Poland, more recently, a woman known only as Izabela died of septic shock, ostensibly after not receiving an abortion, though many of the details remain unclear. In Ireland, it is fair to say that this was one of the major, if not the major, event precipitating legalisation of abortion in the country. In Poland, there have been significant protests domestically and even internationally, since the media has tied this death to Poland’s tightening of its abortion law to remove disability-selective/eugenic abortion as a legal ground for abortion.

The deaths of both of these women are tragedies, and were likely preventable in both cases. Clearly something has gone wrong in both cases, that should be fixed. But what exactly was wrong?

In both cases, we should be clear that – as with virtually every other country in the world – abortion to save the life of the mother was legal. If there was a genuine threat to the mother’s life, as seems to be the case on both occasions, abortion/delivery was legal and should have been performed. There was no legal obstacle to doing so.

The very most that could be said, therefore, is that the law can sometimes be unclear exactly how much of a threat the pregnancy has to be to justify early delivery (leading to the baby’s death), and there may in theory be reasonable clinical disagreement about how urgent the situation is. If so, there seem to be multiple fairly obvious solutions: clear guidance on when delivery is indicated for the few conditions where these problems can arise (chorioamnionitis, pre-eclampsia, and so on) or even a centralised on-call emergency service which physicians could utilise to obtain authorisation for an abortion in cases of doubt. In practice, this is probably unnecessary, since it is generally so obvious that delivery is required in emergency situations that any competent clinician should not need to discuss it with such a service. But such a service – or clear set of guidance – would minimise the chances of these issues arising from incompetent clinicians or rare genuinely difficult clinical scenarios. Such suggestions do seem to have been made by Izabela’s lawyer.

In the case of Savita, the issues clearly identified by the investigation were incompetent clinical recognition of the condition and incompetent clinical decision-making, among other things. I hesitate to speculate on exactly what went wrong in the case of Izabela, since we know so few details – this is one part of why it is so grossly irresponsible for people to use her case in this way.

Let’s look at specifics in both cases.

In Savita’s case, as I mentioned, Irish law clearly allowed abortion in these cases – and this possibility was clearly exercised for many years. The investigation identified 3 ‘key causal factors’ leading to her death: 1) inadequate assessment and monitoring preventing the team from recognising her deterioration; 2) failure to offer all management options to the patient; and 3) non-adherence to clinical guidelines relating to the prompt and effective management of sepsis, severe sepsis and septic shock. While 2) relates to the failure to deliver the child – and delivery of the child might well have saved Savita, though not necessarily – this failure was not because the law prohibited it, but because the clinical team made major errors in recognising the severity of the situation. The report noted that ‘Appropriate monitoring and evaluation of the changing clinical presentation with appropriate clinical investigations would likely have lead to reconsideration of the need to expedite delivery’ – that is to say, if the clinical team had made a correct assessment of the clinical situation, they would likely have delivered the child in accordance with the law. This was fundamentally a clinical failure, not a failure of the law on abortion.

For a more comprehensive examination of Savita’s case, see my blog post on it here.

The circumstances leading to Izabela’s death are more complicated – or, rather, less certain. Very few facts are known at all, with only a few media reports of comments from the lawyer or family members. As I said, this is part of what makes the exploitation of her death for ideological purposes so grossly irresponsible. This is made worse by the fact that almost all the information we have is in Polish, making it easier for non-Polish media outlets to exaggerate, omit or fabricate various facts.

Again, Polish law allows – and always has allowed – abortion where the life of the mother is at risk. As Izabela’s lawyer pointed out, the threat does not even need to be direct or serious. This did not change with the recent tightening of the abortion law to prohibit disability-selective/eugenic abortion. The lawyer – our main source trying to connect the law with her death, even said: ‘I am far from saying that the death of this patient is a simple consequence of the judgment of the Constitutional Tribunal. The wording “consistency” means a cause-and-effect relationship, that is, if it were not for the sentence, the patient would not die. We cannot talk about such a relationship here’. The lawyer at the same time noted that an investigation was underway for ‘medical error’ – most likely the same sort of errors as in Savita’s case, at another point claiming that medical error was likely the main problem. The hospital in question had previously been criticised and fined for bad conditions in general. And the preliminary report suggests that the hospital had similar problems at the time of Izabela’s death, with clinical failings specifically responsible for Izabela’s death. In summary, all the evidence we have suggests that the case was very similar to that of Savita.

It might be thought that, since the baby had a disability which eventually led to the life-threatening situation, allowing abortion for disability would have prevented Izabela’s death by allowing her to abort long before her life was threatened. The problem with this claim is that according to both the lawyer and Izabela’s mother, Izabela wanted to keep the baby despite its disability. Izabela’s mother even claimed that the baby itself was healthy according to the autopsy. Although it remains unclear which condition the baby was thought to have had, a recent New York Times article pointed out that many (though not all) prenatal tests are usually wrong – not only sometimes – when they detect a disability.

Perhaps some of these claims are false. But even if so, that only strengthens my ultimate point, which is that we simply do not have anywhere near enough details about this case in the public domain to say what was the cause of her death, whether the baby was healthy, whether she would have been saved by a different law, and so on.

For this reason, it is wrong – in my view, appalling – to exploit Izabela’s tragic death in service of liberalised abortion laws in Poland or elsewhere. There is no evidence that she would have been saved by a different law, or that she wanted an abortion for disability. Even if there were, there are alternative remedies to prevent deaths attributable to poor clinical decision-making in these situations – clear guidance, or emergency on-call doctors/lawyers to advise in situations of doubt. I think it is fair to call for some of these measures. But to suppose that the law needs to allow for abortion in wider circumstances is completely erroneous and exploitative.

In summary, in neither case was the law responsible for the deaths of Savita or Izabela. In fact, in both cases, the law already clearly allowed for delivery of the child/abortion if the mother’s life was at risk, as it was in both cases. To the best of our knowledge, neither mother sought an abortion until that point, and at that point it was clearly legal. It is highly irresponsible, degrading to the victims, and – in the case of doctors – highly unprofessional to exploit these deaths for ideological purposes.

Does legalising abortion prevent women dying from backstreet abortions?

Perhaps the most common argument for legalising abortion throughout history is that women will get abortions either way; and if it is illegal they will be unsafe and women will die as a result. Hence, it is better to legalise abortion even if you oppose it. This argument is supposed to be powerful because it doesn’t rely on arguing a controversial moral position on abortion. It is a simple practical argument.

Since I’m writing a book on this topic I’ll try my best to keep this as a summary overview, but will write more in time. I’ll just survey the key points rather than giving a full answer. You’re welcome to contact me for more details.

The first thing to say is that one of the key premises is known to be false: abortion restrictions do, in fact, stop abortions. For more on this, see ‘Do pro-life laws work?’

Second, deaths from abortion are frequently fabricated, outdated or misrepresented. For a recent example of fabrication, see here. More commonly they are outdated – from the 1980s or (usually at best) 1990s, since when maternal mortality and abortion mortality have changed dramatically. And they are almost always misrepresented. Sometimes this is by taking the upper end of an estimate rather than the average, or most recent, or most robust. More commonly, almost every statistic refers to ‘abortion’: to most people, ‘abortion’ means ‘induced abortion’. But it also includes spontaneous abortion: that is, miscarriage. In some statistics it even includes ectopic pregnancy as well. So any woman who dies from a miscarriage (these can cause infection and bleeding) or ectopic pregnancy is included in deaths from ‘abortion’. These are then attributed to ‘unsafe abortion’, as if all these deaths were caused by backstreet abortions. Since there is significant evidence that most of these deaths are caused by ectopics and miscarriages, the number attributable to backstreet abortions is usually less than half this number/percentage.

Third, most abortion mortality can be eliminated with good post-abortion care. This is why abortion mortality was reduced to minimal levels in most Western countries before abortion was legalised.

Fourth, most abortion mortality has been eliminated by safer methods of illegal abortion. Of course, unsafe methods remain. But by and large, misoprostol is widely available in the developing world, or surgical abortions (as they were in the US prior to legalisation). And this is widely regarded as safe: so much so that the US Women’s March recently asked people not to use coathanger imagery because it would ‘reenforce right-wing talking points that self-managed abortions are dangerous, scary and harmful’. So what was once the foremost argument for abortion legalisation (and still is in most of the world) is now branded a ‘right-wing talking point’.

Fifth, legal abortion has moved in the opposite direction, moving towards self-managed abortion with misoprostol +/- mifepristone with no in-person contact with a healthcare professional (see ‘Is telemedicine abortion safe?’). Hence illegal and legal abortion are becoming increasingly similar in safety.

Sixth, legalising abortion does not necessarily prevent illegal abortions. In fact, in most countries with studies on legalisation, legalisation has not resulted in any decrease in illegal abortions. It has only resulted in an increase in legal abortions for people who otherwise would not have had one.

Seventh, legalising abortion increases the number of overall unwanted pregnancies, thereby exposing more women to the risks of either pregnancy or abortion.

For all these reasons, the empirical evidence shows a variety of trends:

1. Affluent countries with pro-life laws have very few deaths from abortion. Malta has had no maternal deaths from any cause in the last 10 years, and Poland has the lowest maternal mortality rate in the world. Likewise with Chile, South Korea, (pre-legalisation) Ireland, and many parts of the Middle East.
2. Less affluent countries with liberal abortion laws have many deaths from abortion – South Africa, Mozambique, Ethiopia, Ghana, India, and so on.
3. When abortion is legalised, the trend in maternal mortality and abortion mortality is almost always virtually unchanged. In a few cases, mortality and morbidity increase upon liberalisation, as in the Netherlands, Rwanda and Ethiopia.
4. Countries banning abortion continue to see a gradual decline in maternal mortality and abortion mortality.
5. Hence, the overall picture is that women dying from abortion is a function not of the legal status of abortion, but the quality of emergency obstetric care.

All the references for these claims can be found in my published paper on Malawi, summarised on the Journal of Medical Ethics blog.

In addition, of course, as I pointed out in ‘Abortion is 14 times safer than childbirth’, abortion and its legalisation cause increased mortality of women and children in a variety of other ways. This was particularly noticeable in South Africa, where legalisation was followed by an enormous increase in maternal deaths from HIV/AIDS. While there would have been an increase anyway, we know that abortion legalisation leads to a 60% increase in STD cases, so this probably caused many more deaths than otherwise would have occurred.

Finally, of course, even if legal abortion did reduce maternal mortality, and even if it made no difference to the number of abortions, it would still be unjust. It used to be argued by defenders of the slave trade that keeping the slave trade ‘safe and legal’ was better than unsafe and illegal. They even argued this was better for the slaves: by regulating the trade, you could ensure that ships had sufficient space so that the people kidnapped into slavery to move around, as compared with the horrific conditions in the illegal trade. Of course, this is ridiculous: the slave trade was an affront to the dignity of human beings taken into slavery, and the empirical claim (that laws do not suppress the trade) is implausible by a simple look at the evidence. Both are true in the case of abortion.

As I said, I am working on a book project on this topic, and I know there are many related empirical claims worth examining in more detail. I am happy to answer further questions if needed.