Following the May 2018 Irish referendum largely in favour of abortion, only two European countries, Malta and Andorra, still do not give women the right to decide to end a pregnancy in the first weeks. Outside of Europe, the situations are highly diverse, but abortion remains illegal or restricted to certain conditions in a great number of countries, notably those of the South. This situation not only runs counter to the right of each woman to control her own body but contradicts other internationally recognized rights, such as the rights to equality, health, and free and responsible decision on the number and spacing of children. This “Overview of a Population Question” offers a vast panorama of the issues surrounding abortion, the difficulties in measurement, and the solutions to address them. It reviews current legislation, the most recent data on abortion frequency and methods, as well as the consequences of abortion on women’s health and lives.
1Abortion is an ancient and universal practice. Throughout history, it has taken different forms in different political, social, and cultural contexts. Laws on abortion around the world vary; in some countries, it is available to women on request, while in others it is totally outlawed. The liberalization of abortion is the subject of intense controversy and, once established, is sometimes challenged. Some defend access to abortion as a human right, a woman’s right, a sexual and reproductive right, and a right to health given the dangers of illegal abortions, while others condemn it in the name of the embryo’s right to life.
2Social disapproval of abortion, which remains widespread, is expressed in many ways: from denying abortion rights and omitting this issue from the international agenda – for example, it goes unmentioned in the United Nations Sustainable Development Goals – to implementing (by American conservative administrations) a “global gag rule” to block funding for abortion-related programmes and organizations (Singh and Karim, 2017; Starrs, 2017). It is also expressed in women’s reluctance to talk about their abortions. In certain countries, it results in women being punished and sometimes imprisoned; in discriminatory treatment at health centres both in the performance of abortions and in treatment for complications; and in difficulty finding qualified health professionals to carry out the procedure.
3The social condemnation of abortion is related to the conception of women’s role in society. Those who consider motherhood to be women’s principal role perceive the choice not to have children as deviant (Luker, 1984). Given that gender relations remain unequal in many countries and that the vision of women is still strongly associated with childbearing, the practice of abortion is often stigmatized, albeit to various degrees in different societies.
4Abortion has always been used as a method of regulating fertility in lieu of contraception, and it has played a role in both past and contemporary demographic transitions. It has also been a key tool in certain population policies. In a number of communist countries, such as Bulgaria during the second half of the twentieth century, abortion was used as a population policy instrument in a context where access to contraception was limited. It has also played a noteworthy role in controlling rapid population growth in certain Asian countries. These policies have led to high abortion rates and even to abusive practices, such as forced abortions in China as part of the one-child policy in the late 1970s.
5Abortion is an important subject for demographers and health professionals. Unsafe abortion practices remain a major cause of maternal mortality (WHO, 2011), and the health consequences of abortions still stand at the centre of debates in many countries. Until recently, all illegal abortions were considered unsafe, but this classification has recently been questioned (Ganatra et al., 2014). While legality and safety were closely associated in the 1980s, that is no longer strictly the case thanks to the diffusion of new abortion techniques in the 1990s and 2000s, including in countries with restrictive abortion laws. As a result, three categories of risk are now recognized: safe, less safe, and least safe (Ganatra et al., 2017).
6This article provides an overview of current knowledge on abortion (for definitions, see Appendix A.1) and explores similarities and differences between situations around the world. This is an ambitious task, however, as the question is highly complex, with extremely diverse situations on different continents and in different countries. This overview begins with a description of the methods women use to end their pregnancies, which vary according to the legal context and the services available in their country. Section II looks at the diverse range of abortion laws around the world. Section III presents the terms of contemporary debate surrounding abortion, and illustrates how abortion legislation has evolved under the influence of certain groups. Section IV is methodological: it examines the data and methods used to estimate the scale of the phenomenon, particularly in countries where access is legally restricted and where there are no or few statistics. Section V discusses levels and trends in abortion in the different world regions as well as the specific situation in some individual countries. Section VI examines the process that leads to abortion, the profile of the women who have abortions, and differences in practices depending on living conditions, notably in the case of sex-selective abortion, which occurs mainly in Asia. Section VII addresses the impact of abortion on women’s lives, in terms of morbidity and mortality as well as its psychological and penal consequences. The article concludes with some avenues for future research to improve knowledge on abortion and inform political debates on its status and legalization. This text will be centred on the issue of induced abortion, which we simply call “abortion”.
I – Abortion today: from traditional to modern methods
7Historical evidence shows that women have always obtained abortions using a wide range of methods, passed on from generation to generation, even if their effectiveness was not proven (McLaren, 1990; Van de Walle, 1999). Today, the method used depends on the legal status of abortion, the duration of pregnancy, the available technology (Faúndes and Barzelatto, 2011), as well as the woman’s financial resources and access to providers. It also depends on the qualifications of the practitioner who is to perform the abortion and on the healthcare environment (medical or not).
8Abortion methods are categorized into two main types: traditional methods and medical/surgical methods.
1 – Traditional methods
9Traditional methods rely on popular or folk knowledge and are of variable efficacy. They consist in the use of the traditional pharmacopeia, of manufactured or pharmaceutical products not initially intended for abortion, as well as physical and spiritual methods (Faúndes, 2011; Guillaume, 2004; Guillaume and Lerner, 2007; Singh and Wulf, 1994; Singh et al., 2009). They are mainly used when women do not have access to medical/surgical abortion because it is illegal in their country or because they face obstacles related to their economic situation, age, or family circumstances.
10In the traditional pharmacopeia, plants are known for their supposed contraceptive or abortifacient properties, or their ability to “bring on” delayed menses (Van de Walle and Renne, 2001). Purchased on the market or prescribed by traditional healers, they are prepared either as a tea or infusion to be drunk or used as an enema, or in the form of pessaries (Artuz and Restrepo, 2002; Bankole et al., 2013; Ciganda and Laborde, 2003; Prada et al., 2011; Vallely et al., 2015).
11Some “manufactured” products are also used for their alleged abortifacient properties. These include chemicals, acidic substances such as vinegar, and caustic agents such as bleach, laundry blue, potassium permanganate, etc. Drinks that are alcoholic (e.g., wine, beer), hot, laxative, spicy, or sweet (colas), taken alone or mixed with other products, ingested at high doses, are also considered to have abortifacient properties. They are taken orally or vaginally (Guillaume and Lerner, 2007; Singh et al., 2009).
12Certain pharmaceutical products (not including products such as misoprostol and mifepristone, which are described in the next section) are also known for their abortifacient properties. These are mainly products generally contraindicated in case of pregnancy, such as antimalarials (chloroquine, quinine), hormones (Crinex, Synergon, Metrigen, etc.), aspirin or paracetamol, antibiotics, laxatives, etc. Generally taken in overdose and sometimes combined for greater “efficacy”, they may have serious side effects.
13Among physical methods, one common technique is the insertion of solid, blunt objects into the uterus in order to rupture the membrane surrounding the embryo. These may be stems or roots of plants, metallic or plastic objects such as bicycle spokes, catheters, knitting needles, spoons, pencils, buckles, clothes hangers, umbrella ribs, etc. (Puri et al., 2007). Additional techniques include massage or manipulation of the uterus, extreme physical exertion, blows, and falls (Espinoza and López Carrillo, 2003). Women also use prayers and ritual amulets or gris-gris, which are certainly less dangerous for their health.
14The legitimacy of these traditional methods is grounded in knowledge and beliefs transmitted anonymously, collectively, or by popular reputation (Sanseviero, 2003). They may be used directly by the women themselves, on the advice of family or friends (Grossman et al., 2010), or they may be prescribed by medical professionals or by providers, qualified or otherwise, such as traditional therapists (naturopaths, healers, diviners, etc.).
15Such methods often pose risks to women’s health, especially those involving chemical products, overdoses of medicines or plants, or the insertion of objects into the vagina. Their use most often results in failed or incomplete abortions and complications (see Section VII), with socioeconomically disadvantaged women – the main users of these low-cost methods in countries where abortion is illegal – paying the heaviest price (Espinoza and López Carrillo, 2003; Ouattara et al., 2015; Rashid, 2010; Sundaram et al., 2012; Visaria et al., 2004).
2 – Surgical and medical methods
16Modern abortion methods may be surgical (dilation and curettage, vacuum aspiration) or medication-based. They are used for both legal and illegal abortions, but disparities in access and quality of service, in terms of professional skills and training, are very large between countries where abortion is legal and those where access is restricted (Rashid, 2010).
17Surgical methods are generally practised under local or general anaesthetic. They are based on dilation of the cervix and evacuation of the uterine cavity by either curettage or aspiration. The choice of technique depends on gestational age and on the technical and professional capacities of the local healthcare system.
18Dilation and curettage (D&C) is performed in hospital under anaesthetic using mechanical instruments. It requires appropriate technical infrastructure and skills. This longstanding method creates risks of complications when practised by unqualified personnel, including risks of infection, haemorrhage, and even secondary sterility. For many years, it was the main abortion method and is still widely used in certain countries, such as Malaysia and Sudan, whatever the legal status of abortion (Abdullah and Wong, 2010; Kinaro et al., 2009). The World Health Organization (WHO) considers this method to be “obsolete” and recommends its replacement by aspiration and/or medical abortion (WHO, 2013), although these techniques are not yet available everywhere.
19Vacuum aspiration methods consisting in electric vacuum aspiration or manual vacuum aspiration are generally recommended up to 12–14 weeks of gestation (WHO, 2013). These widely used methods are also applied in postabortion care programmes to treat complications or incomplete abortions. They are gradually replacing D&C in many countries, such as Nigeria and Ethiopia (Okonofua et al., 2011; Prata et al., 2013).
20Methods in which medication is used to bring about abortion are referred to as “medical abortions”, or sometimes “non-surgical abortions” (WHO, 2013). They are used both for abortions and for postabortion care. Two types of medication are generally used: misoprostol, a prostaglandin (hormone) sold primarily under the trade name Cytotec, and mifepristone, an antiprogestogen, known under the trade name RU486. They are used either alone or in combination, depending on the protocols in place in each country and their availability. 
21Beginning in the 2000s, the WHO has included the association of misoprostol and mifepristone on its List of Essential Medicines for abortion, and misoprostol for treatment of incomplete abortion and postpartum haemorrhage (Kumar, 2012; Millard et al., 2015; Shah and Weinberger, 2012).  But not all countries follow these WHO directives. In countries with restrictive laws, misoprostol is often included on the List of Essential Medicines only for prescription in gastroenterology (Raghavan et al., 2012). The WHO’s (2013) manual for health systems on safe abortion describes the type of product that should be used (mifepristone and /or misoprostol), specifying dosages and routes of administration, which vary depending on gestational age.
22The use of medical abortion has expanded considerably since the late 1980s, in countries both with and without legal abortion. Conditions for the supply and use of medications vary, with consequences for the method’s efficacy and effects (Fernandez et al., 2009). In countries where access to abortion is restricted, use of medical abortion has spread over the last 30 years. In particular, misoprostol has acquired a certain “notoriety” for its abortifacient properties, whether administered orally or vaginally (Barbosa and Arilha, 1993). Even where it is not officially recognized by health authorities or available through healthcare services, women obtain it through informal networks or in pharmacies, where it is sold for other indications.
23Brazil is a good example of the informal use of misoprostol for abortion. After its arrival on the market in 1986, demand for Cytotec (a brand name of misoprostol) skyrocketed (Coelho et al., 1993). In the city of Goiânia, for example, sales tripled between 1987 and 1989 (Costa, 1998). However, beginning in 1988, the fact that it was more widely used for abortion than for the treatment of ulcers (its principal indication) sparked controversy. While some gynaecologists argued that misoprostol should be available for the treatment of incomplete abortions, others held that it should be controlled to prevent an increase in the number of abortions (Barbosa and Arilha, 1993). Beginning in 1991, the Brazilian government limited its sale in order to decrease its use as an abortifacient, with restrictions of varying severity in different states (sale on prescription only, use restricted to hospitals or certain authorized locations). This reduced official sales of Cytotec, but it also boosted sales at inflated prices on the parallel market (Coelho et al., 1993). It is still the main method of abortion in Brazil (Diniz and Medeiros, 2012).
24Its use has become widespread in other Latin American countries where the law restricts access to abortion (see the overview by Zamberlin et al., 2012), and has contributed to a decrease in maternal mortality (Shah and Ahman, 2012; Shah and Weinberger, 2012). Today, misoprostol is also widely used in Asian countries with restrictive legislation, such as the Philippines (Gipson et al., 2011). In Africa it is less common, but its use is growing in some countries, such as Uganda, Gabon, and Nigeria (Atukunda et al., 2013; Hess, 2007; Okonofua et al., 2014). It has also gained ground in Spain and Italy by way of Latin American migrants who have spread word of its abortifacient properties (De Zordo, 2016).
25In contexts where abortion is legal, such as Uruguay, Mexico City, France, etc., medical abortion is replacing methods based on aspiration and is now the main method used (Fiol et al., 2016). It represents significant progress for women and health providers. It affords women more autonomy in managing their abortion and can be performed at home (Wainwright et al., 2016), potentially with the partner present (Fiala et al., 2004; Iyengar et al., 2016). In some countries, such as France, medical abortion is under the tight control of health professionals, with two medical appointments required to obtain the medication, while in other countries the practice is less strictly regulated.
26Health professionals are generally less reluctant to perform medical abortions because they are more neutral medical acts, with less personal involvement and associated stigma. Since this type of abortion takes place outside of health centres, professionals are at a greater distance from the act than in the case of aspiration and curettage (Faúndes et al., 2004; Fiol et al., 2016). This method also allows for skills transfer between health professionals because it can be prescribed by nurses and midwives, contrary to the other methods generally practised by physicians only (Barnard et al., 2015; Olavarrieta et al., 2015; Puri et al., 2015).
27In contexts where abortion is a misdemeanour, the medical method poses less risk to women’s health, the potential complications being less serious and visible than with traditional methods. They are more similar to a miscarriage than to an induced abortion.
28Menstrual regulation is another form of abortion. It consists in uterine evacuation without prior medical confirmation of the pregnancy, in women who report recent delayed menses (WHO, 2013). This early form of abortion is performed both with traditional methods (Van de Walle and Renne, 2001) and by aspiration or medication 8 to 14 weeks at most after the last menses, depending on the method and the personnel charged with performing it. It is practised both where abortion is legal and elsewhere. This method can seem more acceptable in a context where abortion is socially rejected for moral, religious, or cultural reasons, or where there is restrictive legislation. Women will consider it as the mere regulation of delayed menses and not as an induced abortion (Faúndes and Barzelatto, 2011).
29This method is widespread in Asia. In Bangladesh, menstrual regulation services were introduced in the 1970s (Dixon-Mueller, 1988) as part of the family-planning programme, despite a highly restrictive abortion law. It is also common in India and Nepal (Tamang et al., 2014). It is estimated that 45–60% of abortions in Vietnam in the 1990s were performed in this way (Goodkind, 1994). Based on the results of the 2002 Demographic and Health Survey, Becquet (2015, p. 167) mentions that “77% of abortions are early abortions, performed at less than six weeks of gestation … which are thus referred to as ‘menstrual regulation’”. In Cuba, where abortion has been legal since the 1950s, menstrual regulation, first used in the 1980s, accounts for a large proportion of all abortions (Bélanger and Flynn, 2009).
3 – The introduction of postabortion care
30The public health problem posed by unsafe abortions led to the creation of postabortion care programmes aimed at reducing maternal morbidity and mortality. There is an international consensus around this type of care, the need for which was reaffirmed at the Cairo Conference in 1994. Article 8.25 of its Programme of Action reads: “All Governments and relevant intergovernmental and non-governmental organizations are urged to strengthen their commitment to women’s health, to deal with the health impact of unsafe abortion as a major public health concern and to reduce the recourse to abortion through expanded and improved family-planning services”. Care for women was emphasized: “In all cases, women should have access to quality services for the management of complications arising from abortion. Postabortion counselling, education and family-planning services should be offered promptly, which will also help to avoid repeat abortions” (United Nations, 1994). Such services were developed in the 1990s for the treatment of incomplete abortions, either miscarriages or induced abortions, with women receiving treatment and follow-up from qualified health professionals. They represent an important advance in countries where abortion is illegal or has been recently legalized. In the latter case, it may be difficult to respond rapidly to legislative change by establishing health programmes and training providers, so unsafe abortions may persist for some time (Fetters et al., 2008; Gebrehiwot and Liabsuetrakul, 2008; Rocca et al., 2013).
31Postabortion programmes are made possible by the introduction and diffusion of the most recent and least invasive techniques, such as aspiration and especially medical abortion, whose use is increasingly widespread (Begum et al., 2014; Bique et al., 2007; Blum et al., 2007; Dao et al., 2007). This type of care can now be provided by health professionals with varied qualifications (physicians, nurses, midwives), and skills can be transferred from physicians to midwives, whose availability at different levels in the health pyramid contributes to a decentralization of care, too often concentrated in large city hospitals. These techniques also substantially decrease hospitalization times and costs (Shearer et al., 2010). Finally, these programmes also provide women with contraceptive counselling and prescriptions in order to prevent a further unwanted pregnancy and thus another abortion. Their development in countries with restrictive laws has led to changes in how health professionals view abortion, by simplifying treatment and legitimizing the practice to a certain extent, even when abortion is illegal.
II – Abortion legislation around the world
32Abortion legislation worldwide is regularly the focus of surveys and other publications. The United Nations publishes periodic World Abortion Policies reports, the most recent of which came out in 2013 (United Nations, 2013), and in 2014 produced a report on reproductive health laws (United Nations, 2014). A very exhaustive database on abortion policies, the Global Abortion Policies database,  has also recently been set up by the WHO (the Special Programme of Research, Development and Research Training in Human Reproduction, HRP) and the United Nations (WHO, 2017). Some non-governmental organizations also contribute to this research effort: these include, among others, the Center for Reproductive Rights, which regularly produces maps on the policy situation around the world (2014, 2017), and the Guttmacher Institute, which produces overviews of the situation both at the global level and in certain countries or regions (Guttmacher Institute, 2015; Singh et al., 2009). 
33Knowledge of the legal framework governing abortion is crucial, as it is an important determinant of service provision and of women’s rights and their health. Around the world, legislation has often oscillated between periods of easing and tightening of restrictions, for moral, religious, health, ethical, or legal reasons. Situations range from a total ban on abortion to unrestricted access at the woman’s request. Between these two extremes, access to abortion is authorized under certain conditions: protection of women’s lives or their physical and/or mental health,  fetal impairment,  rape or incest, economic or social reasons,  or certain other specific causes in some countries. 
1 – A constrained right
34Access to abortion is always subject to a limit in terms of gestational age or weeks of amenorrhea. These limits can be extended or waived if the woman’s life or health is in danger and in certain other situations (rape, malformation, etc.) (CRR, 2014; WHO, 2017). Most countries allow abortion up to 12 weeks of gestation, while some allow it up to 18 weeks (Sweden), 22 weeks (the Netherlands), or 24 weeks (United Kingdom) (Bajos et al., 2004; Hassoun, 2011).
35Authorizations are sometimes required for a legal abortion. In 25 European countries, consent is required from a parent or guardian for a woman who has not reached the age of majority (WHO, 2017). In France, following the 2001 reform of the abortion law, if a woman under the age of 18 is unable to obtain such consent, another adult can substitute for the parental authority; in other countries (Italy, Denmark, Norway, Spain), this role is played by a commission (Hassoun, 2011). In 37 states in the United States, women under the age of 18 need consent from at least one of their two parents (in some cases both parents), and/or the parents must be notified.  This is also a requirement in four countries in Latin America, nine in Africa, and 17 in Asia.  When a woman is in a union, the spouse’s consent is required in some African and Asian countries (WHO, 2017). 
36To end a pregnancy for health reasons, medical approval (and sometime sworn certification) may be required, notably from one or more physicians or from a psychiatrist in case of mental health problems. Abortion after rape or incest often requires legal authorization from a prosecutor or judge and sometimes a police or medical report. These barriers to access particularly affect young women and often result in their being denied the right to abort.  Some legislation also stipulates a waiting period of several days or weeks before authorization is granted; in some cases, this applies only to minor-aged women. While no such waiting period is required in the majority of European countries, it does exist in nine countries and ranges from three to seven days (Nisand et al., 2012).  In France, the seven-day waiting period was rescinded in 2017.
37Conscientious objection on moral, religious, or philosophical grounds (Rehnström Loi et al., 2015) also impedes access to abortion. Health professionals sometimes invoke the conscience clause to avoid performing abortions or treating women with complications. Practitioners may invoke the right to freedom of conscience if they consider abortion to violate their professional ethical commitment to respect for life (Fiala and Arthur, 2014). This right is enshrined in law in some countries,  and even where it is not, many practitioners refuse to perform abortions in Europe (Heino et al., 2013), notably in Spain, Italy, and Portugal (Chavkin et al., 2013; De Zordo, 2017; Hassoun, 2011), in Africa (Lema, 2012), notably in South Africa (Harries et al., 2014), in Latin America (Casas, 2009; Diniz et al., 2014; Faúndes et al., 2004; Maroto-Vargas, 2009), and in the United States (Harris et al., 2011). Nevertheless, even where this right to conscientious objection is legally recognized, health professionals have obligations, such as referring patients to non-objecting practitioners, or treating women in critical condition (Cook et al., 2009). In some cases, public healthcare facilities are also required to have non-objecting personnel in their teams (Islas de González Mariscal, 2008). Some health professionals also declare themselves as conscientious objectors when working in the public sector but perform abortions in private practice where it is more lucrative (Schiavon et al., 2010), or for women with whom they have some personal connection (Faúndes and Barzelatto, 2011). These objectors contribute to reinforcing the stigma around abortion, especially in countries where legal access is restricted.
2 – Contrasting legislation, from highly restrictive to very liberal
38Overall, abortion laws are more permissive in the most developed countries, among which seven in ten – mainly in Europe and North America – authorize abortion at the woman’s request (Figure 1).  The gap widens with increasing differences in level of development: 16% of less developed countries have liberal legislation, compared with only 4% of the least developed countries (United Nations, 2014).
39The only grounds for abortion accepted throughout the world is that of saving the mother’s life, although women do not always have access to abortion even when their life is in danger. Levels of acceptance of abortion for other reasons again vary by level of development: nearly 90% of developed countries permit abortion to protect the woman’s physical or mental health, compared with 50–60% of other countries.
40Differences between regions are even more marked for grounds such as rape, incest, and fetal impairment: twice as many developed countries (86%) as less developed countries (41%) allow abortion on these grounds, compared with less than 30% of the least developed countries. Authorization of abortion for economic and social reasons is also markedly more widespread in developed countries: 82% allow it in this case, compared with 20% of less developed countries and 6% of the least developed countries.
Distribution (%) of countries by legal grounds for abortion and by level of development, 2013
Distribution (%) of countries by legal grounds for abortion and by level of development, 2013
41If we now consider the distribution of women of reproductive age (15– 49 years) in countries classified by their legislation and level of development, the disparities are considerable (Figure 2). While 80% of women in more developed countries benefit from permissive laws (without restrictions), only 37% of women in less developed countries (which includes countries with very large populations such as China) do, and just 6% of women in the least developed countries. Indeed, more than half of women in the last of these groups (53%) live in countries where abortion is totally prohibited or allowed only in order to save the woman’s life.
42In addition to these differences by level of development, contrasting legal situations are found across countries within particular world regions (Figure 3, Appendix Table A.1). Abortion is totally prohibited in 21 countries, but exception clauses exist in some of them (Appendix Table A.1).  Conversely, in some countries, a legally recognized condition may be interpreted in a restrictive way. For example, the right to an abortion may be denied in case of rape, on grounds that the right to life beginning at conception is superior to the woman’s right.
Distribution (%) of women aged 15–49, by legal grounds for abortion and level of development in 2013
Distribution (%) of women aged 15–49, by legal grounds for abortion and level of development in 2013
Legal status of abortion around the world in 2017
Legal status of abortion around the world in 2017
3 – Restrictive laws in Africa
43African countries inherited the restrictive legal framework of the colonial powers, whose laws and criminal code defined the conditions of access to abortion and the associated penalties. In French-speaking African countries, abortion was illegal under the French criminal code of 1810 and was regulated by the French law of 1920 that outlawed “incitement to abort and anti-conceptive propaganda”. English-speaking countries were subject to the 1861 British Offences Against the Person Act; in countries under Portuguese rule, the Portuguese criminal code of 1886 applied; in Belgian colonies, it was the Belgian law of 1867; countries such as Zimbabwe and South Africa were subject to Roman-Dutch law (United Nations, 2001a, 2001b, 2002). Today, legislation in Africa is repressive (with abortion totally prohibited or allowed only to protect the mother’s life), with little change since independence. Only a few countries have eased legal restrictions on abortion. And yet in the Maputo Protocol, ratified in 2003 by 36 African states, article 14 on health and reproductive rights stipulates that signatories shall “protect the reproductive rights of women by authorizing medical abortion in cases of sexual assault, rape, incest, and where the continued pregnancy endangers the mental and physical health of the mother or the life of the mother or the foetus”. 
44In 2017, only six out of 53 African countries, representing 11% of African women of reproductive age, permitted abortion at the woman’s request during the first trimester of pregnancy. These countries were Cape Verde, South Africa, Tunisia, Mozambique, São Tomé and Príncipe, and Angola up to 10 weeks of gestation. In 1973, Tunisia became the first African country, and the first Muslim country, to legalize abortion on request. In South Africa, the law was enacted in 1996 in the context of post-apartheid legislative changes (United Nations, 2002), although access to abortion is still limited by a shortage of abortion facilities, barriers attributable to health professionals such as conscientious objection, demands for medical examinations not required by law, and stigmatization of unmarried women (Gerdts et al., 2015; Hajri et al., 2015; Harries et al., 2014). In Mozambique, São Tomé and Príncipe, and Angola, abortion was decriminalized in the 2010s.
45As of 2017, abortion was totally prohibited in nine African countries, home to 4% of women of reproductive age on the continent, but six of these countries have exception clauses if the woman’s life is in danger (WHO, 2017). In 11 countries, abortion is explicitly permitted if the woman’s life is in danger, but for this reason exclusively in seven of them (Appendix Table A.1). In 27 countries of the region, abortion is permitted to protect women’s physical or mental health. Nearly 20 also allow it on grounds of fetal impairment, although this right is seldom applied in the absence of prenatal screening, in particular for women residing in rural areas. Women’s right to an abortion following rape or incest is recognized in 21 African countries, but it is often hard for women to exercise this right, as they must identify the perpetrator of the sexual abuse, which can be especially difficult when the abuser is a relative or member of the family circle.  In addition to the three countries that decriminalized abortion in the 2010s, 12 countries have improved their laws since the early 2000s by recognizing or expanding legally permitted grounds for abortion.  Still, access to abortion remains problematic (gestational limits, specific authorizations), as both the populations and health professionals of these countries are often unaware of these legal grounds. Even where the practice is legally permitted, it remains socially and morally stigmatized (Coast and Murray, 2016; N’Bouke et al., 2012; Ouattara and Storeng, 2014). In this context of limited legal access to abortion combined with virtually non-existent legal abortion services, women frequently resort to clandestine termination: Sedgh et al. (2012) estimate that in 2008, 97% of abortions in Africa were illegal.
4 – Similarly restrictive laws in Latin America and the Caribbean
46As in Africa, abortion laws in Latin America are restrictive, although there is also a strong movement in favour of legalization, particularly among feminist groups (Kulczycki, 2011). The influence of the Catholic Church and a strong patriarchal culture are obstacles to legislative change (Guillaume and Lerner, 2007). Since the 2000s, abortion laws in certain countries have oscillated between liberalization and restriction, with previously established rights sometimes being challenged through changes to the criminal code or the constitution (Guillaume and Lerner, 2007; Kulczycki, 2011).
47In 2013, at the first Regional Conference on Population and Development in Latin America and the Caribbean, while the status of sexual rights as human rights was reaffirmed, the question of abortion was framed in terms of public health. Countries where abortion is legal were encouraged to provide safe and high-quality abortions, and other countries were urged to consider changing laws, norms, strategies, and public policies concerning abortion in order to protect women’s lives and health (CEPAL and United Nations, 2016, article 42).
48In 2017, abortion was totally prohibited in six out of 34 countries, accounting for 7% of the female population in the region: Dominican Republic, El Salvador, Haiti, Honduras, Nicaragua, and Suriname (Appendix Table A.1). Only three countries and a US territory, representing 3% of the population of women of reproductive age in the region, allowed abortion at the woman’s request: Cuba, Guyana, Puerto Rico, and Uruguay.  Mexico is a specific case since, as a federal country, each of its 32 states has its own legislation: abortion is available on request since 2007 in only one state (Mexico City, formerly known as the Federal District), with a gestational limit of 12 weeks; all other states permit abortion at least in case of rape. Cuba was the first country in the region to decriminalize abortion, in 1965 (Kulczycki, 2011). Available on request, it is commonly practised, although contraception is also widely used. Bélanger and Flynn (2009) even refer to a “culture of abortion” in the country. In Uruguay, a 2012 law allows abortion without restriction up to 12 weeks of gestation (Wood et al., 2016). 
49Between these two extreme groups, some countries allow abortion under certain conditions: to save the woman’s life (in 13 countries, in ten of which only for this reason), to protect her physical or mental health (in eight countries, including two where it is allowed only for this reason), and on socioeconomic grounds in Ecuador and Saint Vincent and the Grenadines. It is also authorized in case of fetal impairment (in eight countries), rape or incest (nine countries plus Mexico, where this is the only restriction applicable at the national level) (Appendix Table A.1). This right to an abortion following rape is often more theoretical than real. Women do not obtain the authorization to have an abortion because they have missed the legal deadline, are pressured by religious authorities (threat of excommunication), or are denied access by legal authorities or health professionals, especially when the latter are conscientious objectors (Taracena, 2004; GIRE, 2013; Machado et al., 2015; Quintero-Roa and Ochoa-Vera, 2015).
50Since the early 2000s, in addition to Uruguay and Mexico City where abortion is now available on request, legal restrictions in the region have somewhat eased. In Colombia (since 2006) and Saint Lucia (since 2004), abortion is allowed in order to save the woman’s life and in cases of a threat to her health, fetal impairment, rape, or incest. In Argentina, the right to abortion in case of rape, which was previously limited to women with mental disorders, was expanded in 2012 to all women. In Chile, where abortion had been totally prohibited since 1989, a law adopted in August 2017 legalized it in case of risk to the mother’s life, fetal impairment, and rape. However, the law has become more restrictive in Nicaragua, where therapeutic abortion was authorized until 2006 (Kane, 2008), and in El Salvador, where it was previously permitted in order to save the woman’s life and in case of fetal impairment (CRLP, 2000). In the 2000s, attempts to extend the right to abortion (to protect the woman’s health, in cases of rape or fetal impairment) failed in Honduras, El Salvador, and the Dominican Republic because of opposition from conservative movements.
51This debate was rekindled following the 2015 outbreak of the Zika virus, which affected a number of countries in the region, particularly Brazil. The WHO estimated that three to four million women were infected. As the virus may cause fetal microcephaly, the possibility of granting infected women the right to an abortion was discussed (Rodrigues, 2016), but no legislative changes were made (Aiken et al., 2016).
5 – Contrasting legislation in Asia
52In contrast, the legislative landscape on abortion in Asia is more diverse and permissive than in Africa and Latin America. A third of countries allow abortion at the woman’s request (covering 44% of women of reproductive age). The only country where it is totally prohibited is the Philippines. All countries in Asia allow abortion to save the woman’s life, a majority for health reasons, and 14 in cases of rape, incest, or fetal impairment.
53Vietnam was a pioneer, legalizing abortion in 1945. As in other communist countries such as Cuba, the ex-Soviet countries, and China, abortion was widely used in lieu of modern contraception, which was difficult to obtain, and it remains common today (Bélanger and Flynn, 2009; Goodkind, 1994; Wolf et al., 2010). China liberalized abortion in 1957, and “abortion can be practised without restriction, at whatever gestational age, and is also sometimes coercively imposed on pregnant women regardless of the quotas defined by the official birth control policy” (Attané and Barbieri, 2009, p. 63). India legalized abortion in the early 1970s in response to high levels of maternal morbidity and mortality as well as a high birth rate, allowing the procedure for social and economic reasons, to preserve the woman’s health and life, in cases of rape, incest, fetal impairment, and failed contraception. Despite this law, the majority of Indian women are not able to have an abortion in healthcare facilities (whether public or private) because of a lack of adequate services, and significant disparities in access between states remain (Ramachandar and Pelto, 2010).
54In the last 20 years, some Asian countries have eased legal restrictions on abortion, shifting legislation from a total ban or limitation to cases where the woman’s health is in danger, to laws that include other grounds for abortion (Singh et al., 2009). In 2002, Nepalese law was drastically changed and now allows abortion at the woman’s request up to 12 weeks of gestation, up to 18 weeks in case of rape, and for any duration if the woman’s life or health is in danger (Upreti, 2014). Previously, it was tolerated only to save the woman’s life, and the agreement of two physicians was needed (Shakya et al., 2004). Women who had an abortion for any other reason could be sentenced to up to 20 years in prison (Ramaseshan, 1997). In Fiji, in addition to health grounds and social and economic reasons, abortion has been allowed in cases of rape and fetal impairment since 2009. Since 1997, Cambodia has permitted abortion at the woman’s request up to 12 weeks of gestation (Hoban et al., 2010). In other countries, changes have been more limited. In 2004, Bhutan legalized abortion to protect the woman’s life and in cases of rape, incest, or mental health problems. In 2009, Indonesia legalized abortion in three cases: to protect the woman’s life, rape, and fetal impairment. In Bangladesh, although access to abortion is highly restricted, menstrual regulation is allowed up to 10 weeks of gestation and is widely practised in healthcare facilities (Rashid, 2010; Singh et al., 2012).
55The diversity of abortion laws in Asia reflects differences in sociopolitical and demographic contexts. Population policies have contributed to the liberalization of abortion and its frequent use in a number of countries, as in China with its one-child policy, as well as Vietnam, India, and Thailand (Attané and Barbieri, 2009). Sex-selective abortion is also a factor behind the high abortion rates in certain countries, although some governments condemn the practice (see below).
6 – More permissive laws in Europe, North America, and Oceania?
56In Europe, abortion laws have evolved according to very different timetables in different countries. Russia pioneered the legalization of abortion in 1920 (David, 1992). Under its influence, other Eastern European countries followed suit in the 1950s (Blayo, 1991). In these countries, abortion was the principal means of birth control for many years, as access to modern contraception was limited and sex education was poor (Sobotka, 2003). But these laws have constantly swung back and forth between periods of easing and tightening of restrictions. Romania, which had legalized abortion in 1956, drastically reversed this move in 1966 with the Ceaus¸escu government’s adoption of a pronatalist policy (Hord et al., 1991). This reversal led to a sizable increase in maternal mortality because of a rise in clandestine abortions (David, 1992). With the fall of Ceaus¸escu in 1989, this policy was abandoned, abortion became legal again, and abortion-related mortality decreased (Hord et al., 1991; Serbanescu et al., 1995). Another reversal occurred in Poland, which had liberal legislation until 1993. Following a series of legislative changes, abortion is now allowed only on grounds of health, fetal impairment, or rape. An attempt to impose a total ban failed in 2016 following mass protests within the country and across Europe.
57In 2017, among the 43 countries of Europe, 29 allowed abortion at the woman’s request (covering 72% of European women of reproductive age), and four allowed it on social and economic grounds. The majority of countries revised their legislation in the 1970s and 1980s. Access to abortion remains limited in the other countries; it is totally prohibited in Andorra, Malta, San Marino, and the Vatican. In Ireland, it has been legal since 2013 in case of danger to the woman’s life, while the case of fetal impairment was rejected. Liberalizing Ireland’s abortion laws was put to a referendum in May 2018, where a majority of voters voted in favour of reform. The laws should be modified by the end of 2018 to allow for unrestricted abortions up to 12 weeks of gestation and up to 24 weeks in exceptional circumstances.
58Legal restrictions have also been eased in recent decades in some European countries. Since the late 1990s, Albania, Estonia, Luxembourg, Portugal, Spain, and Switzerland have allowed abortion on request. Since 2009, it has been authorized in Monaco in cases of fetal impairment, rape, or incest and to preserve the woman’s life or health. At the level of the European Union, there are no common directives regarding abortion. But a 2008 resolution of the Council of Europe invites member states to “decriminalize abortion within reasonable gestational limits, if they have not already done so”, “guarantee women’s effective exercise of their right of access to a safe and legal abortion”, and “lift restrictions which hinder, de jure or de facto, access to safe abortion”. 
59In Oceania, abortions can be performed legally at the woman’s request only in Australia and in New Caledonia. In other countries, access is limited to situations where the woman’s life or health is in danger. In Canada and the United States, abortions are permitted on request. In the United States, abortion has been a constitutional right since the 1973 Supreme Court ruling in the case of Roe v. Wade, but each state can set its own regulations on the issue, and there have been many attempts to tighten restrictions.  In 2017, more than half of all states further limited access to abortion by imposing new rules: specific regulations for healthcare facilities that perform abortions; mandatory abortion counselling and waiting period; the requirement that a parent be present if the woman is under the age of 18; and bans on the use of public funds for abortions.  These restrictions represent obstacles to abortion and have reduced the availability of abortion services in some states (Jones and Jerman, 2017). The position taken by the Trump administration as soon as it came into office, with the reinstatement of the “global gag rule”, led to a ban on all funding of non-governmental organizations and other groups with programmes providing abortion services or information on abortion, in the United States and in the Global South (Starrs, 2017). 
7 – Practices of circumvention
60This international overview of abortion legislation highlights major disparities across the world. Knowledge of the law varies substantially. According to a recent study, depending on the country, between 0% and 71% of women have a good understanding of the law (Assifi et al., 2016).  Knowledge is quite poor in countries where the law has recently changed (South Africa, Ethiopia, Nepal) but also in places where it has been established for longer, as in India and Armenia. There are also considerable differences within countries, depending on women’s level of education and place of residence. This is the case, for example, in South Africa, where women who have completed some higher education are twice as likely to have good understanding of the law as women with no formal education (76% versus 32%), and women living in urban areas are better informed than those in non-urban areas (62% versus 39%).
61When faced with obstacles to legal abortion in their country, women often circumvent the law. One strategy is to travel to a nearby country or region where the context is more favourable, either because the law is more permissive or because the gestational limit is longer.  In Europe, many women travel to England, the Netherlands, and Spain, where abortions are permitted until relatively late in the pregnancy. Italian women often travel to other cities or regions within the country, as many physicians are conscientious objectors (De Zordo, 2017); if they are unable to do so, many resort to clandestine abortions. Irish women travel to England or other European countries with more liberal laws (Bloomer and O’Dowd, 2014). In Mexico, women travel to Mexico City, the only state where abortion is available on request (Senderowicz et al., 2016). In Chile, women travel across the border to a Peruvian city where clinics offer these services, even though legal access to abortion is restricted in Peru (Freeman, 2017). In the Caribbean, women travel to islands where safe services are available, legally or illegally (Pheterson and Azize, 2006). Another geographical strategy is provided by the NGO Women on Waves, which performs abortions on specially commissioned ships anchored outside the territorial waters of countries where abortion is illegal (Gomperts, 2002; Viall, 2017). 
62Women also circumvent abortion bans by obtaining abortifacient medication through informal networks. Pharmacists are important providers of these products, delivered either with a prescription or in more or less clandestine fashion (for a review, see Sneeringer et al., 2012), but they do not always provide information on dosage or routes of administration (Hendrickson et al., 2016; Huda et al., 2014; Lara et al., 2011; Senderowicz et al., 2016; Sherris et al., 2005; Tamang et al., 2014). Women obtain information on their use from websites or hotlines (Drovetta, 2015).  It is also increasingly common for women to obtain these medications online, where numerous websites offer abortifacient products (mifepristone or misoprostol, depending on the country). Some international NGOs (such as Women on Waves, Women on Web, and Women Help Women) and local NGOs sell and deliver these products, even in countries where abortion is illegal, enabling women to use them at home and in safety (Sheldon, 2016; Viall, 2017).
III – Positions and debates on abortion
63Abortion is often subject to social and moral disapproval. Even in countries where it has long been legal, “abortion continues to be seen as a practice that raises many ethical, philosophical, and scientific questions, and the legitimacy of resorting to it is still seen as problematic” (Bajos and Ferrand, 2011, p. 44). In some societies, abortion is considered a sign of a sexuality that deviates from prevailing norms, such as sexuality outside marriage. It reveals difficulties in preventing pregnancy, with women often considered responsible for this failure or for a lack of contraception (Bajos et al., 2002). It also enables women to refuse motherhood. For this reason, the issue of abortion is deeply intertwined with that of gendered roles and gender inequality.
64Abortion is a complex and hotly debated subject, with various interested parties defending divergent and sometimes highly radical positions. It touches on both the public and private spheres. In the debates on abortion, the desires, needs, and living conditions of women are not always taken into account.
65Debate on the legalization of abortion has focused on various questions. The recognition of abortion as a crucial public health question is very widespread. Unsafe abortions, illegal abortions in particular, are a cause of maternal morbidity and mortality (Shah and Ahman, 2012), while mortality in countries where it is legal is almost nil. Those who oppose the decriminalization of abortion reject these arguments, minimizing the number of such deaths and dismissing the link between illegality and the associated risks.
66Another recurring debate concerns the impact of legalization on the number of abortions. Opponents of legalization hold that decriminalizing abortion leads to increased abortion rates. However, scientific data show that in countries where abortion is legal and accompanied by abortion services and prevention programmes, as in Western and Northern Europe, prevalence is low and relatively stable (Sedgh et al., 2012, 2016).  This is not the case in countries where it is illegal (Lerner et al., 2016; Sedgh et al., 2012). Also often debated in the context of abortion are women’s rights and the respective rights of the woman and the fetus.
1 – Actors in the debate
67Many actors may become involved in the debate surrounding abortion: the State and its executive, legislative, and judicial branches, other political actors, health professionals, civil society groups (NGOs, associations, feminist groups), religious institutions, international bodies, researchers, the media, legal professionals, and others (Hessini, 2005; Lerner et al., 2016). Broadly speaking, there are two contending positions: one described as “pro-choice” and one as “pro-life”. The first is seen as defending women’s rights and human rights, while the second as defending the life of the fetus without giving much consideration to women’s lives (Faúndes and Barzelatto, 2011; Ortiz Millán, 2009).
68State and political bodies (senators, deputies, legislators, political parties, supreme courts, etc.) are key actors in this debate: it is up to them to define the legal status of abortion, debate the constitutionality of the law, determine any penalties, and propose legislative or constitutional changes.
69Health professionals are also important actors in the debate, as they play a role in their patients’ reproductive decisions, grant or deny the right to an abortion, and participate in defining and implementing health policies. They generally recognize abortion as a public health problem where it is illegal as well as a social justice issue, as the most disadvantaged women are much more exposed to the risks of unsafe abortions. The legalization and practice of abortion are controversial issues almost everywhere, although to differing degrees in different countries. While some medical professionals favour abortion, others oppose it or are reluctant to practise it, judging it to be contrary to their medical ethics based on respect for life and their perception of the fetus as a human being. On the basis of these convictions, founded on moral, cultural, or religious precepts, they may refuse to perform legal abortions or to treat women with complications, and may even report them to the authorities (CRLP, 2001). Some declare themselves as conscientious objectors (see above).
70Many national, regional, and international NGOs are active participants in the debate and in activist campaigning. Some lobby for the right to abortion to be recognized as a woman’s right, and denounce the punishments and prejudice endured by the women concerned. In some cases, they run programmes to reduce maternal mortality through postabortion care by providing information on medical abortion (through a hotline or a website) or by providing services (Gomperts, 2002). To support their advocacy and defence of the right to abortion, some gather data on abortion trends and carry out opinion surveys in specific populations (medical professionals, women, etc.). Anti-abortion groups are highly active in defending the right to life beginning at conception, either through advocacy (political debate, demonstrations, websites) or through attempts to block access to abortion in countries where it is legal.
71Religious authorities have a wide variety of positions on the issue: the texts of all religions can be interpreted in different ways, leading to positions that range from progressive to conservative. For example, the Catholic Church opposes abortion in the name of defending the right to life beginning at conception, and considers fetuses as human beings. Church doctrine likens abortion to homicide, and those who perform abortions or who support others in doing so can face sanctions and even excommunication.  Where the Catholic Church is powerful, as in many Latin American countries, its relations with governments contribute to a conservative political climate. The Church can have a marked influence on sexual and reproductive health policies, especially those concerning contraception, emergency contraception, abortion, and medically assisted procreation. It sometimes plays a decisive role in debates on the legalization of abortion, demanding that access be restricted or even opposing any form of legal abortion, whatever the grounds. And yet in surveys, Catholics generally state that religion should not interfere in sexual and reproductive life, and that they support the legalization of abortion in certain circumstances (Aldaz et al., 2013; Felitti, 2015). For example, in a survey performed in Mexico in 2014 on a representative sample of 2,700 Catholics, a majority was favourable to abortion: 80% of respondents favoured its legalization if the woman’s life is in danger; around 70% in case of rape, if the woman’s health is in danger, or if she is living with HIV; and 57% in case of fetal impairment. More than half of all respondents (53%) supported legalization in all circumstances (Católicas por el Derecho a Decidir, 2017). Within the Catholic Church itself, there are also diverse textual interpretations, with some currents raising questions about when a human life actually begins and defending women’s rights, such as the NGO Católicas por el Derecho a Decidir (Catholics for the right to decide) (Lerner et al., 2016). Abortion is not prohibited in all countries where Catholicism is the dominant religion, as in France in 1975 when the law was changed to allow abortion on request, or in other European countries such as Portugal, Spain, and Italy.
72Abortion is not explicitly mentioned in the Koran, which considers life as sacred (Yari et al., 2011). Tolerance towards abortion differs between different schools of thought, and within them, depending on the circumstances and the duration of pregnancy. The “ensoulment” of the fetus which, according to different schools of thought, occurs at 40, 90, or 120 days after conception (Hessini, 2007), serves as a watershed. Before this time, abortion may be allowed, while afterward it is forbidden unless the woman’s life is in danger (Alamri, 2011). Gruénais (2017, p. 190) cites a physician’s contribution to the debate on the legalization of abortion under certain conditions in Morocco. He recalls that in the Maliki tradition, the predominant religious current in the Maghreb, “abortion is forbidden after 40 days (6 weeks), except in cases of urgent necessity, which must be debated”. In the MENA region (Middle East and North Africa) where Islam is the state religion, the legal status of abortion ranges from availability on request in the first trimester (Turkey and Tunisia) to availability on specific grounds only (physical and/or mental health in six countries, fetal impairment or rape in three countries). All countries allow abortion to save the woman’s life, although actual access to abortion is not always guaranteed.
73In Judaism, positions range from highly strict to more liberal. As emphasized by Khorfan and Padela (2010), while Judaism assigns supreme value to human life, the fetus is considered as a “prehuman” life and not a complete human being; only at birth does it acquire its full rights. But where abortion is prohibited, Jewish law makes an exception if the woman’s life or health are threatened. 
74These three religions share a common characteristic: while they tend to adopt conservative positions overall, in practice they typically allow abortion for certain reasons. Membership of a religion does not necessarily entail agreement with all its values, and the behaviours of different populations are notably influenced by their level of religiosity. But many other factors play a role, such as levels of education (of the woman and the couple), economic resources and social capital, as well as political and legislative positions in different countries (Morán Faúndes, 2015).
2 – Abortion as a public health problem: the positions of international organizations
75International organizations have confronted the question of abortion at various conferences, most notably at the International Conference on Population and Development in Cairo in 1994, but “none of these conferences has agreed on a right to abortion, which has explicitly been said to depend on national authorities” (Ouattara and Storeng, 2014, p. 111). The international debate has focused on the health consequences of unsafe abortions, which are considered a major public health problem. The report of the Cairo Conference urges governments and NGOs to “strengthen their commitment to women’s health, to deal with the health impact of unsafe abortion as a major public health concern and to reduce the recourse to abortion through expanded and improved family-planning services”, and to ensure “quality services for the management of complications arising from abortion” (United Nations, 1994, § 8.25). It also emphasized abortion prevention through universal access to family-planning services, and the need for governments to ensure that individuals can exercise their sexual and reproductive rights. Recognized as human rights, these include the right to decide freely and responsibly on the number and spacing of one’s children. These recommendations, reaffirmed at the Fourth World Conference on Women in Beijing in 1995, place emphasis on social discrimination: the women most affected by unsafe abortions tend to be young and poor. While these conferences promoted the recognition of sexual and reproductive rights as human rights, they continued to treat abortion mainly as a public health problem. Nevertheless, they had a positive impact, contributing to the definition and implementation of postabortion care for medical complications from induced or spontaneous abortions, particularly in countries where legal access is restricted (Rasch, 2011).
76In short, the right to abortion is not yet recognized at the international level as a woman’s right, despite the demands of NGOs and feminist movements (Haslegrave, 2004; Ouattara and Storeng, 2014), and despite the reiteration of those demands at regional conferences, such as the 2003 Maputo Conference (see above) and the 2013 Montevideo Conference (CEPAL and United Nations, 2016).
77Following these conferences of the 1990s, synergies developed among different bodies within the United Nations, such as the Centre for Civil and Political Rights and the Committee on the Elimination of Discrimination against Women (CEDAW). Together with governments and NGOs, they have contributed to improving the laws that limit access to abortion in various countries (Hessini, 2005).
78The World Health Organization took part in this movement, recognizing abortion as a public health problem (WHO, 2007). Combating unsafe abortion is one of the priorities of its global reproductive health strategy, a strategy based on international human rights treaties (WHO, 2013). The WHO is a key actor in formulating recommendations for providing abortion services and treating associated complications. Among other initiatives, it has published practical guides on abortion methods and has included the pharmaceutical products required for medical abortions on its Model List of Essential Medicines (WHO, 2013, 2015, 2016a).
79The United Nations does not explicitly mention the question of abortion either in its 2000 Millennium Development Goals or in its 2015 Sustainable Development Goals, although both initiatives set a goal of substantial reductions in maternal mortality (Basu, 2005). Target 3.7 within the third SDG, on health, refers for example to “universal access to sexual and reproductive health care services, including for family planning, information and education” but without mentioning abortion. Objective 5 on gender equality calls for “universal access to sexual and reproductive health and reproductive rights as agreed in accordance with the Programme of Action of the International Conference on Population and Development and the Beijing Platform for Action” – two earlier documents that clearly treated abortion as a public health issue.
3 – Abortion: a woman’s right
80Women’s movements in various countries, whether the feminist movements of the 1960s (notably in France) or women’s groups at the international level in the 1990s, have demanded the rights to abortion and contraception.  Through slogans such as “My body is mine”, “My body, my choice”, “A child when I want, if I want”, and “Let her decide”, women have defended their right to control their own bodies, to enjoy safe sexuality dissociated from procreation, and to decide freely whether to have children or not. They have argued that childbearing should be individually desired and chosen, and no longer held as a women’s duty and a biological inevitability.
81Since the 1960s, the place of women in society has changed in many countries, with an increase in labour force participation, in levels of education, and in age at marriage. Childbearing no longer occupies the same place in their lives. The development and growing availability of family-planning programmes and the liberalization of abortion have supported these changes in women’s relationship to childbearing. A new “procreative norm” has arisen, enabling women and couples to choose the right time to have a baby (Bajos and Ferrand, 2011). The decision to terminate a pregnancy is often explained by the absence of a plan to become parents, of a stable partner, or of a partner who wishes to have a child (Bajos and Ferrand, 2006). In this context, the legalization of abortion affords women greater control over their reproductive lives, as contraception cannot prevent all unplanned pregnancies. Segdh et al. (2014) estimate that in 2012, 40% of pregnancies around the world were unplanned and that half of these ended in an abortion.
82However, many societies continue to be based on patriarchal values and exercise control over women’s sexuality and reproduction. In these societies, women’s demands for the right to control their reproduction are limited. In such contexts, arguments on the legalization of abortion focus more on health than on women’s rights.
4 – The rights of the embryo
83Discussion of the rights of the embryo leads to questions about when a life begins. Interpretations of this issue differ between scientific, legal, bioethical, moral, and religious perspectives. What is the status of the fetus? Is it a human being, with the same rights as those who have already been born? That is the premise of opponents to the decriminalization of abortion, for whom life begins at conception or at fertilization. They see the embryo as a potential human being and not a developing being. This position is not based on the notion of fetal viability, i.e. the ability to live independently outside the mother’s womb. According to the WHO, a fetus is viable after at least 22 weeks of gestation or when it has reached a weight of 500 grams (WHO, 1977). This limit has been debated in some countries, as technological progress in medicine improves the survival chances of severely premature infants (on this issue, see Pignotti, 2009). Bioethicists point out that the embryo’s neurological development only begins at the twelfth week of gestation.
84The legal protection of the right to life applies to individuals from birth to death but not during the pregnancy. In English law, the term “person” is only used after a fetus has been born alive. International treaties on the rights of persons do not mention the embryo or the fetus and do not treat the rights of individuals as applicable before birth (Cook and Dickens, 2003). According to the Universal Declaration of Human Rights (1948), “Everyone has the right to life”, and the European Convention on Human Rights (1950) contains a similar clause, but neither document defines the notion of “life”. The American Convention on Human Rights (1969) mentions that “Every person has the right to have his life respected. This right shall be protected by law and, in general, from the moment of conception”. The use of the qualifier “in general” leaves open the question of whether or not the right to life is to be recognized from conception.
85With increasing scientific intervention on humans and embryos, bioethical questions have emerged and spurred reflection on laws regulating abortion. As pointed out by Labrusse-Rioux and Bellivier (2002), “first of all, the words embryo and fetus were not used in legislation until medical biology and requests for the legalization of scientific or medical interventions in developing human organisms following conception required legislators to use them; previously, civil and criminal common law applied the terms conceived child or unborn child, and, more recently, human being.” In France, in 1984, for the first time, the National Consultative Committee on Ethics (Comité consultatif national d’éthique) referred to the human embryo and fetus as a “potential person”, but legislators set aside the question of the embryo’s legal status, focusing only on its scientific reality: the embryo is potentially the beginning of a human life. The complex question of how to treat the potential human life of the fetus has been resolved in some countries, such as Mexico, by weighing the rights to life of the fetus against those of the pregnant woman, and considering that the rights of the woman, already a living being, outweigh those of the embryo, which is a potential life. The debate reveals the complexity of defining the beginning of life, and the divergence of possible interpretations.
IV – Available sources and measures of the phenomenon
1 – Main sources of data
National abortion statistics
86In countries with liberal legislation, statistical data on abortions are most often provided by the healthcare facilities that perform the procedure. These figures, when they are complete, show without exception that abortion is a common phenomenon; in France, for example, a third of women have an abortion in their lifetime (Mazuy et al., 2014). By revealing that abortions are common and that all categories of women have them, abortion statistics can contribute to normalizing the practice (Kumar et al., 2009). They also provide information on abortion services offered, their safety and compliance with the law, inequalities of access, and are used to observe changes over time in unplanned pregnancies and their characteristics.
87Abortion statistics record the medical aspects of the procedure (gestational age, technique used), sometimes adding information on postabortion contraceptive counselling. They include indications of the women’s characteristics (age, marital status). Other information, such as parity, number of abortions the woman has had, her level of education, or nationality, is less common. Users of these data must take precautions (Sedgh and Henshaw, 2010). Sometimes, these figures also include spontaneous abortions or exclude some induced abortions, such as those for fetal anomalies. Similarly, these national or local statistics include non-resident women who have travelled to have an abortion in the country or region, and it is often impossible to distinguish them from residents. In some cases, as in Canada, hospital (re-)admissions for complications are also treated as abortions, leading to double counts. Finally, and more generally, in recent decades the increasing privatization of abortion services and the shift toward non-hospital facilities (notably with the growth of medical abortion) may have led to increased under-reporting. The institutions responsible for compiling abortion statistics sometimes draw on other sources of health data to fill the gaps in their systems.
88Out of the 77 countries and territories around the world in 2008 with liberal abortion laws,  only 24 – all developed countries and most of them in Europe – had national abortion statistics that were at least 90% complete (Sedgh et al., 2011).  The largest proportion of countries with liberal legislation (34 out of 77) thus had abortion statistics that were incomplete or of unknown completeness in 2008. Nineteen former communist countries only record abortions performed in the public sector.  In Canada and the United States, responsibility for abortion statistics lies with the provinces or states. This leads to missing data, as some states lack data collection systems or make little effort to obtain data from facilities that fail to report their statistics. Statistics were also incomplete or of unknown completeness in 2008 in Greece, China, Hong Kong, Japan, Mongolia, Vietnam, Cuba, Puerto Rico, and Tunisia. In Mexico City, India, Nepal, and South Africa, not only do statistical systems have difficulty counting all legal abortions, they also do not count illegal abortions, which remain common. Finally, 19 countries with liberal legislation, including Australia, Ethiopia, Cambodia, Turkey, Taiwan, and South Korea, have no statistical system or did not respond to the survey by Sedgh et al. (2011).
National surveys of abortion providers
89In countries with liberal laws but where routine statistics are flawed or lacking, surveys of abortion providers can supply reliable figures. In the United States, for example, the Guttmacher Institute produces complete figures at the national level through periodic censuses of all health professionals who perform abortions (Sedgh and Henshaw, 2010). Taking particular care to establish an exhaustive list of these professionals, they use a number of strategies to minimize the amount of missing data: out of a total 1.2 million abortions counted in 2005, 76% were drawn from the census of providers, 12% from state statistics, 9% were estimated based on information from local informants, and 3% were projected on the basis of other figures (similar providers or previous years).
90Sample surveys of abortion providers are also used to document their opinions and experiences with the practice. Another common approach is to interview women who have had an abortion, from a sample of healthcare facilities that offer abortion services, on their satisfaction with the services provided, the events that led to the abortion, the choice of technique, etc. The survey carried out in France in 2007 by the Ministry of Health’s research and statistics department (Direction de la recherche, des études, de l’évaluation et des statistiques, DREES) is one such example (Vilain et al., 2010).
91Surveys through abortion providers are also helpful in countries of the South where the practice has been legalized. In Cambodia, these surveys have made it possible to track national trends in abortion rates since its legalization in 1997 (Fetters and Samandari, 2015). In restrictive contexts, on the other hand, it is generally impossible to perform large-scale quantitative surveys from abortion practitioners, not least because they are not identified in any directory. It is sometimes possible to perform studies at the local level, as in a survey of Kenyan pharmacists using the “mystery clients” technique (Reiss et al., 2016). 
Postabortion care statistics
92In countries where abortion is illegal, statistics on postabortion care are the most widely used sources of data. These figures concern women who have had an induced abortion or a miscarriage and who come to a healthcare facility before the products of conception (embryo, placenta) have been completely expelled. These statistics serve to evaluate the implementation of the postabortion care programmes generalized in the 1990s in countries with restrictive legislation and to document their progress in decreasing maternal morbidity and mortality. They can also be used to describe the consequences of abortions for women’s health, and their costs to healthcare systems.
93The quality of these data varies, as do the health information systems of these countries as a whole. In some more developed countries in the South, researchers can compile data from hospital registers, although this information is not always aggregated or published at the regional or national level. In the least developed countries, specific surveys are needed. The standard procedure is to train healthcare workers to record the information of interest as patients arrive. Most quantitative data on abortion in the 1980s and 1990s from countries with restrictive legislation are drawn from studies of this type performed in a few hospitals, often as part of doctoral theses in medicine. Surveys of women treated for complications in healthcare facilities are sometimes added. For example, a recent multisite survey performed by the WHO in 30 countries across the five world regions asked women presenting serious complications from an abortion about the conditions in which the procedure was performed (Kim et al., 2016).
94The major limitation of such data collected upon admission for complications, aside from their often local nature, is that they cannot be generalized to all abortions. Only the most dangerous methods lead to hospitalization, and not all women with complications have access to healthcare facilities. Furthermore, there is no way to correct these hospital data, as the distribution of abortions in the population is unknown. Another difficulty with these data is that complications may be the consequence of both induced and spontaneous abortions. It is clinically impossible to distinguish them unless the method has left visible traces or sequelae (an increasingly rare occurrence thanks to the growing prevalence of medical abortion), or the woman reports that she induced the abortion herself, which is uncommon. The nature of the abortion (induced or spontaneous) thus cannot be recorded in registers.
95To circumvent this problem, studies sometimes look only at the most serious complications, which are rarely due to spontaneous abortions. The study of serious morbidity has also become increasingly important as the abortion-related mortality rate has declined. A recent review, however, has shown that data on complications from abortion (symptoms, severity, treatment) remain extremely disparate across countries and across data collection operations, making meta-analysis impossible (Adler et al., 2012). Registers of maternal deaths do not clearly identify abortion-related deaths (Gerdts et al., 2013). A recent WHO survey of severe cases treated in an obstetrics and gynaecology department (near misses ) arrived at the same conclusion (Souza et al., 2013). In other words, because families of deceased or seriously ill women are reluctant to reveal what brought their relative to the hospital, it seems that only data collection efforts aimed at identifying and following patients who arrive with an incomplete abortion can allow a rigorous counting of these severe outcomes (Kim et al., 2016). Data on abortion-related deaths and severe complications must therefore be used with caution.
General population surveys
96All the data mentioned above – statistics on abortion or complications, surveys of providers of abortions or postabortion care, and surveys of users of these services – are collected directly or indirectly from the healthcare system. General population surveys could offer a useful counterpoint to these strictly healthcare-based approaches, both in the North and the South, but with a few notable exceptions, this type of source remains to date relatively little used. In countries where abortion has long been legal and where the practice is widespread and less stigmatized (typically the countries of the former communist bloc), women report a high number of abortions in surveys. In these countries, the national Demographic and Health Surveys and the Reproductive Health Surveys serve to measure the scale of the phenomenon, probably underestimating it (Johnston et al., 2010; Westoff, 2005). For example, in Georgia, the annual abortion rate calculated from the 1999 Reproductive Health Survey is 125 abortions per 1,000 women aged 15 to 44, which is seven times higher than the rate calculated from healthcare statistics. This number remains the highest documented abortion rate in the world (Serbanescu et al., 2007). In France and Great Britain, reporting of abortion in general population surveys is also relatively good: women reported 66% of abortions in the 2010 FECOND survey in France (Fécondité – Contraception – Dysfonctions sexuelles) and 72% in the 2010 Natsal-3 survey in Great Britain (National Survey of Sexual Attitudes and Lifestyles) (Scott, 2017). However, in other countries where abortion has been legalized, the estimated rate of reporting in general population surveys does not exceed 50% or cannot be estimated in the absence of complete statistics (Rossier, 2003).
97In countries with restrictive laws, questions on abortion in general population surveys often meet with limited success. Nonetheless, a substantial number of responses have been obtained in certain countries, such as the coastal countries of West Africa, despite restrictive legislation and pronatalist values (Guillaume and Desgrées du Loû, 2002; N’Bouke et al., 2012). In addition to optional questions on abortion in the pregnancy history, the Demographic and Health Surveys include a module on the conditions of the abortion (providers, methods, locations). This module has been successfully used in recent years, for example in Ghana (Rominski et al., 2014). When the number of responses is high, one can assume that these data are relatively representative of all abortions, at least in comparison to the results of hospital surveys, which are the only available sources of information in many countries.
98Different techniques for improving response rates in general population surveys have been tested (Johnston et al., 2010). One strategy consists in first asking less sensitive questions on unplanned pregnancies before going on to record abortions (Huntington et al., 1993); more recently, audio computer-assisted self-interview (ACASI) has been used (Lara et al., 2004). However, neither of these two measures seems to have made a real difference. Using qualitative approaches either to adapt quantitative questionnaires to suit the specific context or to talk more freely with respondents during the interview seems to be more successful, at least in liberal contexts such as India, but these approaches remain relatively costly (Johnston et al., 2010). Other approaches are more promising, such as the Randomized Response Technique: the respondent is shown two binary yes/no questions, one uncontroversial and with a known probability, the other on abortion; the woman draws one of the two questions at random, reads it privately, and then gives the answer to the interviewer, who does not know which of the two questions she has answered. Another method is the Sealed Envelope Technique: the woman answers the question on abortion in private and in writing, inserts the answer in a sealed envelope; the interviewer puts it in a bag along with the other envelopes; a code links the envelope to the anonymous principal questionnaire. Another more recent variant is the List Experiment, where the sample is separated into two groups: the first group of respondents receives a list of non-sensitive characteristics and must say how many apply to them; the second receives the same list, with the sensitive characteristic added. In all of these cases, large samples are needed to obtain relatively complete responses, but which pertain only to the frequency of the occurrence of an abortion. Another approach, the Anonymous Third-party Reporting method (or the Best Friend method, in another variant), has been successfully tested in contexts where access to abortion services is difficult and where women turn to their close family and friends to find someone who can help them terminate their pregnancy (Owolabi, 2017; Rossier et al., 2006; Yeatman and Trinitapoli, 2011). Surveyed women are asked to provide a list of their close female relationships, without names, and are then asked to say who in this list has had abortions over the preceding years. This approach works with smaller sample sizes and can be used to collect other information on abortions.
99Beyond the completeness of responses, the crucial question when direct interviews are used is that of differences in under-reporting by the women’s characteristics. The few studies on this question have produced mixed results: in France, women who reported abortions in the 2000 COCON survey (COhorte sur la CONtraception) had the same sociodemographic characteristics as in national statistics (Lelong et al., 2005), whereas in the United States women from racial minorities, women with low incomes, and young women tended to under-report their abortions in the National Survey of Family Growth in 2002 (Jones and Kost, 2007). This subject deserves more systematic study. In the absence of this type of validation, the quantitative study of abortion through general population surveys remains problematic (Jagannathan, 2001).
100On the other hand, such surveys are well suited to capturing the normative climate surrounding abortion. In developed countries, research on opinions about abortion blossomed in the 1960s and 1970s as abortion was legalized. This research remains highly relevant in countries that are preparing, or have recently undergone, legislative change. In recent years, researchers have increasingly sought to measure the stigmatization of abortion, in terms of intensity and the way it is expressed, in different countries. Question modules have recently been validated for women who have had an abortion in the United States (Cockrill et al., 2013), at the community level in Mexico (Sorhaindo et al., 2016), and in Ghana and Zambia (Shellenberg et al., 2011). The question has also been studied among abortion practitioners in the United States (Martin et al., 2014).
101Given the limitations mentioned above, the contributions of qualitative studies are key to improving knowledge in this domain. Such studies provide unique information in contexts of illegality: for example, on medical abortions in Latin America (Grindlay et al., 2013), sex-selective abortion in Nepal (Lamichhane et al., 2011), or the choice of technique and provider in Zambia (Coast and Murray, 2016). Additionally, sociological and anthropological research can show how different normative contexts in relation to gender, parenthood, and sexuality affect individual abortion trajectories. In Sweden, a study with adolescent girls who had had an abortion showed that their family, friends, and partners strongly pressured them to terminate the pregnancy, and that in reality they did not have the choice of early motherhood (Ekstrand et al., 2009). Qualitative studies also explore in detail the viewpoints of different actors. They look, for example, at practitioners’ experience of conscientious objection or of task-sharing following the introduction of methods based on medication or aspiration. A few studies have focused on describing men’s representations and experiences. For example, a study in the United States showed that men’s perceptions vary considerably: some do not feel responsible for the pregnancy and are completely uninvolved in the decision to have an abortion, while for others the opposite is the case (Reich and Brindis, 2006).
102At another level, qualitative research with actors in the different institutions involved in regulating and performing abortions (activist NGOs, healthcare facilities, ministries of health) sheds essential light on the ideological issues traversing this field and their implications. A recent qualitative study in Senegal showed that the Ministry of Health was making efforts to count postabortion care as treatments for spontaneous abortions, in an ideological reframing of reproductive health as maternal health. This not only delays the opening of a debate on induced abortion – as the phenomenon remains invisible at the national level – but also obliges providers to subject the women concerned to a prolonged and stigmatizing interview in order to avoid liability in the event of a police investigation (Suh, 2017).
Estimates of the number of illegal abortions at the national level
103In countries with restrictive legislation, despite the contributions of the various sources mentioned above, the scale of abortion cannot be measured directly. To remedy this problem, the Guttmacher Institute developed the Abortion Incidence Complications Method (AICM) in the 1990s, which estimates the number of illegal abortions in a country based on the number of complications identified at the national level. The first estimates were established in Brazil, Chile, Colombia, the Dominican Republic, Mexico, and Peru (Singh and Wulf, 1994), followed by the Philippines and Bangladesh (Singh et al., 1997). In these studies, the number of complications at the national level is obtained from hospital registers. This method has since been replicated many times. When hospital registers cannot be used, less intensive methods are applied: data on the number of complications in the past months are collected from heads of obstetrics and gynaecology departments in a sample of healthcare facilities that is representative at the national level. However, a 2014 study in Zambia highlighted the lower quality of these retrospective data in comparison to data collected as women arrive at the hospital (Owolabi, 2017).
104The next step in the estimation process is to distinguish between complications from induced abortions and those from spontaneous abortions. In the AICM, a theoretical number of spontaneous abortions requiring medical treatment is calculated (derived from the number of births in hospitals in the country). This number is then subtracted from the total number of complications to obtain the number of complications from induced abortions (Singh et al., 2010). Next, to estimate the total number of abortions, the AICM interviews around a hundred experts in the country (largely healthcare professionals) on the proportion of abortions that do not end in a complication at the hospital. Each of these steps constitutes a source of uncertainty in the final estimates, and the direction of the associated biases remains unknown. But despite its limitations, AICM is the most common approach for estimating the scale of abortion at a national level in contexts where it is illegal.
105The residual estimation technique, derived from the Bongaarts model of the proximate determinants of fertility, is another way to estimate national abortion rates (Johnston and Westoff, 2010). The estimated number of births per woman avoided through the use of contraception, sexual inactivity, and postpartum amenorrhea is subtracted from the total theoretical number of births per woman. The difference between this figure and the observed fertility rate represents the number of births avoided through induced abortions, which can be used to calculate an abortion rate. This approach has not been widely adopted, however, as the model is highly sensitive to small variations in the measurement of the proximate determinants (contraception, sexual activity, and amenorrhea).
106The WHO and the Guttmacher Institute have been developing methods for estimating the number and safety of abortions at the world level since the 1990s (see Appendix Document A.1).
V – Abortion rates around the world
1 – Trends in the frequency of abortion at the regional and world levels
The number of abortions
107Using the methods described (Appendix A.2), it is estimated that today (2010–2014), on average, at the world level, 35 out of every 1,000 women aged 15 to 44 years have an induced abortion each year (Table 1) (Sedgh et al., 2016). This corresponds to a total of 56 million abortions per year, or 25% of pregnancies, or approximately one abortion for every three births.
108Rates in developing countries, where 88% of abortions in the world take place, are higher (36 per 1,000) than in developed countries (27 per 1,000) (Table 1). A large proportion of countries with liberal legislation have relatively low rates, as in Western Europe (16 per 1,000), Northern Europe (18 per 1,000), and North America (17 per 1,000), because of the existence of successful sexual education and family-planning programmes. However, an equally large number of countries with liberal legislation (for instance, those of the former Soviet bloc) have long favoured abortion over modern contraception as a way to regulate births. Although contraceptive use has been increasing and abortion decreasing in these countries, abortion rates remain relatively high (42 per 1,000 in Eastern Europe in 2010–2014). The incidence of abortion in countries with restrictive laws is high (37 per 1,000, on average); these are also countries with a less developed culture of prevention and lesser access to contraception, leading to high numbers of unplanned pregnancies (Sedgh et al., 2016). The Caribbean (59 per 1,000) and South America (48 per 1,000) stand out with significantly high incidence. Rates in the other subregions of the developing world – Africa, Asia, and Central America – are close to the world average. But underlying the relative similarity of these regional figures is a great deal of heterogeneity within regions, i.e. at the country level.
Abortion rate per 1,000 women aged 15 to 44 years, estimated in 1990–1994 and 2010–2014, by geographical area
Abortion rate per 1,000 women aged 15 to 44 years, estimated in 1990–1994 and 2010–2014, by geographical area(a) United Nations Population Division definition.
Interpretation: It is estimated that for the world as a whole, between 1990 and 1994, for every 1,000 women aged 15 to 44 years, 40 had an induced abortion each year. The confidence intervals indicate that there is a 90% chance that the real number is between 39 and 48 induced abortions.
Note: The confidence interval is a measure of the precision of the estimate. This precision depends on the number of data points used and the degree of uncertainty attributed to each of these points. Confidence intervals tend to decrease between the two periods, as more data have become available, with the exception of Latin America, which very few studies have examined in the recent period.
109Over the last decades, abortion rates for all regions combined have slightly but significantly decreased from 40 per 1,000 in 1990–1994 to 35 per 1,000 in 2010–2014. Between these two periods, the absolute number of abortions increased somewhat, from 50 million to 56 million abortions per year, because of population growth.  The small decrease in abortion rates at the world level reflects contrasting trends in developing countries, where incidence has remained nearly stable (39 per 1,000 in 1990–1994 to 36 per 1,000 in 2010–2014) and in developed countries, where it decreased by nearly half (from 46 per 1,000 in 1990–1994 to 27 per 1,000 in 2010–2014). Progress in developed countries is mainly due to sharp decreases in Eastern Europe, where the rate went from 88 per 1,000 in 1990–1994 to 42 per 1,000 in 2010–2014. Slower decreases were also observed in Southern Europe (from 37 per 1,000 to 26 per 1,000). In developing countries, however, trends remained relatively stable in all subregions, with the exception of moderate (and, given confidence intervals, non-significant) decreases in three of four subregions in Asia (Southeast Asia, from 46 per 1,000 to 35 per 1,000; Western Asia, from 42 per 1,000 to 34 per 1,000; East Asia, from 43 per 1,000 to 36 per 1,000). The spread of modern contraception in the last two decades in the countries of the South thus has not led to a notable decrease in abortions, probably because the demand for children decreased, and the increase in contraceptive use has only partially responded to this change.
The proportion of unsafe abortions
110Estimates of abortion safety in different world regions indicate that just over half of the 56 million induced abortions each year (55%) in 2010–2014 were “safe”, meaning that they involved recommended techniques and providers (Table 2) (Ganatra et al., 2017). The remaining 45% of abortions did not meet the recommended medical standards. These 25 million unsafe abortions per year pose a real danger to women’s health and lives. Around a third (31%) of all abortions were “less safe”, meaning that they were performed in conditions that, while not optimal, involved either a technique or a provider that met WHO standards. No less than 14% of abortions were “least safe”, with none of the safety criteria being met. These results clearly highlight the substantial efforts that are still needed to ensure access to safe abortions.
111Estimates of the degree of safety (Table 2) describe a situation of extreme contrasts between the most developed countries and less developed countries: in 2010–2014, 88% of abortions in the first group were safe, compared with 51% in the second. In the North, Eastern Europe stands out as the only subregion where the proportion of less safe abortions is not very low (14%), which reflects healthcare providers’ continuing widespread use of dilation and curettage. The trend in developing countries is the reverse. More than three-quarters of abortions are unsafe in nearly all subregions. The gravest situation is in Africa, where most abortions are highly unsafe; women in Africa still resort to the most dangerous and invasive methods. The situation in Central Africa is the most serious, followed by West Africa, East Africa, and North Africa. Southern Africa stands apart from the rest of the continent, as three-quarters of abortions there are safe. While in Latin America the proportion of safe abortions is similar to that in Africa, the situation there is comparatively better, as abortions tend to be in the “less safe” category because of the abandonment of the most dangerous methods in favour of the (often informal) use of misoprostol. The situation in Asia is heterogeneous, with some countries having made little progress on abortion safety, while others liberalized their legislation relatively early and offer abortion services in healthcare facilities to the entire population. Aggregating the various national situations, the proportion of safe abortions is above 50% in Southeast Asia and West Asia; the safety profile of East Asia (which includes China) is even close to that of the developed countries. Oceania is another world region with extremely heterogeneous conditions. The degree of safety is optimal in Australia and New Zealand but extremely low in the Polynesian islands.
Proportions of abortions by estimated degree of safety and geographical zone, 2010–2014
Proportions of abortions by estimated degree of safety and geographical zone, 2010–2014(a) United Nations Population Division definition.
Interpretation: The proportion of safe abortions in the world in 2010–2014 is estimated at 54.9%, and there is a 90% chance that the real figure is between 49.9% and 59.4%.
112In short, the goal of safe abortions has been reached in nearly all developed countries, with the exception of Eastern Europe. The situation in developing countries is highly diverse yet worrisome overall. These estimates indicate that safe abortions are the norm in middle- and high-income countries (according to World Bank definitions), while only 22% of abortions are safe in low-income countries. The law also plays a role. In countries where women can end their pregnancy on request, 87% of abortions are safe, compared with 25% in countries where abortion is prohibited or allowed only to save the woman’s life. In countries that allow abortion, 13% of abortions are nevertheless unsafe, not only because some providers use dated techniques but also because legalization is not always followed by the creation of a sufficient supply of abortion services. In countries with restrictive legislation, 25% of abortions are safe because women with the means to do so are able to obtain clandestine abortions from trained personnel in conditions that meet medical standards. But level of development seems to have an even greater impact than legislation. Only 0.3% of abortions in high-income countries with restrictive laws are in the “least safe” category, compared with 31% in low-income countries with an equivalent legal situation. Restricting access thus mainly disadvantages women in developing countries, especially poor women in these countries. In developed countries with restrictive laws, such as Ireland and Poland, women are better able to circumvent legally imposed restrictions. These figures show that liberalization can extend the availability of safe abortion to all social strata, provided that services of sufficient quality and number be put in place.
2 – The diversity of situations among countries with reliable statistics
113The eight case studies below (France, Spain, Bulgaria, United States, Uruguay, Nepal, China, and Tunisia) illustrate what is known about the frequency of abortion in different contexts. These countries were chosen because of the availability of statistics and published studies concerning them, and to illustrate the diversity of issues around legal abortion.
114Since induced abortion was legalized in France in 1975, various decisions have reinforced its access: reimbursement by social security in 1982, extension of the gestational limit in 2001, and legalization of medical abortion in private practices in 2004 and in family-planning centres in 2009. According to DREES and INED statistics, the frequency of induced abortion decreased slightly between 1975 and 1990.  Over this period, use of modern contraceptive methods spread, and couples were increasingly successful at avoiding unplanned pregnancies (Mazuy et al., 2014). Since the early 1990s, however, abortion rates have remained stable. In 2016, 14 of every 1,000 women aged 15–49 years in metropolitan France had an abortion. Rates vary by a factor of two between regions, with the highest rates in Île-de-France and the South (Vilain, 2017). The abortion rate in French overseas territories in 2016 was 25 per 1,000.
115One of the most notable recent trends in abortion is the decrease in the ages of the women who have them. Since the mid-1990s, abortion rates below the age of 25 years have increased while remaining stable at other ages. This tendency must be understood in the context of delayed childbearing. Young women have fewer and fewer unplanned pregnancies and end them increasingly often when they do occur (Mazuy et al., 2014). Another major shift is in the techniques used, with the proportion of medical abortions increasing from 31% to 64% between 2001 and 2016 (Vilain, 2017). This switch to medical methods has had a major effect on the locations where abortions occur since 2004 (when abortions in non-hospital facilities became legal): in 2016, 18% of abortions took place outside of hospitals. Gestational age at abortion decreased markedly over the 2000s for the same reason, from 7.1 weeks in 2001 to 6.4 in 2011 (Mazuy et al., 2014). A final significant trend is the increase in repeated abortions, to which we will return below.
116In 1985, Spain legalized abortion in cases of rape, fetal impairment, and risk to the woman’s physical or mental health. From that time on, women could legally access safe procedures, conditional on providing the required medical certificates or police reports. It was only in 2010 that abortion on request was legalized.
117Spain is an interesting case, as it is the only country with complete statistics where the abortion rate has recently increased: from eight out of every 1,000 women aged 15 to 44 years in 2003, to 12 in 2008 (Sedgh et al., 2011). One study of abortion registers from four Spanish regions covering an earlier period (1991–2005) shows that immigrant women’s abortion rate was three times higher than that of Spanish women and that 76% of the increase in the rate between 1991 and 2005 was due to the increase in this population (Orjuela et al., 2009). Such contrasts between women of different geographic origins have been observed in many countries of the North and are due to immigrant populations’ greater difficulties accessing contraception. A 2007 study of women accessing induced abortion services in Spain showed that while most immigrant women were using contraceptive methods just before they became pregnant, they were mainly using condoms and the pill, which are sensitive to their conditions of use (Serrano et al., 2012).
118The history of abortion in Bulgaria, as in many countries in Eastern Europe, has been turbulent. Abortion on request was legalized in 1956. In 1968, the government prohibited it for childless women out of fear of depopulation. Other restrictive conditions were introduced in 1973, which were relaxed in 1974. Beginning in 1990, after the end of the communist regime, abortion on request during the first trimester of pregnancy was allowed once again. Bulgaria is the country where abortion rates have fallen the most quickly over the last two decades, out of all those with complete abortion statistics (Sedgh et al., 2011).
119According to official statistics, abortion rates increased after 1956 (Marston and Cleland, 2003), reaching a very high level, around 70 abortions per 1,000 women, in 1973. At the time, aside from traditional contraceptive methods, abortion was the only available method for regulating births. The restrictions of 1973 and the introduction of the first modern contraceptives in 1975 stabilized the trends, but abortion rates decreased slowly in the 1980s. Beginning in the early 1990s, there was an acceleration. In 1996, the abortion rate was still 51 per 1,000, but it plunged over the following years, reaching 22 per 1,000 in 2003. This remarkable decline has been attributed to the establishment of effective sexual and reproductive health programmes in the early 1990s. In the 2000s, abortion continued to decline but at a more modest pace. In 2008, the country’s rate was close to the Western European average, at 16 per 1,000.
The United States
120The United States offers another illustration of the importance of national abortion statistics. The right to abortion on request was established at the federal level in 1973. Since then, the abortion rate has steadily declined, reaching 14.6 per 1,000 in 2014, as indicated by the latest census of abortion facilities (Jones and Jerman, 2017). Rates vary between states, being highest in urban states (33 per 1,000 in the District of Columbia, 30 per 1,000 in New York) and lowest in rural states (1.1 per 1,000 in Wyoming, 3.5 per 1,000 in South Dakota), whose residents travel to neighbouring states to access abortion services.
121Between 2011 and 2014, the abortion rate fell by 14%. It happens that, during the same period, legislators increased their efforts to limit the provision of services in a number of states, typically by implementing new rules on minimum equipment and personnel requirements for abortion clinics (where the great majority of abortions take place) or on the nature of their affiliation with hospitals. These rules were struck down by the Supreme Court, and the clinics reopened. They also do not seem to have been a barrier to access, as statistics do not show greater decreases in the frequency of abortion in the states where these measures were implemented. Other sources indicate an increase in the use of long-lasting contraceptive methods (IUD, implants) in the second half of the 2000s. The recent decrease in abortion in the United States thus seems to reflect better contraceptive coverage. However, it cannot be ruled out that it may also reflect more frequent self-administration of abortion medication. A study in Texas found that this practice seems to be relatively widespread (Grossman et al., 2015).
122Decreases in abortion-related mortality in Uruguay did not follow legalization, unlike the case in Nepal (discussed below), but are attributable instead to a harm reduction strategy. In 2001, after one too many maternal deaths, a small group of physicians opened a centre within their hospital to inform women facing unplanned pregnancies about the home use of abortion medication (Labandera et al., 2016). This initiative immediately decreased maternal mortality and severe abortion-related complications in the hospital. This experiment was progressively recognized by the country’s professional associations and then extended to the national level. This recognition of women’s right to sexual and reproductive health by a part of the medical profession also reshaped public discourse on the issue, leading to the legalization of abortion on request in 2012.
123The abortion rate in 2013 was 12 per 1,000 (Antón et al., 2016), a low level that is apparently explained by high contraceptive prevalence. Considering estimated levels before liberalization (nearly 30 per 1,000), it is possible, however, that many women continue to resort to informal medical abortion. In any case, the maternal mortality rate decreased from 26.6 deaths per 100,000 births in 2001–2005 (of which 37% were due to abortion) to 15.4 in 2011–2015 (8%, or three deaths due to abortion) (Briozzo et al., 2016).
124One of the particularities of the 2012 law is that it makes the use of medical abortion mandatory, apart from exceptional cases. As in many other countries in the region, the law also allows for conscientious objection, whose frequency varies widely between regions. 
125China legalized abortion in 1953. From the beginning, a family-planning programme was launched, and abortion and contraception were promoted jointly. The authorities decided to tackle the population problem head-on in 1973, extending contraception and abortion services to rural areas. Facing high birth rates, they instituted the one-child policy in 1979. It was strictly applied until the early 1990s (Wang, 2014). The policy includes mandatory use of an IUD for women with a child, of sterilization for couples with two or more children, and of abortion in case of unauthorized pregnancy. Between 1980 and 1983, campaigns of forced IUD insertion, sterilization, and abortion were carried out at the end of each year. The annual abortion rate reached its peak in the 1980s (56 per 1,000). The prevalence of sterilization also peaked during this period (46% of women aged 15–49 years). Following the Cairo Conference in 1994, China moderated its policy, promoting the free choice of contraceptive methods, first in pilot projects and then in the entire country from 2000 onward. In 2002, couples were also given the option of paying a hefty fine rather than ending an unauthorized pregnancy. Abortion rates declined quickly in the second half of the 1990s before stabilizing at around 19.5 per 1,000 between 2001 and 2010. While the massive prevalence of forced abortions at the height of the one-child policy is incontestable, researchers differ on the effects of Chinese population policy on the high incidence of sex-selective abortion (Goodkind, 2015).
126Today, attention is focused on the increasing abortion rates among young people who have sex before marriage without access to subsidized contraceptives, and on the many abortions of rural-to-urban migrant women (Zeng et al., 2015). Repeated abortions raise the question of postabortion contraceptive services, which are virtually non-existent today (Tang et al., 2017).
127The pattern in Nepal is quite distinct, which demonstrates the utility of abortion statistics even if they are incomplete. Abortion was liberalized in Nepal in 2002, which has contributed to the rapid decline of maternal mortality, decreasing from 580 deaths per 100,000 live births in 1995 to 190 in 2013 (WHO, 2014). Ever since abortion services were launched, the authorities have taken a particular interest in monitoring and evaluating them (Samandari et al., 2012). Public healthcare facilities produce monthly abortion statistics, but private facilities are not subject to this obligation. The Ministry of Health and Population pays close attention to the questions of postabortion contraception and complications. When it observes too high a number of complications or too few postabortion contraceptive consultations, targeted interventions are initiated.
128In 2009, as abortion statistics ceased to benefit from specific support, their quality deteriorated. The figures that have been produced nonetheless indicate that the number of women benefiting from legal abortion services continued to increase, from around 84,000 women in 2009 to 95,000 in 2011, and that the proportion of postabortion care remained stable at around 10% of all abortion services between 2009 and 2011. However, a large proportion of the population continues to use informal services, which are not captured in statistics. According to a 2014 AICM survey aimed at measuring all abortions (Puri et al., 2016), there were 323,000 abortions in Nepal in that year, which represents a high rate of 42 per 1,000 women aged 15 to 49 years; only an estimated 42% of these abortions were legal.
129In 1965, Tunisia became the first Arab country to liberalize abortion but only for women with more than five children. It legalized abortion on request in 1973. It was also the first African country to legalize medical abortion, in 2000. It was already used for 70% of abortions in the early 2000s (Hajri et al., 2004). The abortion rate in the country is comparatively low (9 per 1,000 women aged 15 to 44 years in 2009), and has been since the mid-1990s (Sedgh et al., 2011). It should be noted that the statistics in the country are considered relatively complete (at least 80% complete) (Sedgh et al., 2007).
130Despite the successful spread of medical abortion, access remains often difficult in rural areas. According to some observers, there has recently been a deterioration in access to abortion services in the public sector. This situation apparently dates to the budget cuts of 2004 and seems to have worsened following the Arab Spring in 2010–2011. A qualitative study with 22 women who were denied an abortion in Tunis in 2013 (interviewed in two healthcare facilities, one public and the other private) reveals how complicated the path to obtaining an abortion can be in the country (Hajri et al., 2015). Some women waited entire days in healthcare facilities without obtaining a consultation or were sent from one facility to another without being seen. Those who obtained a consultation were denied the procedure either because of a health problem or because it was too early or too late in their pregnancy (the same women who had previously been sent away because it was too early). Most of these women had to resort to costly private-sector abortion providers.
VI – The process that leads to abortion
131The variations in the frequency of abortions between countries and between population subgroups (discussed in the previous section) raise the question of abortion’s underlying factors. Studies on the causes of abortion agree on the need to distinguish two key moments in the process that leads to the voluntary termination of a pregnancy: first, the occurrence of an unplanned pregnancy; and second, the decision to end it. 
1 – Unplanned pregnancy: contraception in question
132The literature on the first part of this process is vast. The analyses that dominate the field tend to point to “unmet contraceptive needs”  and the efficacy of the contraceptive methods used as the top two determinants of unintended pregnancies (Singh and Darroch, 2012; Singh et al., 2009). These two parameters, along with marital status, are used in modelling the incidence of abortion worldwide (Sedgh et al., 2016). Studies from countries where long data series on contraception and abortion are available do show a close link between the spread of contraception and the decline of abortion (Marston and Cleland, 2003). Examination of decreases in fertility in Western countries, Eastern Europe, and Japan has shown that abortion rates increase at the beginning of these transitions, when couples start wanting to regulate births but lack contraceptives. With the spread of natural contraception, and then of modern contraception beginning in the 1960s, abortion rates have most often decreased in these countries (Davis, 1963; Frejka, 1985; Tietze and Bongaarts, 1975). This theory has been confirmed for fertility transitions in Latin America and Asia (Ahmed et al., 1998; Frejka and Atkin, 1996; Hollerbach, 1980; Singh and Sedgh, 1997; Westoff et al., 1998). In sub-Saharan Africa, where the decrease in fertility is still in its initial phase, a recent meta-analysis confirms this pattern: abortion rates are higher in the population subgroups that have fewer children and show greater contraceptive use (Chae et al., 2017). It is important to remember that at the world level, rates are also lowest in countries with the most effective sexual education and family-planning programmes (Ganatra et al., 2017). In other words, there is a profound change in the reproductive model during historical fertility transitions and the contraceptive revolutions that accompany them. Couples first control births through abortion (combined with abstinence, breastfeeding, or relatively ineffective traditional methods) and then move on to using modern contraception.
133This focus on contraception should not lead us to neglect other factors that directly affect unplanned pregnancies, such as sexuality and fecundability, as well as ambivalent fertility intentions. The latter has not been well identified as a factor of abortion thus far, as it is difficult to quantify. A recent study, based on weekly journal data collected from adolescent women, highlighted a clear link between ambivalent fertility intentions (indecision as to whether or not to have a child) and the occurrence of unplanned pregnancies (Miller et al., 2013). Similarly, in Finland abortions are apparently more frequent at the time of divorces (Väisänen, 2017). Intimate partner violence, another factor that has not been widely studied, is nevertheless associated with unwanted pregnancies and abortions (Pallitto et al., 2013).
2 – To continue or end a pregnancy
134Contraceptive difficulties and unplanned pregnancies mark only the beginning of the process that leads to an abortion. The woman must then decide to end the pregnancy and succeed in doing so. This second part of the process is less often studied. One way to explore the decision to abort is to examine the reasons women give at the time of the event. A literature review has analysed the results of 19 quantitative studies on this topic from eight developed countries between 1996 and 2007 (Kirkman et al., 2009).  It classifies the reasons given by women into three groups: those concerning the woman herself (wrong time, health, not wanting a child or another child); those that concern others involved (no partner, partner absent, partner does not want a child at the time, well-being of the unborn child or of older children); and material reasons (lack of financial means, etc.). Women always cite multiple reasons at once, which are often linked; for example, a large family, difficult material conditions, and lack of support from her partner. The authors conclude that the reported reasons reflect, above all, the concern of women or couples with being “good parents”. However, qualitative studies show that the situation is more complex. While women and couples mention good reasons not to want a child at the time, they often evoke equally legitimate reasons to want a child or not use contraception, which brings us back to the topic of ambivalence. These contradictions reveal tensions between different social expectations regarding parenthood, sexuality, and gender roles, and/or disagreements within couples on these issues (Bajos et al., 2002). In another recent literature review on reasons for abortion, Chae et al. (2017) review the results of 14 quantitative studies carried out between 2002 and 2012 in different countries of the North and South.  The countries where socioeconomic reasons are dominant are also those where women who have abortions are most often young and unmarried. In the countries where limiting family size is the principal reason, women who have abortions are mainly married, already mothers, and older.
135Condemnation of abortion by people close to the woman or the couple, their own negative perception of abortion, and difficulties accessing abortion services are other factors in the decision to have an abortion. To explore the role of these factors, research must compare women who ended their pregnancy with those who did not. In one of few such studies, Adamczyk (2008) showed that in the United States, women are more likely to end an unplanned pregnancy if they have career ambitions, are not affiliated with a Protestant or conservative religious group, and live closer to an abortion clinic as well as in a county where public funding for abortion is greater. In countries where access to abortion is legally restricted and stigmatized, the decision process is even more complex, and the trajectories women take to find a method and a provider who can help them are long and varied (Adjamagbo et al., 2014; Coast and Murray, 2016; Gbagbo et al., 2015; Puri et al., 2007). While some women are easily able to access public or private healthcare services, others must resort to traditional methods or to medical abortion (whether by prescription or through self-medication), and sometimes make multiple attempts.
3 – Who are the women who have abortions?
136As mentioned above, abortion happens frequently in every country in the world, and women in all categories, throughout their life course, have abortions. Nonetheless, the categories of women most likely to have an abortion vary between countries. To understand the relationship between a particular characteristic (such as the woman’s age, marital status, or socioeconomic status) and the propensity to have an abortion, the two key moments in the abortion process must be distinguished, as a given factor may have a negative influence on unplanned pregnancies and a positive one on the decision to end them (Rossier et al., 2007).
137Chae et al. (2017) reviewed research on the characteristics (age, parity, level of education, place of residence) of women who have abortions in 28 low- and middle-income countries. They found that women aged 20 to 29 years have nearly half of abortions worldwide. At the national level, the proportion ranges from 32% in Turkey to 61% in Armenia in the 28 countries included in the study. Most women in their twenties are sexually active and in a union, and wish either to delay childbearing or limit the size of their family. However, variations are seen around this general tendency. In sub-Saharan Africa, the highest proportions of abortions are in younger women: those aged 15–19 years in Nigeria and 20–24 years in the DRC, Ethiopia, Ghana, and Gabon. The same is true in Latin America, at least in Haiti and Mexico City, the only territories included in the study. In the Asian countries with documentation (Armenia, Azerbaijan, Bangladesh, Georgia, Kyrgyzstan, Nepal, Pakistan, Philippines, Tajikistan, Turkey, Uzbekistan, Vietnam), the proportion of abortions under the age of 20 years is low, but the proportion between the ages of 30 and 39 years is high, from 32% (Armenia) to 53% (Turkey). In Europe and North America, most abortions occur between the ages of 20 and 24 years, although in English-speaking countries and Finland, the proportion between the ages of 15 and 19 years is also relatively high (Sedgh et al., 2013). Finally, in the eight least developed countries in Europe included in the study (Albania, Belarus, Bulgaria, Moldavia, Montenegro, Romania, Serbia, Ukraine), most abortions take place between the ages of 25 and 29 years, or between 30 and 34 years in Serbia and Albania.
138With regard to parity (according to the same 2017 study), in sub-Saharan Africa, women who do not have children or have only one child are more likely to have an abortion, as in Mexico City and Haiti; for those who do not have children, abortion is a way to delay parenthood (Guillaume, 2004; N’Bouke et al., 2012). In many countries in the region, premarital sex is relatively widespread, but it is still condemned, and young people seldom use contraception. In Asia, however, abortion is more common among women (in a union) with two or more children, except in Nepal and the Philippines, where nulliparous women are the most likely to have an abortion. Finally, according to a somewhat older review, childless women have 30–50% of abortions in rich countries (Bankole et al., 1999).
139Chae et al. (2017) only had data on marital status for ten countries, which nonetheless reflects a great variety of situations. In Ethiopia, women who are not in a union have 60% of abortions, compared with only 5% in Albania. According to another study (Sedgh et al., 2016), the latest estimates of worldwide abortion rates show that they are higher among women in a union (36 per 1,000) than among women who are not in a union (26 per 1,000). But this average conceals contrasting situations. In three world regions – Asia, Latin America, and Europe – women in a union have 70–80% of all abortions, compared with only 50% in the three other regions (Africa, North America, and Oceania).
140Several regional profiles of abortion practice emerge from these various studies. In sub-Saharan Africa and Latin America, abortions most often take place at the very beginning of reproductive life and before the beginning of union. In Asia, they most often occur at the end of reproductive life and are aimed at limiting family size. In the richest countries, they are most frequent among young adults who will soon start a family. These disparities result from different normative constraints on sexuality and parenthood in different regions as well as from differing contraceptive use. In Asia, premarital sex remains rare, and family formation begins early; over time, abortion can serve to limit the number of children if contraception fails, or to obtain children of the desired sex (see below). In Africa, in contrast, the desired family size remains large, which limits the number of abortions within marriage and at the end of reproductive life; and while women are more likely to engage in sexual activity outside of a union than in Asia, it is nonetheless subject to social disapproval and often poorly protected, which can lead to unwanted pregnancies and abortions. The profile of Latin America is intermediate, between Asia and Africa: closer to Africa for the beginning of reproductive life, but as in Asia, many women also seek to limit births at the end of reproductive life. In Europe and North America, because of women’s frequent use of contraception both at the beginning and the end of reproductive life, abortions are more linked to the phase of family formation, when fertility intentions are ambivalent and changing, and contraceptive failures more frequent.
141As for women’s socioeconomic characteristics, the data clearly show that in low- and middle-income countries around the world, the chances of having an abortion increase with level of education, degree of urbanization, and living standard (Chae et al., 2017). A combination of factors may explain this pattern; more affluent and educated women may want fewer children and to control births. They also certainly have better access to abortion services (whether legal or informal), greater decision-making autonomy, as well as the economic and cultural resources to avoid an unwanted birth. However, both an earlier literature review by Bankole et al. (1999), which included developed as well as developing countries (South Korea, Italy, Kazakhstan, Kyrgyzstan, Uzbekistan) and another by Rossier et al. (2007) (United States, France, Italy) found that the link between level of education and propensity to abort varies between countries: in certain countries, the most educated women have fewer abortions as a result of more effective contraceptive practices.
4 – Repeated abortions and contraceptive practices after abortion
142Some women have multiple abortions over the course of their lives. Academic interest in repeated abortions has intensified in recent years in both the North and the South, as the phenomenon has evolved. First of all, the proportion of repeat abortions among all abortions has increased in countries with liberal legislation, such as Sweden (from 19% in 1975 to 38% in 2008) and New Zealand (from 23% in 1991 to 38% in 2011) (Rowlands et al., 2014). In France, among women having an abortion, the proportion who had already had at least one abortion was 18% in 1990, 28% in 2002, and 41% in 2011 (Mazuy et al., 2014). Given that the frequency of abortions has decreased or remained stable in these countries, the increase in the proportion of repeated abortions paradoxically reflects improvements in the contraceptive practices of the population as a whole, and a concentration of abortions among women who repeatedly go through situations that generate contraceptive failure. Contraceptive practice is better overall because the proportion of women who have an abortion in their lifetime has greatly decreased; but those who do end a pregnancy do so more times on average. This repetition results from the extension of the period preceding family formation, more often involving occasional sexual activity and the use (by certain individuals and over time) of less effective contraceptive methods, such as condoms. It is also explained by the choice of a contraceptive method that is not well adapted to the person and her situation (for example, a person who tends to forget to take the pill but who does not change method). This type of mismatch, if it is not corrected after a first abortion, increases the risk of repeated abortions and raises the question of postabortion contraceptive counselling (Bajos et al., 2013).
143In countries with restrictive legislation, little data is available on the frequency of repeated abortions. Postabortion care programmes, used on a large scale in the 1990s, initially focused on combating maternal mortality and on interventions to save women’s lives (Curtis et al., 2010). In 2001, an evaluation of these programmes showed that the proposal of contraceptives was an often-neglected component, increasing the risk of repeated abortions (Cobb et al., 2001). In the last decade, the inclusion of contraception counselling services in postabortion care programmes has become a strategic focus for reproductive health interventions. A series of initiatives to reinforce the role of contraceptive services as an integral part of abortion services have been successfully tested over the last decades (Tripney et al., 2013).
144The dimensions explored in studies on repeated abortion are diverse. A 2014 review of 46 studies around the world found that the factors most often associated with repeated abortion are a history of domestic violence, the use of a barrier-type contraceptive (such as condoms) or oral contraceptive, and adverse life events (divorce, employment difficulties, etc.) (McCall et al., 2014). But some researchers consider that in reality the differences between groups of women (no abortion, one abortion, multiple abortions) identified in these quantitative studies are relatively minimal, as is the case, for example, in the United States (Jones et al., 2006). A recent study in England taking a combined quantitative and qualitative approach concluded that the diversity of women’s histories is so great that generalization is impossible, beyond the relatively unremarkable observation of the occurrence of contraceptive failures (Hoggart et al., 2017). Moreover, according to these authors, studies on repeated abortion, by seeking to characterize at-risk women rather than situations that create risk, add to the construction of stigma around women who have abortions.
5 – Sex-selective abortion
145Sex-selective abortion, or prenatal sex selection, is a practice that was first described in the literature in the 1980s.  It consists in the elimination of female fetuses and tends to replace other practices of discrimination towards girls (selective infanticide, neglect, abandonment, etc.). This type of abortion has been growing since the 1980s with the emergence and widespread diffusion of new reproductive technologies and imaging techniques, such as medical ultrasound, that make it possible to discern fetal sex before birth.
146As a result of this phenomenon, male births have become predominant, as measured by changes in the sex ratio at birth. The universal, biological norm for this ratio is 105–106 births of boys for every 100 girls. When this ratio is between 110 and 115, or even higher, it reflects real discrimination against girls. This phenomenon is very pronounced in some Asian countries. In China, since the 2000s, the ratio has been between 115 and 125 depending on the province and area of residence (Guilmoto, 2015b). In Vietnam, the imbalance continues to grow, with the ratio at 113 to 114 (Becquet, 2015), again varying by place of residence and region as well as by children’s birth order (den Boer and Hudson, 2017; Guilmoto, 2010). In Indonesia, this imbalance is found only in certain ethnic groups (Guilmoto, 2015a). In South Korea, however, the phenomenon of excess male births quickly disappeared after a ratio that reached 110 to 115 in the 1970s and 1980s, but the country is a regional exception (Chung and Gupta, 2007). In certain northeastern regions of India, the ratios are 110 or 120, whereas in other regions in the south and east, they are normal. While countries such as Sri Lanka and Bangladesh seem not to show this type of discrimination, the opposite is true in Nepal and Pakistan. The practice of prenatal selection has also been observed in the southern Caucasus, Armenia, Azerbaijan, Georgia, and in the Balkans, including in Albania, Kosovo, Montenegro, and northeast Macedonia (Guilmoto, 2015a; Guilmoto and Duthé, 2013). This type of discrimination is also seen in the diasporas from these countries living in North America and in various European countries (Almond and Edlund, 2007; Auger et al., 2009; Dubuc and Coleman, 2007; Singh et al., 2010; Verropoulou and Tsimbos, 2010).
147While these practices reflect a very strong preference for boys in these societies, why are they only found in certain countries? They seem to be closely associated with kinship systems. They are only found in societies with strong patrilineal norms, where having a son is indispensable to continue the lineage. They are thus based on the economic and social roles assigned to boys as supporters of family (specifically of elderly parents) and resource producers. When high-quality ultrasound screening is available, this strong preference for boys leads to sex-selective abortion, and often in a context of low fertility where the risk of having only girls is much higher (Guilmoto, 2009).
148Certain countries have taken legislative measures in order to regulate these practices (restrictions on access to prenatal diagnostic technologies, legal regulations, etc.), but they have not had the anticipated effects (Guo et al., 2016; Rahm, 2017). Rahm (2017, p. 26) points out that, in South Korea, “the country’s return to a normal sex ratio at birth in Korea is explained by profound societal transformations in the country and by various reasons that have nothing to do with policies aimed specifically at this objective: notably the elevation of women’s social status and level of education, change in family structures and the loss of influence of the extended family, socioeconomic development, and urbanization”. Such abortions have also been condemned by the international community at various conferences, where they have been described as a form of discrimination or violence against women (Rahm, 2017).
VII – The consequences of abortions for women’s lives
149Unsafe abortions have different consequences on women’s health and lives. Beyond the fact that they pose a major public health problem, they can affect the living conditions of women and their families, and strain family budgets (Langer, 2003; Leone et al., 2016; Singh et al., 2009).
1 – Unsafe abortions: a cause of maternal mortality
150The annual number of deaths in the world due to an abortion has decreased over the last decades, from 69,000 deaths in 1990 to 56,000 in 2003, and then 47,000 in 2008 (WHO, 2011), the most recent year for which a figure is available. The decline in mortality due to abortions follows a parallel trend to that of maternal mortality overall: the number of maternal deaths went from 523,000 in 1990 to 289,000 in 2013 (WHO, 2014). In 2011, the WHO estimated that the proportion of maternal deaths caused by an abortion remained relatively stable between 1990 and 2008, at around 13% (WHO, 2011).
151But this type of result is sensitive to the methods used. According to a more recent WHO estimate of the distribution of maternal deaths by cause between 2003 and 2009 in 115 countries (produced using a hierarchical Bayesian model), only 8% of maternal deaths are due to abortion (Say et al., 2014). The authors emphasized that this result cannot be compared to the previous figure of 13% because of the divergence in the methods used in the two cases. The same year, a team from the University of Washington published yet another different estimate, this one for the year 2013, using a distinct dataset and Bayesian model (Kassebaum et al., 2014). They concluded that 15% of maternal deaths are due to abortion.
152Only the WHO estimate for 2008 (WHO, 2011) provides the case fatality rate – the ratio of the number of deaths caused by an abortion to the number of abortions – which was 220 per 100,000 abortions worldwide (or one death for every 455 abortions). The case fatality rate is extremely high in sub-Saharan Africa (520 deaths per 100,000 abortions), relatively low in Latin America and Eastern Europe (30), and virtually nil in developed countries and in East Asia, with the rates in other regions of Asia at intermediate levels (between 70 and 200). The situation in the Polynesian islands is also concerning (400 deaths per 100,000 abortions). Variations in case fatality rate between world regions result from differences in the safety of abortions and the effectiveness of obstetric emergency services, which in turn highly depend on the legal conditions of abortion, level of socioeconomic development, women’s financial resources, available abortion techniques, and the stigma that surrounds the practice (Ganatra et al., 2017). This mortality is distributed unequally among women and particularly affects the most disadvantaged women in the poorest countries (Grimes et al., 2006).
153A number of factors have contributed to the decreasing abortion-related mortality observed in recent decades. First of all, the legalization of abortion, by improving access to safe services and postabortion contraceptive counselling, has had a substantial effect on maternal mortality. An emblematic example is South Africa, where abortion was liberalized soon after the fall of apartheid, in 1996. Once safe abortion services (aspiration and medical abortions) had been established in the public hospitals, abortion mortality seems to have immediately decreased, from 425 deaths per 1,000 abortions in 1994 to 32 deaths per 1,000 abortions in 1998 (Jewkes et al., 2005). Since then, the case fatality rate for abortions has remained low, although many abortions are performed in the informal sector because of the insufficient availability of legal services. The techniques introduced in healthcare facilities seem to have spread to all providers, including those working outside the law.
154Beyond change in the laws and the availability of services, two other factors contribute to reducing abortion-related mortality: the improvement of family-planning programmes that decrease the number of unplanned pregnancies that could lead to abortions, and the effectiveness of obstetric emergency services that are able to treat women presenting with life-threatening abortion complications. In this regard, the case of the population observatory in Matlab, Bangladesh (where menstrual regulation was legalized in 1979) is revealing (Benson et al., 2011). This observatory is separated into two areas, each of which benefited in the 1980s and 1990s from the same state programmes for access to menstrual regulation, modern contraception, and obstetric emergency care. One of the two areas also took part in a community-based contraceptive and midwife-assisted childbirth programme.  The results showed that mortality due to abortion was high in both areas in 1976–1985 (107 and 99 abortion-related deaths per 100,000 pregnancies respectively) and that it decreased substantially afterward. The decreases were greater in the area with additional programmes, where there were 12 deaths due to abortion for every 100,000 pregnancies in 1996–2005, compared with 24 in the other area (Chowdhury et al., 2007).
155Finally, part of the decrease in mortality due to abortion worldwide is no doubt attributable to the spread of medical abortion, including in restrictive contexts. In addition to relatively well developed healthcare infrastructures, the predominance of this technique in many Latin American countries seems to explain the low lethality in the region, which is close to that of developed countries (WHO, 2011) (Figure 4). Modelling the impact of misoprostol on abortion mortality in countries of the South, Harper et al. (2007) estimated that two-thirds of deaths could be avoided if 80% of women used this medication to abort.
Case fatality rate by world subregion in 2008
Case fatality rate by world subregion in 2008* Except Australia, Japan, New Zealand.
Note: Subregions with a negligible proportion of unsafe abortions are not represented.
2 – Medical complications from unsafe abortions
156While the number of women dying from unsafe abortions is shrinking, complications remain frequent. The most common problems are incomplete abortions, haemorrhage, infections, and, in rarer and more serious cases, septic shock, uterine and intestinal perforation, and peritonitis. These problems are treatable, but the later the treatment, the more serious the consequences. Abortion can also have long-term effects on women’s health, such as sterility, anaemia, persistent weakness or pain, and inflammations of the pelvis or the reproductive tract. For example, it is estimated that each year 17 million women around the world suffer from secondary infertility following an abortion, while 3 million suffer chronic reproductive tract infections (WHO, 2007).
157According to Singh and Maddow-Zimet (2016), in 2012, 6.9 million women worldwide were hospitalized following complications from induced abortions, which does not reflect all complications, as many women do not receive treatment. The same authors estimate that around 60% of women who suffer complications from an abortion go to a healthcare facility for treatment, and that the total number of women suffering from a health problem immediately after an abortion was around 10 million in 2012. In other words, they estimate that 40% of the 25 million women who have an unsafe abortion each year experience a complication, 3 million of whom do not receive treatment.
158While the frequency of deaths due to an abortion has decreased worldwide, the overall number of hospitalizations for complications seems to be increasing, from 5 million in 2005 (Singh, 2006) to 6.9 million in 2012. This apparently reflects improvements in access to treatment rather than a greater frequency of abortions or of problems linked to unsafe abortions. In Latin America, a region where access to treatment is not increasing much over time because healthcare coverage is already high, the rate of hospitalizations actually decreased between 2005 and 2012 (Singh and Maddow-Zimet, 2016), no doubt because of the more widespread use of misoprostol.
159The most severe complications are still present in many countries, particularly in sub-Saharan Africa. The paucity of data on the severity of admissions, as well as the current lack of standardized definitions and tools, limit the scope of comparisons (Adler et al., 2012). A few case studies offer information on the frequency of severe complications in certain countries. According to a national survey performed in Ethiopia in 2007–2008 (just after the easing of legal restrictions on abortion), 41% of admissions for incomplete abortion involved moderate or severe morbidity (Gebreselassie et al., 2010). In Kenya, where the law is still restrictive, according to the 2012 national survey, 40% of complications were moderate and 37% severe (Ziraba et al., 2015). Complications as well as deaths from abortion are distributed in a highly unequal fashion across social groups.
160Finally, in addition to the costs paid by individuals, particularly the poorest, postabortion complications also carry high costs for healthcare systems, notably in the least advanced countries. These costs have been estimated at USD 232 million each year in developing countries (Singh and Darroch, 2012). Singh and Darroch estimate the funding needed to provide all affected women with access to quality care at USD 560 million.
3 – Psychological consequences of abortion
161Abortion’s consequences on women’s health are a major topic of debate among both opponents and defenders of the decriminalization of abortion. Ending a pregnancy is always a difficult decision for women to make, even more so when it is prohibited and/or socially stigmatized. It is a source of tension for women and for couples, who are torn between the social and sometimes religious norms that reject abortion and a personal (material, emotional, work) situation that pushes them towards it (Arslan Özkan and Mete, 2010; Palomino et al., 2011). In countries with easy access to modern contraception, women often feel responsible for an unplanned pregnancy, which they consider avoidable (Bajos and Ferrand, 2011). The decision to abort is still more difficult and stressful in contexts where the act is illegal because in addition to the health risks, women may be punished by law or stigmatized by health professionals and the community.
162While possible abortion-related psychological disturbances are often discussed, various scientific studies cast doubt on this relationship. Data suggest that the observed mental disturbances are linked to a state of health existing before the pregnancy (APA Task Force on Mental Health and Abortion, 2008; Bajos and Ferrand, 2011; Munk-Olsen et al., 2011). Charles et al. (2010) and Robinson et al. (2009) performed in-depth analyses of the studies cited by abortion opponents as establishing a link between abortion and psychological problems. They concluded that these studies suffer from numerous methodological biases: they are based on non-representative populations and do not take into account the women’s living conditions at the time of the pregnancy, in particular their psychological history (Lerner et al., 2016). In fact, when women decide on their own to have an abortion, it provides relief and allows women to regain control over the direction of their lives (Faúndes and Barzelatto, 2011; Flores Celis, 2016; Ortiz, 2008).
163The consequences of the denial of abortion for women and for their children have not been widely investigated. Some studies have emphasized the negative consequences of this denial on children’s health, well-being, and development, sometimes continuing into adulthood (David, 2006; Upadhyay et al., 2013). Similarly, the denial of the right to an abortion to women who want one has negative effects on their lives (Upadhyay et al., 2013), particularly when they have been victims of rape or incest. This raises the problem, among others, of the acceptance of the child, as different Latin American cases have shown (Taracena, 2002; Farmer, 2000; Lamas et al., 2000).
4 – The criminalization of abortion: punishing women
164Laws criminalizing abortion violate not only women’s reproductive rights but also their rights to health, liberty, safety, and potentially their right to life (Guillaume and Lerner, 2007). They accentuate inequality in gender relations, punishing only women – even in case of rape – and not the men responsible for the unwanted pregnancy.
165Abortion is often punishable for reasons prohibited by law; without the woman’s consent; and if performed outside legally sanctioned contexts or beyond the gestational limit. These punishments mainly affect women, but also the health professionals or other people who help them by providing information or abortifacient products.
166In countries with restrictive laws, where abortion is classified as a misdemeanour and in some cases even as homicide, it is punishable by prison sentences that can range from a few days to several months or years, depending on the country.  Few countries provide information on the number of people who are incarcerated under abortion laws, but activist groups in El Salvador and Mexico have compiled data on these cases (CRLP and La Agrupación Ciudadana, 2014; GIRE, 2013). Other punishments include fines, community service, and mandatory psychological treatment aimed at reaffirming the value of motherhood and family. The latter is used, for example, in some Mexican states (Puebla, Veracruz, and Yucatán, in 2016). In Latin America, a number of NGOs are taking legal action to free women imprisoned under these laws, or to ensure that their rights are recognized when they are denied legal abortions, as in cases of rape in Mexico and Brazil. Health professionals can also be suspended from practice. 
167The criminalization of abortion differs widely between countries and world regions, particularly for women. Punishments stipulated by the law are not always applied, as in France in the 1970s before the Veil law liberalized abortion (Bajos and Ferrand, 2011). When these laws fall into disuse, the practice continues to be considered a crime, but it is increasingly tolerated socially (Sanseviero, 2003).
IX – Final considerations
168This overview of abortion highlights the great diversity of situations at the level of laws and practices, its frequency, as well as its consequences for women’s lives and health. Beyond the various differences that we have highlighted, two general observations clearly emerge: whether legal or illegal, there are many abortions around the world each day, reflecting the difficulties in preventing unwanted pregnancies and contraceptive failures, difficulties that are unequally distributed between countries and socioeconomic strata. Despite this, the right to abortion is not universally recognized, and it continues to be punished, legally and socially, in many countries.
169As we close this review, we emphasize two points that remain at the centre of debates around abortion: the right to abortion and women’s access to safe abortion services. We end with a look at avenues for future research.
1 – Progress and challenges in improving access
170The struggle for the right to abortion has been ongoing for half a century (albeit at different levels of intensity in different countries and regions), but progress remains slow. NGOs and civil society actively promote women’s rights, while other movements are active in protecting the rights of the embryo. As a result, even when the right to abortion has been established, it is frequently challenged. In Nicaragua, for example, therapeutic abortion has been banned, despite all of the resulting risks for women’s health and lives. There is active, widespread support to reverse such steps at both the national and international levels. In late 2016, the Polish government abandoned its plan to limit access to abortion, but a new law has been in preparation since January 2018, aimed at restricting the right to abortion to two situations – in case of risk to the mother’s life or health, or if the pregnancy is a result of rape or incest – eliminating the possibility of abortion in case of fetal impairment.
171The risks associated with illegal abortion reflect social inequalities because they affect women differently depending on their social situation. As Langer (2002, p. 197) points out, “A society that allows some of its women to access an induced abortion without risk or difficulty, while others run great risks, is not a democratic society”. The negation of the right to abortion contradicts rights recognized at international conferences, in commitments adopted by countries around the world, such as the rights to equality, health, and free and responsible decision on the number and spacing of children, as well as sexual and reproductive rights as stipulated at the Cairo Conference and in the CEDAW recommendations.
172Even in some countries where the right to abortion seems to be firmly established, abortion is still subject to widespread stigma. Conscientious objection, denial of treatment, and fear of others’ reactions all continue. The growing literature on stigmatization reflects a new determination to combat both the concrete and symbolic dimensions of this gendered form of discrimination, which is not only directed exclusively at women – despite the fact that men are equally responsible for unwanted pregnancies – but would have no reason to exist if women’s freedom to seek fulfilment in non-family roles were recognized. This refusal to afford women reproductive autonomy is tied more broadly to the defence of a traditional vision of family, which is also reflected in stances on other societal issues, such as opposition to same-sex marriage, medically assisted procreation, and gestational surrogacy.
173The introduction of medical abortion and postabortion care are two major advances in healthcare systems’ handling of abortion. The growth of medical abortion, both legal and illegal, has transformed the conditions of abortion. It has improved the safety of the act; even when it is performed informally, it presents fewer risks than other methods, and the complications that might occur are less severe, as has been clear from the Latin American case. However, the use of these products continues to depend on governments’ health policies as well as on their availability, including on informal markets. Many women, particularly in Africa, still do not have access to them.
174The development of postabortion care has reinforced the role of health professionals and conferred a greater legitimacy on their treatment of women in this context. Nonetheless, there are limitations on introducing postabortion care: in countries with restrictive legislation, it confines the question of abortion to the domain of health. As Ouattara and Storeng (2014, p. 120) point out with regard to Burkina Faso, this can prevent a debate on the legalization of abortion, which some actors see as an unattainable societal change. And yet the decriminalization of abortion is a key factor in guaranteeing women’s rights and protecting their health.
2 – Research prospects
175This article reviews the scientific literature on the most commonly discussed issues surrounding abortion, i.e. legalization, measurement, and safety. In contexts where few advances have been made, there is clearly a great need for information on these issues, which is immediately useful when it becomes available. While there is an abundant literature on these issues (with highly variable coverage in different countries because of the difficulties that researchers face in the most restrictive contexts), research rarely strays from these well-worn paths.
176This thematic focus is understandable given that the majority of scientific research on abortion is carried out by non-governmental organizations, often in the context of operational research. These actors choose their subjects and target populations in accordance with their domain of action. They thus conduct studies in countries where the health burden of abortion is greatest and on themes such as knowledge and opinions concerning abortion legislation, the role of health professionals, and the consequences of abortion. There is still too little fundamental research on the subject. At the 28th International Population Conference in South Africa in 2017, out of 31 conference papers or posters on abortion, a little over a quarter were produced by research institutes or universities; the rest were from NGOs. There are several reasons for this relative lack of interest from the academic world. We should underline, first, that abortion issues are not widely studied in countries where contraceptive practice is widespread and access is legal (as in Europe), since in these countries abortion no longer constitutes a public health problem, a right that needs to be fought for, or a methodological challenge; yet major issues around stigmatization remain. In countries with restrictive laws – even when abortion is a hotly debated issue – it remains difficult to collect data because of its illegality and/or the associated stigma. Major ethical dilemmas arise when researchers seek to perform field studies on illegal phenomena. Financial support for such studies can be problematic to obtain from funding agencies. These various difficulties thus limit research in the area.
177This withdrawal of the academic world is, of course, compensated by research performed by activist organizations. And yet greater involvement on the part of researchers from various disciplines (history, gender studies, sociology, anthropology, economics, demography, public health), as well as greater interdisciplinarity, could help sharpen and expand analytic approaches, enable the development of more fundamental reflection, and diversify the themes of research. It would be interesting and helpful, for example, to use longitudinal data to demonstrate the impact of abortions on women’s trajectories; to study the question of abortion as a step on the path towards women’s control over their lives; to analyse the normative tensions around the decision to have an abortion when contraception fails; to investigate all the nuances of representations around abortion; and to unpack the factors involved in the stigmatization of abortion and their links to other dimensions of the status of women in different contexts. The fight for the right to an abortion within the feminist struggle deserves to be situated and documented. Another subject that merits further investigation is why medical abortion has spread in some contexts and not others. All of these avenues of research could widen the range of topics in the literature, offering a broader vision of abortion, which is often treated simply as a legal, statistical, or health issue. In the meantime, here we offer a few avenues for research.
Insufficient data collection
178In countries where abortion is legal, collection of abortion statistics is often flawed and incomplete (53 of 77 countries). Routine data collection could thus be greatly improved in many countries. The data collected provide only minimal information on the characteristics of the woman, the pregnancy, and the method used. Other variables would be useful, such as the number of abortions the woman has had, socioeconomic characteristics (level of education, occupation, place of birth, etc.), place of residence, postabortion contraception, and complications. These could help identify particular problems with access to contraception and inadequacies in abortion services. If the requirements of maintaining exhaustive and continuous statistics are too heavy, then periodic surveys (potentially on a sample) of healthcare providers who perform abortions could be undertaken.
179It is worth noting that the complexity of statistical data collection on abortion is likely to increase with the spread of medical abortion. Private-sector prescriptions for home-use abortion medication and any cases of procurement through informal networks (notably the Internet), even in countries with legal access, could complicate the recording of these acts in the future.
180To make up for the lack of national data, estimates of the number of abortions, their degree of safety, and certain characteristics of the women who have them are calculated at the regional and world levels. The recent shift towards Bayesian estimation methods makes the procedures used to produce global figures more transparent. But they are only estimates and not exact figures, and thus must be used with caution. Only ample, high-quality local data will make it possible to improve the precision and accuracy of higher-level estimates.
181Beyond statistics collected through healthcare systems, we believe that data sources must be diversified, specifically through surveys. One point merits greater attention in the interpretation of data collected in this way: the issue of differential under-reporting. Who are the women who under-report? It would also be helpful to better understand why the answers in surveys in some countries are nearly complete, while in others they very often fall short. Understanding these differences would allow us to identify the types of countries where it is possible to introduce direct questions on abortion into general population surveys.
182Furthermore, methodologies recently used to work on sensitive subjects, such as drug use, must continue to be tested in quantitative studies on abortion to improve their completeness, particularly in contexts where women are reluctant to talk about their abortions.
183Finally, qualitative approaches make essential contributions to the production of in-depth and local knowledge about abortion, and merit further development. Knowledge about the process and representations of abortion are insufficient in most countries, despite the great diversity of situations. What is known about one country often cannot be generalized to another. Some of the important themes discussed below can be explored through qualitative studies.
Key thematic directions for future research
184Several research topics merit special attention for future study. First of all, priority must be given to documenting the consequences of the negation of the right to abortion and the inability to end a pregnancy. What are the effects on women’s lives and their physical or mental health, as well as those of their children? This subject continues to be largely neglected in the social science literature (David, 2006; Upadhyay et al., 2013).
185Research must be developed according to changes in legislation, technology, and healthcare contexts. For example, as the WHO (2016b) recommends, special attention should be paid to the role of abortion in the Zika epidemic that affected several Latin American countries (notably Brazil), causing fetal impairment. What are the specific health and social consequences of the illness?
186The barriers women face in accessing abortion, both in legal and illegal contexts, are poorly understood in many countries and remain insufficiently studied. Research on the circumvention of the law also needs to be further developed, including the examination of information and distribution channels organized by NGOs, the use of ICTs (information and communication technologies, e.g. Internet, hotlines), or movements between cities, regions, or countries, etc. Informal networks supplying medications for use in medical abortion must be investigated through studies on these markets, surveys of pharmacists, etc. in order to better understand the conditions in which abortions occur, the advice that women receive, and treatment in case of complications.
187Generally speaking, the role of men in the abortion process has been little studied. Research is needed to characterize men’s place in these decisions, particularly where demands for paternal rights are emerging. Investigating the role of men entails examining the role of sex (whether planned or unplanned, desired or forced) in the process that leads to abortions. The question of ambivalence in desires for children, which appears to underlie a large proportion of abortions, has also been little studied.
188At the intersection of gender studies and public health, research on the stigmatization of abortion is now growing rapidly. This question is not posed with the same force in countries where abortion is legal compared to those where it is illegal and where research around abortion as a whole is insufficiently developed. It aims to measure the negative representations of abortion, their impact on trajectories of access to services, as well as on the quality of the services offered and the related health and social consequences for women. Such studies should be encouraged because they shed light on the heavy ideological burden that still weighs on this method of controlling births in most societies, the influence of healthcare actors in the field, and the embeddedness of abortion in the question of gender inequalities.
189Finally, a dimension that has not been widely studied is the involvement of actors – whether within the State, private sector, or NGOs – who play a major role in making abortion services available, defining postabortion care programmes, making medication available on the market, etc. In particular, research is needed in restrictive contexts with actors involved in the abortion debate, in order to better identify the levers for and brakes on the needed changes in legislation, in a field often riven with powerful ideological conflicts.
Document A.1. Definitions and types of abortions
191Medically, an abortion is defined as the expulsion or extraction from the uterus of a product of conception that is presumed to be non-viable, i.e that has not reached a certain period of gestation (less than 22 weeks, according to the WHO, 1977) or, in some cases, a certain weight (500 grams, according to the WHO) or height. These norms vary depending on the definition used in each country, and the criteria defining viability can also vary between countries depending on progress in medical technology (Pignotti, 2009).
Spontaneous and induced abortions
- Spontaneous abortions (miscarriages) are cases where the fetus is expelled without any deliberate action on the part of the woman or another person. This definition generally applies up to seven months of pregnancy (beyond which, it is considered a “stillbirth”).
- Induced or voluntary abortions follow a deliberate action performed by the woman or another person in order to end the pregnancy. These include therapeutic abortions performed for medical reasons, often due to an anomaly, an illness that threatens the life of the fetus, a risk of serious sequelae after birth, or danger to the mother’s life or health.
Abortions: legal and illegal, safe and unsafe
193Abortion is also classified according to the legislation and conditions of healthcare in each country.
194When it is performed in a context where it is allowed by law, it is referred to as legal abortion, and otherwise it is known as illegal, illicit, or clandestine abortion.
195The WHO distinguishes two types of abortions: safe and unsafe.
196Abortions are considered unsafe if they are performed by people who lack the necessary training, in an environment that does not comply with the minimum medical standards, or both (WHO, 1995), creating risks to the woman’s health or life.
197This definition was in use until recently and classified all illegal abortions as unsafe. The spread of new abortion techniques, first aspiration and then medical abortion (see section I.2), has allowed growing numbers of women to access (relatively) safe abortions, even in countries where abortion is not legal. In some countries with legal abortion, though, providers continue to use methods that are no longer recommended, such as dilatation and curettage.  These trends have created an increasing dissociation between legality and safety, to the point that the legal framework alone is not enough to characterize the level of safety of an abortion.
198In 2014, the WHO set out a new conceptual framework aimed at more precisely measuring the degree of safety of abortions in different countries (Ganatra et al., 2014). It proposes a new classification of abortions into three risk groups (Ganatra et al., 2017):
- Safe abortions, which are performed by a qualified person with a recommended technique;
- Less safe abortions, which fulfil only one of these two conditions;
- Least safe abortions, where neither of these safety criteria are met.
199The last two groups combined form the category of unsafe abortions. WHO recommendations for techniques and training continue to evolve, and this classification may change as well.
Incomplete abortions and postabortion care
200The WHO classifies an abortion as incomplete in case of “failure to remove or expel all of the pregnancy tissue from the uterus”.
201Postabortion care (PAC) aims to reduce the morbidity and mortality associated with complications from induced and spontaneous abortions or in case of incomplete abortion. It also includes providing contraceptives to prevent future unwanted pregnancies (WHO, 2013).
202Today, the indicator most widely used to measure the frequency of abortions is the abortion rate. It is calculated as the number of abortions in a year (or other period) per 1,000 women of reproductive age (15–44 years or 15–49 years). It is a measure of incidence. Over time, it has come to be preferred over two other indicators: prevalence (the proportion of women aged 15 to 49 years who have already had an abortion) and ratio (the ratio of the number of abortions to the number of births in a year). Prevalence counts abortions that may have happened more than 20 years earlier, while the ratio is not suited for use in comparing populations with different fertility levels.
203The total abortion rate is similar to the total fertility rate. It is the sum of age-specific abortion rates in a given year. It is equal to what the number of abortions per woman would be in a population where abortion rates across all ages are the rates for that year.
204The intensity of mortality due to an abortion is often measured simply by counting the number of such deaths in a given country or world region. But this figure is sensitive to the number of women of reproductive age in each country.
205Using the number of deaths due to an abortion per 100,000 births instead enables comparison of the intensity of mortality due to abortions in different world regions. Births are used rather than the number of women of reproductive age (which would be preferable, as fertility varies) because hospital data, where these deaths are often counted, only record births.
206The case fatality rate is the ratio of the number of deaths caused by an abortion to the number of abortions. This is a better indicator of the intensity of mortality by or due to abortion, as it takes into account the number of abortions, which varies widely between regions.
Document A.2. Strategies for estimating the number and safety of abortions around the world
207Since the early 1990s, the WHO, in collaboration with the Guttmacher Institute, has been compiling data on the number of induced abortions worldwide, using all of the quantitative sources cited above. Estimates have been produced for the years 1995, 2003, and 2008, and for the period 2010–2014 (Ganatra et al., 2017; Henshaw et al., 1999; Sedgh et al., 2007, 2012, 2016). The results are published not at the individual country level but at the level of world regions, on the premise that inevitable estimation errors at the country level should compensate one another at the regional level. In the first three estimates (1995, 2003, 2008), a binary classification of abortion safety was used. All abortions in countries with a liberal law were classified as “safe”, while all those in countries with restrictive access were classified as “unsafe”, as they were performed outside healthcare facilities and were not registered, except potentially where complications led to treatment in a hospital. The total number of safe abortions was thus obtained by summing the figures available for countries with liberal legislation (statistics, surveys of providers, general population surveys), possibly after corrections.  To estimate the number of unsafe abortions, a literature search was performed on studies concerning countries with restrictive laws. All available estimates (AICM, population surveys) were assessed and in some cases corrected.  These results were then extrapolated to countries without data one by one, using a qualitative approach, according to their similarities with well-documented countries, particularly with respect to contraceptive use and abortion complications as attested in local hospital studies. Finally, the incidence of abortion – the abortion rate for a given year or period – was calculated for each region by taking the ratio of the annual number of abortions to the population of women aged 15 to 44 years. 
208With the diversification of safety conditions in countries with both restrictive and liberal laws resulting from the spread of new techniques, a new logic was used to estimate abortion figures for the most recent period (2010–2014). For the first time, the calculation of the incidence of abortion was separated from calculations aimed at identifying the proportion of unsafe abortions. Both were estimated using Bayesian methods.  This innovation was made possible by the publication of (again, Bayesian) estimates of data such as the prevalence of contraception and unions for all countries. It follows the general trend in the estimation of public health statistics (such as the number of deaths by cause) at the world level. The use of systematic modelling based on factors associated with abortion so as to infer figures for countries for which no data are available (instead of a qualitative, case-by-case approach) is an improvement because it allows the estimates to be replicated and verified. In principle, these publications contain enough details to allow research teams to reproduce and improve the results.
209The model created to estimate the worldwide abortion rate provides estimates for the period 2010–2014, and retrospectively for each five-year period going back to 1990–1994.  The results obtained are somewhat higher than previous estimates. For example, the new estimate of the abortion rate for 2005–2009 is 35 abortions per 1,000 women aged 15 to 44 years, compared with 28 for the previous estimate. The difference is due to particularly large gaps in a few subregions, notably the Caribbean. The researchers emphasized their use of conservative hypotheses (the lower of two different reference levels is chosen) in extrapolating data for previous estimates, as well as their use of corrective factors that were doubtless also conservative.
210The new estimate for the proportion of unsafe abortions in each region covers only the period 2010–2014, as empirical data are lacking for the previous periods. The new non-binary categorization of safety (safe, less safe, least safe) (see Appendix Table A.1) identifies abortion techniques that do not meet recommended standards but that nonetheless decrease risk. The estimation process consists in extracting the distribution of abortions by technique from national statistics, and in performing a literature review of studies at the national and subnational level which offer indications of the distribution of abortions by provider, method, and/or location of the procedure. A Bayesian hierarchical model is then used to extend these figures to the countries in the same region for which data are not available, based on a series of variables measured at the national level, chosen according both to a theoretical model of the factors affecting the safety of abortions and to the availability of data. Finally, estimated distributions for each country are applied to the number of abortions derived from the estimation of incidence (number of abortions in a given period) described above, providing total numbers of abortions and their distribution by safety category for the different world regions and subregions.
211This exercise (Ganatra et al., 2017) produces results in agreement with previous estimates, which in 2008 showed that 49% of abortions in the world were unsafe (Sedgh et al., 2012). The new method estimates the figure at 45% for the period 2010–2014, and breaks it down into the two levels of inadequate safety. However, this procedure is subject to limitations. First, there are few studies describing the methods, practitioners, or locations of abortions in countries where a large proportion of abortions take place outside the healthcare system; and even in legal contexts, abortion statistics do not always contain the necessary information. The number of data points used to produce these estimates thus remains relatively low. Second, the abortion safety factors used to extrapolate these data points can only be captured in a highly approximate fashion,  as there are few databases covering all countries in the world. Today, it is essential to improve and expand the collection of data on the safety of abortions in order to improve the quality of these estimates.
Legal conditions of abortion in 193 countries
Legal conditions of abortion in 193 countriesFor countries marked *: Center for Reproductive Rights, http://worldabortionlaws.com/map/
(1) Considered illegal, but with exceptions:
- In Cameroon, it is permitted by the criminal code if it has been “proved necessary for the saving of the mother from grave danger to her health.” It is classified as “to save the woman’s life” only.
- In Comoros, abortions are permitted for very serious medical reasons, recognized in writing by at least two physicians.
- In Djibouti, the interruption of a pregnancy by a physician for a therapeutic reason in accordance with the public health law is not considered to constitute an abortion.
- In Democratic Republic of Congo, abortion is prohibited by the criminal code. Its legitimacy in exceptional cases – when the mother’s life is seriously threatened and therapeutic abortion seems to be the only way to save her – is a matter of ongoing debate.
- In Madagascar, it is up to the State to ensure the availability of specialized, high-quality healthcare services that are accessible to women presenting complications linked to pregnancy, postpartum, and the postnatal period, as well as abortion.
(3) In these countries, the Harmonised Codes of Ethics and Practice for Medical and Dental Practitioners of the ECOWAS (2013), which defines practitioners’ general obligations, may apply (but it is not considered an official government document). It specifies that “voluntary interruption of pregnancy cannot be practised unless otherwise provided for by law”. However, section V, concerning practitioners’ obligations towards patients, article 141 of the Code stipulates that “a therapeutic abortion may be performed if the intervention is the only means to save the life of the mother”.
(4) In Mexico, each state has its own law. The only case for which it is legal in all states is rape. It is permitted without restriction in Mexico City.
(5) Australia is a federal country where each state has its own legislation. In most cases, it is permitted on request.
(6) In Chile, a law permitting abortion for three reasons – life, rape, and fetal impairment – was passed in August 2017.
Note: Permitted in case of rape (R), incest (I), fetal impairment (FI), or for women with a mental disability (D).
Derivatives of these products are sold under different names and formulae (molecules on their own or combined with others).
In the early 1970s, the WHO chose four characteristics to define essential medicines: therapeutic effectiveness, safety, satisfaction of the health needs of the population, and affordability (Whyte et al., 2002).
This database (http://srhr.org/abortion-policies/), covering all countries and world regions, provides information on laws and their exceptions, legal sources (excerpts of criminal codes), required authorizations, gestational limit, sanctions, etc.
The Guttmacher Institute periodically publishes a report on abortion in the world. Its latest report (https://www.guttmacher.org/report/abortion-worldwide-2017) was published in early 2018. The present article has not been able to take into account this report’s data.
There has been discussion in Latin America about adopting a broader definition of health that also includes its social dimension, with reference to the WHO’s concept of health. It was pointed out that the consequences of abortions are measured not only in terms of morbidity and mortality but also in terms of access to quality healthcare services, without stigma or violence. This social dimension refers to the protection of women’s well-being and to the consequences of an unwanted pregnancy for their life plans (González Vélez, 2011).
Few countries define what they consider to be a fetal impairment. Sometimes the law specifies genetic diseases, serious congenital malformation, incurable diseases, mental deficiency (Indonesia, Qatar, South Korea, Honduras, Peru), or pathologies incompatible with life outside the womb, such as anencephaly (Uganda).
In a few countries, the law specifies that “account may be taken of the pregnant woman’s actual or reasonably foreseeable environment” (Barbados, Belize, United Kingdom, Zambia). Specific family and social situations may also be considered: being unmarried, death or divorce of the spouse, number and spacing of children, poverty (Germany, Guyana, Iceland, Kazakhstan, Macedonia).
For honoris causa (in cases of socially condemned premarital or adulterous conception) in Costa Rica, in case of forced artificial insemination in Colombia, Peru, and Costa Rica, or for other reasons: imprudence, mental disability, HIV infection, age 16 years or under or 40 years or older, contraceptive failure, etc.
Argentina, Angola, Armenia, Bahrain, Bangladesh, Barbados, Brazil, Cambodia, Cape Verde, Cuba, DRC, Georgia, India, Ivory Coast, Kazakhstan, Kyrgyzstan, Malaysia, Mauritania, Mongolia, Morocco, Mozambique, Nepal, Panama, Rwanda, São Tomé and Príncipe, Saudi Arabia, Syria, Tajikistan, Turkey, Uzbekistan, Yemen.
Bahrain, Indonesia, Japan, Kuwait, Morocco, Qatar, Republic of Korea, Saudi Arabia, Syria, East Timor, Turkey, United Arab Emirates, Yemen.
Court authorizations are required in certain countries: Bolivia, Eritrea, Georgia, Macedonia, Namibia, Panama, Rwanda, Seychelles, Zimbabwe, etc. and the filing of a complaint: Argentina, Bolivia, Colombia, Cyprus, Finland, Hong Kong, Latvia, Mauritius, Monaco, Saint Lucia, Uruguay, etc. (WHO, 2017). For more on the situation in Latin America, see Bergallo and González Vélez (2012). No authorization of this type is required in Ethiopia.
Three days in Albania, Germany, Spain, Hungary, and Portugal, five days in the Netherlands, six days in Belgium, and seven days in Italy and Luxembourg.
It is included in the majority of European countries: Denmark, France, Italy, Norway, etc.; in the United States; in Latin America: Bolivia, Colombia, Panama, Uruguay, Mexico City; in Asia: Nepal, Singapore; and in Africa: Ghana, Guinea, Mozambique, etc.; it also features in the Harmonised Codes of Ethics and Practice for Medical and Dental Practitioners of the ECOWAS (Economic Community of West African States), where its recognition is combined with an obligation to refer women to non-objecting professionals.
See the United Nations list of countries by level of development at http://www.un.org/en/development/desa/population/publications/pdf/policy/ WPP2013/wpp2013.pdf (p. 37). Three categories are distinguished: more developed countries, less developed countries, and least developed countries.
These exception clauses may be set out in the criminal code or in health policies or programmes to broaden the right to abortion. In Comoros, where abortion is totally prohibited, an article in the criminal code stipulates that “abortion can be performed for very serious medical reasons recognized in writing by at least two physicians” (WHO, 2017). Some analyses group countries where abortion is totally prohibited with countries where it is allowed only to save the woman’s life, whether or not this reason is explicitly mentioned.
Also known as the Protocol to the African Charter on Human and Peoples’ Rights on the Rights of Women in Africa; see http://www.achpr.org/instruments/women-protocol/
In Ethiopia, following a recent change in the law to improve access, identifying the perpetrator is no longer required.
Namely, to save the woman’s life in Somalia, to preserve the woman’s life and health in Kenya, with in some cases the addition of rape, incest, and fetal impairment (Mauritius, Lesotho, Niger, Rwanda, Swaziland, Togo). The grounds for legal abortion have been expanded in Chad, Mali, and Benin, where it was previously only permitted in order to save the woman’s life, and in Guinea and Ethiopia where it was only allowed if the woman’s life and health were in danger (Center for Reproductive Rights, 2014).
It is also allowed in French Guiana and the French Antilles, which are under French law.
The limit is up to 14 weeks in case of rape, and there is no limit if the woman’s health is threatened or in case of fetal impairment; in 2008, it was permitted only in case of rape, serious health problems, or fetal impairment (Wood et al., 2016).
PACE, Resolution 1607 (2008), “Access to safe and legal abortion in Europe”, retrived from http://assembly.coe.int/nw/xml/XRef/Xref-XML2HTML-EN.asp?fileid=17638&lang=en
Row v. Wade is a decision of the Supreme Court of the United States that recognizes abortion as a constitutional right, thereby invalidating all rules criminalizing, prohibiting, or restricting it.
Guttmacher Institute, Induced Abortion in the United States [Fact sheet], retrieved from https://www.guttmacher.org/fact-sheet/induced-abortion-united-states
This policy prohibits the distribution of any US funding to international organizations that provide legal abortion services or offer complete information on sexual and reproductive health.
A study based on a review of the literature published between 1980 and 2015, covering 3,126 articles and 16 reports. This analysis bears on 24 articles presenting the results of surveys of women in 13 countries.
Some authors refer to this type of travel as “abortion tourism”, an inappropriate term when considering the financial and psychological costs for these women.
Campaigns have been organized in the following countries: Guatemala, Ireland, Morocco, Mexico, Poland, Portugal, Spain.
The NGO Women on Waves has set up hotlines in Poland, Africa (Kenya, Malawi, and Morocco), Asia (Bangladesh, Indonesia, Malaysia, Pakistan, Thailand), Latin America (Argentina, Chile, Ecuador, Peru, Venezuela, Uruguay).
Although abortion sincrease immediately following legalization, as un reported clandestine abortions become legal and appear in statistics, the rates then decrease (Juárez Carcaño, 2008).
A Brazilian bishop recently excommunicated the mother of a nine-year-old girl following a rape, along with the doctor who performed the abortion, despite the fact that abortion for this reason is legally permitted. In his 2016 apostolic letter Misericordia et Misera (Mercy with Misery), Pope Francis writes that “abortion is a grave sin, since it puts an end to an innocent life”, and he “grant[s] to all priests, in virtue of their ministry, the faculty to absolve those who have committed the sin of procured abortion”. Retrieved from https://w2.vatican.va/content/francesco/en/apost_letters/documents/papa-francesco-lettera-ap_20161120_misericordia-et-misera.html
For more information on abortion and religion, see Faúndes and Barzelatto (2011), Maguire (2003), and Schiff (2002).
One example from France is the “manifesto of the 343 women declaring they had had an abortion, published on the initiative of activists from the women’s liberation movement in 1971” (Pavard, 2009, p. 79).
Allowed on request or for social and economic reasons, within certain gestational limits.
Bulgaria, Czech Republic, Hungary, Slovakia, and Slovenia in Eastern Europe; Denmark, Estonia, Finland, Iceland, Norway, and Sweden in Northern Europe; Belgium, England, France, Germany, the Netherlands, Scotland, and Switzerland in Western Europe; Italy, Spain, and Portugal in Southern Europe; as well as Israel and Singapore in Asia, and New Zealand in Oceania.
Albania, Belarus, Croatia, Latvia, Lithuania, Macedonia, Moldavia, Montenegro, Romania, the Russian Federation, and Ukraine in Europe; Kazakhstan, Kyrgyzstan, Uzbekistan, Tajikistan, Turkmenistan in central Asia; Armenia, Azerbaijan, and Georgia in West Asia.
In this survey technique, various interviewers pretend to be potential clients of illegal abortion providers. They record the information collected after the interview, without revealing the provider’s identity.
Near misses are situations where the healthcare system prevents a death. They are all thus observable by definition in healthcare facilities. When a type of death (such as maternal death) becomes rarer, the study of near misses helps to improve healthcare systems.
The number of women aged 15 to 44 years worldwide went from 1.26 billion in 1990–1994 to 1.64 billion in 2010–2014 (United Nations, 2017).
In the 1990s, to complete the statistics drawn from abortion notifications (bulletins d’interruptions volontaires de grossesse), which were established in 1970 and are poorly completed in private clinics, the Ministry of Health turned to other administrative sources, including the annual statistics of healthcare facilities (SAE, Statistique annuelle des établissements de santé) and the medical statistics database (PMSI, Programme de médicalisation du système d’information). As procedures performed outside of hospitals are not recorded in SAE and PMSI, the Ministry of Health estimates the number using figures for the reimbursement of the procedure by social security (Vilain, 2017).
According to the concepts usually used in this field of research, an unplanned or unintended pregnancy is one that happens when the woman or the couple wishes to avoid having a child; the unintended pregnancy can be unwanted or mistimed.
“Unmet need” for contraception is a measure of the propensity not to use contraception among women who are sexually active, not sterile, and not protected by postpartum amenorrhea, and who do not want a child in the near future.
The countries are Australia, Canada, Denmark, Greece, the Netherlands, Norway, Sweden, and the United States.
Armenia, Azerbaijan, Belgium, Republic of the Congo, Gabon, Georgia, Ghana, Jamaica, Kyrgyzstan, Nepal, Russia, Sweden, Turkey, United States.
This section is based on the review by Guilmoto (2015b).
Health support workers (auxiliaires de santé) are trained to work not in healthcare centres but directly with the population, in particular through periodic home visits.
Multi-year prison sentences can be as high as 20 years in Benin, 16 years in Colombia, 10 years in the Bahamas, etc. (WHO, 2017).
In French-speaking Africa (Burkina Faso, Chad, Guinea, Niger, Senegal, etc.), Monaco, Republic of Korea, Venezuela, suspensions can be a minimum duration of five years in certain cases (WHO, 2017).
The term provider designates people who perform abortions, who may have highly variable levels of skills and training. These people can operate both in the formal and informal sectors.
Legal abortion figures in different sources are corrected on the basis of the published literature and in consultation with local experts in different countries.
Population survey data were corrected on the basis of the mean response rate in the few studies comparing survey and statistical data.
These data are also used to calculate the number of unplanned pregnancies worldwide, by adding together the number of abortions and the estimated number of unplanned births.
Bayesian estimation is a method that allows the probability of an event to be inferred from the probabilities of other events that have already been evaluated, as a function of the characteristics of the statistical units involved. A hierarchical model is used to model probabilities at multiple levels (country, region).
The predictors of the number of abortions in countries without data – which were chosen based on a theoretical model of the determinants of abortion and on the availability of data – are the numbers of women aged 15 to 44 years divided into four groups: women not in a union, and, among women in a union, those who use contraception, those with an unmet contraceptive need, and those who do not require contraception. The model exchanges information between countries with and without data in a given subregion on the basis of these predictors in order to estimate the missing data. Note that the possibility of using several country-level variables (age distribution of women, level of education, and GDP) in order to calibrate these exchanges of information was tested. As none improved the model, they were not used.
For example, the availability of safe abortion services is estimated based on the inclusion of misoprostol (for any indication) and mifepristone on the list of medicines authorized in the country.