Although abortion occurs in every society, and a substantial proportion of pregnancies are resolved by abortion worldwide, there is little empirical research on why women obtain abortions. This lack of information is part of an overall scarcity of data on abortion. Legal, moral and ethical issues surrounding abortion make research on all aspects of abortion difficult to undertake, and also affect the quality of the information obtained. Collecting good information on reasons for abortion may be especially difficult, because it requires asking women to articulate the often complex and sensitive process that led to the decision.

Some might argue that we already know why a woman obtains an abortion—she does not want the pregnancy—and that we need look no further. However, while at one level almost all abortions result from unintended pregnancies, there can be many steps between acknowledging an unplanned pregnancy and having an abortion. Moreover, many women who have an unintended pregnancy either do not seriously consider having an abortion or do not consider it at all. Some will simply adjust to the pregnancy; for others, an initial desire to discontinue a pregnancy can change because they were either ambivalent themselves or because they acceded to the preferences of others. On the other hand, conditions that were either unknown or were less serious before conception may also change, so that a pregnancy wanted at the time of conception is no longer wanted later on.¹

In addition, not all women who decide to seek an abortion will succeed in obtaining one. They may face personal and social barriers such as their husband's objections or community values that oppose abortion. In countries where safe abortion services are scarce, only affluent women who can afford the fees of a private doctor will obtain an abortion, along with poorer women who are so determined they are willing to risk their health and life in seeking out unsafe clandestine services.

Even though the planning status of a pregnancy does not tell us the full reason why women choose abortion, understanding the prevalence of unplanned pregnancy and its proximate cause—nonuse of contraceptives or contraceptive failure—is essential for understanding the context within which women seek abortion.

Evidence abounds that a high proportion of women become pregnant unintentionally, in both developed and developing countries. In the United States and in some Eastern European countries for which data are available, about one-half to three-fifths of all pregnancies are unintended, and a large proportion of these are resolved through abortion.² And in many developing countries, the proportion of recent births that are unintended exceeds 40%; even in regions where most couples still want large families, 10-20% of births are unplanned.³

This level of unintended pregnancy for developing countries would be even higher if more accurate abortion information were available, since most abortions represent, by definition, unintended pregnancies. The limited available data show that high proportions of unintended pregnancies are resolved by abortion in Tanzania (61%) and in six Latin American countries (ranging from 43% in Mexico to 63% in Chile).⁴ In the former Soviet republics of Kazakstan and Uzbekistan, more than one-third of mistimed pregnancies and about four-fifths of pregnancies among women who have all the children they want resulted in abortions.⁵

While unintendedness is clearly a first level of explanation, for many women it covers a wide range of more specific underlying factors.⁶ Although the available body of research on these underlying explanations has many limitations, with careful interpretation, a review and synthesis of findings from the published literature can advance our knowledge in this area, while providing ideas for new research.

In this article, we first discuss women's intentions to postpone or prevent pregnancy, and whether they support these intentions by practicing contraception. This contextual information is based on data from nationally representative surveys for 52 countries.

We then explore the reasons women give for why they obtained an abortion, using the limited information available from a review of published findings of 32 studies conducted in 27 countries, as well as original analyses of survey data from three of these countries (the Czech Republic, Turkey and the United States). Within the limitations of the available data, we assess whether women's reasons for seeking abortion vary by region. Finally, we examine how these reasons are related to the characteristics of women who obtain abortions, since despite the intuitive plausibility of a strong association between women's characteristics and their reasons for abortion, few studies have investigated this relationship.

Data Sources

Data on Fertility Intentions

Forty-nine of 52 fertility surveys that provide background data are from the series of Demographic and Health Surveys (DHS) conducted by Macro International in collaboration with individual national governments. These surveys are nationally representative and include all women of reproductive age (ages 15-49 in most countries), except in Asia and in North Africa and the Middle East, where only ever-married women were interviewed. (The exceptions are Morocco and the Philippines, where all women, regardless of marital status, were interviewed.) In addition, we examine comparable fertility intentions data from three developed countries—the Czech Republic, Japan and the United States.⁷

Data on Reasons for Abortion

The second, and more important, source for our analysis is existing research on reasons why women obtain abortions. To identify relevant studies in both developing and developed countries, we undertook an extensive search using databases (i.e., Popline, Medline and Population Index) and available bibliographies. We also sent letters of request to organizations and individuals, asking them to send us related unpublished work or references.

From the small number of studies carried out between 1967 and 1997 that included reasons for abortion, we selected only those that solicited information directly from women who had had an abortion, that were based on open-ended questions or precoded responses with a wide range of response alternatives, and that presented quantitative information on reasons for abortion. Because of the relative paucity of this research, we included some studies with small sample sizes. The 32 studies selected for analysis are listed in the Appendix.

There are five main types of surveys, and each presents advantages and limitations:
•National surveys of abortion patients. These surveys, based on national samples of providers and conducted at the abortion facility, have the advantage of representing all women having abortions; as such, they avoid the biases inherent in retrospective studies.* However, such national studies of abortion patients are very rarely undertaken.

•Subnational hospital- or clinic-based surveys of abortion patients. These studies, which are limited to particular areas of a country or are based on a nonrepresentative group of hospitals or clinics, collect information on women hospitalized for abortion complications or for the abortion procedure itself. The data are collected directly from the women or are abstracted from medical records or both. In countries where abortion is highly restricted by law, most of these studies include only women treated for abortion complications; as such, the data probably suffer from selectivity bias. (For example, women who have complications but fail to obtain hospital treatment, and those who receive a safe abortion and do not develop complications, will not be included.) This bias is reduced in countries where abortion is legal, but other bias can stem from the sample of clinics or providers being too small to be nationally representative.

•Official government statistics on abortion patients. These studies are relatively rare, and are available only in countries where abortion is legal under broad conditions. The data quality is affected by the completeness of coverage and by the type of data collection approach used. For example, if women are required to report their reasons for having the procedure on an official form or if they must answer questions posed by medical personnel, they may be less forthcoming than if they are surveyed less formally.

•National fertility surveys. Some fertility surveys collect information about abortion, and also ask about women's reasons for having one. This source has the advantage of being nationally representative. However, by definition, the data are limited to women who acknowledge having had an abortion in the survey interview. Depending on the extent of underreporting† and on whether it occurs selectively according to a woman's stated motive for the abortion, the data might not represent the full range of reasons. An additional source of bias in fertility surveys is that data collected retrospectively tend to be less accurate than those gathered at the time of the event.

•Subnational surveys of women. Such community surveys often sample a cross-section of all women in a designated area, but occasionally can sample only a selected group. The possible limitations of this type of study include lack of national representation, the exclusion of some women of reproductive age, underreporting of abortions and, in some cases, small sample sizes.

Many other sources of bias in data on women's reasons for abortion are not specific to any one type of source. These include the marital status composition of the sample, the size of the sample and the questionnaire design. For example, most studies that cover the subject ask only a single question about women's most important reason for having an abortion, and respondents are not given the option of mentioning other contributing reasons, even though their decision may have been motivated by more than one.

This restriction on responses prevents a more nuanced understanding of the reasons why women have abortions, especially when women have more than one reason or find it difficult to rank reasons in order of importance. However, some research allows multiple answers to the question; for example, one U.S. study found a mean of 3.7 reasons, with 63% reporting 3-5 and 13% reporting 6-9. Only 7% of women in that study gave just one reason for obtaining an abortion.⁸

Data quality can also be affected by the format of the data collection process (i.e., gathered through a personal interview by a trained interviewer, collected by means of a self-administered questionnaire, or assembled by medical providers and entered into official records). Moreover, the timing of the interview (i.e., during a hospital stay, when the woman is being treated for a complication, when she is attending a clinic to obtain an abortion, or even months or years later) may also influence the accuracy and quality of women's responses.

Finally, the legal status of abortion may affect women's willingness to report fully on the reasons for their abortion. And, regardless of the legal climate, women may (consciously or not) give socially acceptable reasons rather than their actual reasons.⁹

All of these potential biases, which can differ from one study to another, might affect the quality of the information collected. In summary, some of the differences in the findings of the various studies might result from a lack of comparability of the data, stemming from one or more of the factors mentioned above. The many limitations of the studies available for presentation here must be borne in mind when the results are interpreted.

Categories of Reasons

The 32 published studies (including three for which we were able to retrieve actual data tapes and undertake original analyses) differed somewhat in how they grouped women's reasons for having an abortion, and exactly comparable categories were not always available. However, since some reasons were similarly worded across studies, we used our judgment to create nine broad, comparable categories of women's reasons for their abortion.

Specific reasons that were distributed over nonoverlapping categories must necessarily be interpreted in light of all of the categories used by a particular study.‡ For the few surveys for which we had the actual data tapes (the Czech Republic, Turkey and the United States), we created groupings of reasons that would be most comparable to those created from the other published studies.

Results

Pregnancy Intentions and Method Use

Unintended pregnancy, the fundamental and immediate cause of abortion, is a reality worldwide: Overall, the combined proportion of married women aged 15-49 who want to postpone their next birth or to stop childbearing at any given point in time§ ranges from 39% in the Central African Republic to 89% in Japan (Table 1). This proportion is greater than 50% in 48 of the 51 countries for which data are available.

Despite these women's desire to postpone or stop childbearing, a substantial proportion are not using any method of family planning in many developing countries, especially in Sub-Saharan Africa. There are many reasons why women who want to avoid pregnancy do not use contraceptives, including ambivalence about pregnancy, lack of knowledge about contraception, their own or their partner's opposition to family planning, poor access to contraceptive services, fear of side effects and the woman's perception that she cannot become pregnant.¹⁰

Moreover, some of the methods used by married women who want to avoid pregnancy have high failure rates. Although the proportion who rely on such less effective methods as periodic abstinence, withdrawal and other traditional methods is relatively small, these women contribute disproportionately to rates of unintended pregnancy in some countries: Among married women who want to delay or stop childbearing, at least 10% in 17 countries, and more than 20% in seven of those countries, are using less effective methods. The reasons why women opt for less effective methods may include a lack of knowledge about modern methods, religious values that proscribe modern contraceptive use, concern about side effects, partners' objections and difficulty in paying for or obtaining a modern method.

Contraceptive use does not necessarily provide complete protection against pregnancy; each method can fail, even when it is used perfectly. U.S. data from the late 1980s, for example, show that the estimated first-year failure rate for the pill is 8%, while that for the condom is 15%.¹¹ (Failure rates for less effective methods, such as periodic abstinence, are even higher—e.g., 26%.) The DHS data indicate that in 16 of the 18 developing countries for which data are available, more than 10% of contraceptive-use discontinuations in the five preceding years were precipitated by a method failure, and this proportion surpassed 20% in seven of these countries.

Some DHS surveys asked sexually experienced unmarried women who were not using contraceptives whether they would be happy, unhappy or indifferent if they became pregnant soon; in half of the 14 countries for which these data are available, 49-59% said that they would be unhappy with a pregnancy, while in the other half, 60-93% said they would be unhappy. In many of these countries, contraceptive prevalence is very low among young never-married women, even though many of them do not want to become pregnant.¹²

Underlying Reasons for Abortion

Table 2 presents percentage distributions of 26 samples of women from 23 countries according to their most important reason for seeking abortion. (This measure combined responses both to questions that asked women to give their most important reason and to questions that simply asked why they had had an abortion.)

For many women, more than one factor undoubtedly contributed to their decision. In such situations, it may be difficult to identify a single factor as the most important one. Even if a woman identifies one overriding reason, pertinent information would still be lost, because the whole range of reasons guiding the decision would not be measured. Thus, using questions that allow women to give multiple reasons adds another dimension to understanding the factors that underlie the abortion decision.

In Table 3, we present results from seven studies (from four developing countries, all in Asia, and three developed countries) that allowed women to give multiple reasons for why they sought an abortion.** These interpretations should be considered exploratory, however, especially for those countries for which only studies with small sample sizes are available, or where the samples are not nationally representative.

•Timing births and controlling family size. The desire to postpone a birth or to stop childbearing is a very common reason given by women seeking abortion. In almost half of the 23 studies (in 20 countries) with this information, about 50% or more of women gave the birth-timing and family-size control cluster of reasons as their most important reason (Table 2).

In addition, in three South Asian countries (Bangladesh, India and Pakistan), one-half to two-thirds of those giving multiple reasons cited postponing or stopping childbearing, and in the Philippines, roughly one-third did so (Table 3). This category was also important in the three developed countries in Table 3: The proportions citing it as one of many reasons ranged from 20% to 35%. Moreover, 49-67% of Czech and Romanian women, respectively, cited a desire to postpone or stop childbearing as their most important reason for seeking an abortion (Table 2).

Often, this reason stems from other factors in a woman's life that make the timing of the pregnancy undesirable; it thus may reflect a wide range of issues that are detailed under the categories in Table 3. For example, women may need to postpone childbearing because of their or their children's health, or in societies where young unmarried mothers are commonly ostracized for having a child or where early childbearing disrupts education, women may particularly want to postpone the first birth. Moreover, poverty, unemployment and inability to afford to educate any additional children may be behind reasons for restricting family size.

In about half of the countries for which the postponing and limiting reasons could be separated, the proportion of those who cited a desire to limit births as their main reason was higher than that of women who considered postponing to be most important. Where the reasons could not be separated (i.e., in Romania) or where only one of the two categories was reported (typically, the "desire to stop" category), the absent category is likely to have been implicit in the reported one. (For example, if a woman's main reason for having an abortion is that she does not want the pregnancy, this may mean that she does not want it at all or that the timing is bad. Additional probing is needed to clarify and thus separate such responses.)

•Poverty and economic reasons. Economic reasons or women saying that they could not afford to properly care for a child come second overall in importance. The proportion who gave this reason was more than 20% in six of the 19 studies with relevant information (Table 2). The importance of women's economic situation as the main reason for their seeking an abortion was evident in developed as well as developing countries. (U.S. women, for example, tended to explain this reason with a more specific one, such as a baby would disrupt employment or schooling, that the woman or her partner was unemployed and that she lacked support from her partner.¹³) When women were allowed more than one response (Table 3), 30-68% cited poverty as contributing to their decision in four of the seven countries with available data.

Combining the data on the impact of a birth on a woman's education and on her financial situation yields a broader, more inclusive category of socioeconomic reasons. The proportion of women who cited such overall socioeconomic reasons as their primary one for having an abortion is less than 10% in five studies, 10-29% in nine studies, 30-55% in nine and 80-86% in two.

•Relationship problems. Relationship problems, including the partner's objection to carrying the pregnancy to term, are moderately important in explaining why women have abortions. The proportion of women citing such problems as their overriding reason for the abortion reached 25-42% in four studies (Chile, Honduras, Mexico and Nigeria). It was the main reason for fewer than 10% of respondents in nine studies, and for 10-20% in seven studies (Table 2).

Only relatively small proportions (4-14%) of women in the three developed countries with information on relationship problems (the Czech Republic, Romania and the United States) cited it as their main reason for seeking an abortion. Some 19% of women in the Australian study and 16% of those in the Dutch study cited problems with their husband or partner as contributing factors, and 51% of U.S. women and 29% of Australian women mentioned not wanting to be a single mother. Underlying this general reason are such specific ones as that the partner threatened to abandon the woman if she gives birth, that the partner or the woman herself refuses to marry to legitimate the birth, that a break-up is imminent for reasons other than the pregnancy, that the pregnancy resulted from an extramarital relationship, that the husband or partner mistreated the woman because of her pregnancy, or that the husband or partner simply does not want the child. Sometimes women combined these reasons with not being able to afford a baby, suggesting the importance of having a partner who can offer both emotional and financial support.

•Young and unmarried. Being too young or fearing that parents or others would object to the pregnancy is a fairly common reason for having an abortion. In 10 countries, more than 10% of women gave this as their main reason, and 20-37% did so in five of them (three in Latin America and two in Sub-Saharan Africa).

This reason was an especially common primary reason in Honduras and Mexico, where it was cited by about one-third of women. It also was a prominent contributing factor in Australia: One-quarter of Australian women mentioned that being "too young" was a factor in their decision to have an abortion and 15% cited not wanting their parents or others to know about the pregnancy. In the Netherlands, 13% mentioned as a contributing reason for their decision that they were too young, and in the United States, at least 30% cited either being too young or fearing their parents' objections as contributing reasons for their abortion.

•Risk to maternal health. This reason was somewhat important overall, having been cited as the main reason by 5-10% in seven countries and by 20-38% in three (Kenya, Bangladesh and India). This factor is apparently less important in Latin America and in the developed countries included here.

The category of maternal health risk may include risks to either physical or mental health; another area of uncertainty is whether the potential health problem has been identified by a doctor or by only the woman herself. Because a threat to maternal health is often an exception to the law in countries where abortion is illegal, many women may cite this reason because it is socially acceptable and provides a legal or moral justification for abortion. Nevertheless, pregnancy probably poses a real threat to many of these women, because at least a small proportion in almost every country cite it as their overriding reason, regardless of the legal status of abortion.

•Fetal defect. Women rarely report that fetal defects or potential problems for the baby motivated their decision to have an abortion. This probably stems from one or more factors, including the low actual incidence of birth defects, the fact that most women obtain abortions before such defects could be known, and fetal defects are generally not detected in developing countries (where advanced testing and modern medical care are not widely available). Furthermore, in many surveys, this reason may not have had its own separate category, but may have been grouped into an "other" catch-all category. Finally, the reason may have been omitted altogether in some studies.

This reason was recorded in only one-third of the countries, with Indian women the most likely to have given fetal defects as the most important reason (11%); 5-8% of women in three other developing countries (South Korea, Taiwan and Thailand) also cited this as their main reason. In all four of these Asian and South Asian countries, sex selection is believed to play a role in abortion, and in such instances, some women may report that "fetal defect" was the main reason for their abortion.¹⁴ In the United States, only a small fraction (8%) of the women who reported fetal defect as a contributing reason said that they had been advised by a physician that the fetus may be deformed or abnormal, suggesting that many women may be making this determination on their own.

•"Other" reasons. Almost all studies have a residual category of "other reasons." However, fewer than 10% of women cited them as primary in most studies, although the proportion doing so reached 14-16% in the Czech and Sri Lankan studies. In most studies conducted with women who have abortions, the factors that fell into the "other" category were usually unspecified, especially since the studies mostly gave little attention to why women seek abortion in general.

Even though some studies (e.g., the Czech Republic study) listed a category as "other" in the precoded responses offered to respondents,¹⁵ there is no reason to believe that such a category has uniform meaning to women. However, some of the more sensitive reasons for abortion that are likely to be categorized as "other" include rape or incest (which are rarely mentioned), sex selection and pressure exerted by others to have the abortion.

Regional Patterns

•Sub-Saharan Africa. The limited quantitative and qualitative evidence available for Sub-Saharan Africa suggests that the most important reasons why women have abortions in this region relate to socioeconomic factors—specifically, that women perceive pregnancy as disrupting education and employment. This finding agrees with the general perception that the majority of Sub-Saharan African women who obtain abortions are young and unmarried.¹⁶ For adolescents who engage in a sexual relationship with an older man because of financial need, this same need will most likely be their main reason for seeking an abortion if an unintended pregnancy results.

Women in this region also turn to abortion to postpone or limit births, with this proportion as high as 53% in Zambia, 35% in Benin and 13-21% in the Nigerian studies. The pronatalist values prevalent in the region suggest that the majority of women who cite this cluster of reasons are probably seeking an abortion to delay rather than stop childbearing.

Relationship problems are another important reason why women have abortions in this region. Although the quantitative data indicate that at least 20% of women mention this as their overriding reason in only two of the five studies from the region, qualitative data collected in Ghana and Kenya show this to be an important reason behind women's decision to seek abortion;¹⁷ the category includes extramarital relationships, the husband's or partner's denial of paternity, and a lack of willingness or readiness for marriage. While the desire to continue schooling or working is the most important motivation for abortion among young unmarried women in Sub-Saharan Africa, marital problems appear to be a motivation for abortion among married women in the region.¹⁸

•Asia. The most commonly reported primary reason for abortion in the Asian countries studied was the desire to postpone or stop childbearing. Of the two components of this reason—delaying vs. limiting—the latter is more prevalent. This finding is consistent with the widespread preference for smaller families in most Asian countries and with the fact that most abortions in the region are to married women. All of the other reasons tended to play smaller roles in the decision. Although only a minority reported fetal defect and "other" factors as their most important reason, these two categories were nevertheless more commonly reported in Asia than in other regions. Sex selection may be a factor.

•Latin America. Based on limited data for the four Latin American countries listed in Table 2, three categories of reasons compete for the position of primary importance: socioeconomic reasons (combining not being able to afford a baby and disruption to education and work), relationship problems and reasons related to being young.

The first two categories of reasons are probably linked to the high prevalence of consensual unions in Latin America. Such unions have higher dissolution rates than legal marriages, and usually imply less commitment by the cohabiting partners. In deciding whether to carry an unintended pregnancy to term, women in consensual unions are likely to weigh whether they could support the child on their own should the union be dissolved or their partner be unable (or refuse) to provide support.¹⁹ Further, a study of Colombian women who had had an abortion suggests the importance of partnership problems: Only 39% of these women were living with their partner at the time of the pregnancy, and so most may not have been committed to the relationship.²⁰

•Developed countries. In the six developed countries for which information is available, two clusters of reasons appear to be important. In the Czech Republic and Romania, postponing and limiting childbearing is by far the most important reason women gave for why they had had an abortion. However, the richer information obtained from questions on multiple reasons in Australia, the Netherlands and the United States indicates some aspects of the motivation behind "to delay or stop:" For about two-thirds of Australian women, the reason behind postponing or limiting childbearing was not being able to afford a child (Table 3).

A high proportion also reported one or more of a cluster of reasons related to having to postpone a birth at a particular stage of life, such as "having a child would change my life in a way I do not want," "I feel I should establish my career before I have a child," "having a child now will affect my schooling/education" and "I am not ready for that responsibility." Problems of timing are also reflected in other generally stated reasons, which include being too young, not wanting to be a single mother, and the perception that a woman's parents would object to her having a child at that time. Relationship problems and the objections of a husband or partner are also somewhat important in the five developed countries represented here.

Correlates of Reasons

How closely are a woman's reasons for abortion related to her socioeconomic and demographic characteristics? We address this question by examining how the reasons vary by three characteristics—the woman's age, marital status and level of education—in 10 countries. (For two of these countries, Australia and the Philippines, the data are based on the number of responses to an item that allowed women to specify multiple reasons.)

•Age. A woman's age is only moderately associated with why she seeks an abortion (Table 4). In four of the five countries for which data are available on postponing childbirth as a reason for abortion, women younger than age 25 were more likely than those aged 25 and older to say the reason for their abortion was to postpone childbearing. The exception was Zambia, where only 29% of younger women said they sought their abortion for timing purposes, compared with 71% of older women.

On the other hand, in all countries except Romania—for which postponing childbirth and limiting births could not be separated as reasons—older women were more likely than younger women to say their abortion was motivated by a desire to stop childbearing. For example, among ever-married women in Turkey, 65% of those aged 25 and older cited wanting no more children as their reason for seeking abortion, compared with 34% of women younger than 25.

However, in three of the eight countries, younger women were more likely than older women to mention socioeconomic factors as their reason (Turkey, the United States and Zambia); the reverse was true in just one country (the Czech Republic), and there was virtually no difference by age in the remaining four (Australia, Colombia, Finland and Romania). As expected, in all five countries in which the "too young" reason was studied, younger women were more likely than older women to note that they were too young or feared their parents' objection.

•Marital status. A desire to stop childbearing and socioeconomic circumstances appear to be the most prominent reasons why married women have abortions,*† while socioeconomic factors and young age or parental objections are the two most important ones among unmarried women. Marital status makes no difference in the likelihood of citing a desire to postpone childbearing as the main reason for having an abortion. However, in five of the seven countries with available data, unmarried women were at least as likely as married women to cite socioeconomic reasons as most important; as expected, the proportion citing being underage and parents' objections as their main reason was consistently higher among unmarried women than married women.

Unmarried women were more likely than married women to say their abortion was mainly motivated by relationship problems (for example, 22% vs. 5% in the Czech Republic, and 30% vs. 4% in the Philippines). In all seven countries, but especially in Colombia, the Czech Republic and the United States, reasons of maternal or fetal health tended to be more important among married than unmarried women.

•Education. No clear association emerged between women's educational attainment and their main reasons for seeking an abortion. Studies conducted in five countries show that the profile of reasons why women have abortion is very similar among both more and less educated women.

Discussion

The universality of the phenomenon of unintended pregnancy illustrates that, worldwide, women and couples have great difficulty in successfully planning births. In the majority of the 49 developing countries for which we examined fertility survey data, a high proportion of women would like to postpone having a child or to stop altogether, but are not using an effective contraceptive method. Even where effective use is quite high, women continue to experience unplanned pregnancy, because of either contraceptive failure or unanticipated changes in their life circumstances, or sometimes as a result of their own ambivalence.

The analysis of the reasons women give for why they had an abortion shows that the most commonly reported ones are postponing childbearing to a more suitable time or stopping altogether to focus energies and resources on existing children. The fact that these two reasons were less important in Latin America and the United States than in Asia and some of the other developed countries may partly be explained by the high prevalence of sterilization at relatively young ages in these first two regions, which reduces the need for abortion to limit family size. The desire to delay or stop childbearing probably reflects a number of underlying, more specific reasons for not wanting to have a child at that time.

The second most commonly reported reason consists of socioeconomic factors, such as being unable to afford a child—either in terms of the direct costs of raising a child or the opportunity costs to a woman who, to care for a child, must interrupt her education or work. This set of reasons is particularly prominent in Sub-Saharan Africa, where the majority of women who seek abortion tend to be young and unmarried, and where pregnancies that end in abortion are likely to occur in unstable relationships.*‡

In the Latin American countries for which we have information, relationship problems are among the most important reasons why women seek abortion; in these societies, where many women are in consensual unions, the issue of being able to support the child should the relationship end is probably a major concern. Being unable to afford a child is also an important reason why women obtain an abortion in the United States.

While at least a small proportion of women in most countries mentioned the risk to their health as their primary motivation for the abortion, this reason was relatively more prevalent in Sub-Saharan Africa and South Asia than in other regions. This finding is not surprising, since we expect abortions for maternal health reasons to be related to large family size and close birth spacing, factors that are much more common in these two regions than in the others.

Only small proportions of women mentioned a risk to fetal health either as the most important reason for having an abortion or as a contributing one. However, in some studies, particularly those in Asian countries, a substantial minority of women mentioned fetal defects as the most important reason for their abortion. We speculate that this category may include women who chose an abortion because of the sex of the fetus. The generally low prevalence of such reasons may stem from women's poor access to modern diagnostic tests in most developing countries rather than to any real population difference in the prevalence of fetal defects.

Social acceptability may play some role in determining the reasons women give for seeking abortion. Reasons such as postponing or limiting births may be the first or immediate response that comes to mind because these two may be viewed as expected or acceptable reasons. Other reasons, such as sexual coercion and abuse, do not feature prominently in any available studies, probably because relatively few pregnancies result from such situations. But, given the sensitivity of the circumstances surrounding sexual coercion or abuse as a reason for abortion, many women may be reluctant to report them, even when they might be true.

The patterns in the relationships between women's characteristics and their reasons for obtaining abortions suggest that these reasons are not random, but relate closely to the woman's current situation and aspirations. For example, older women who have had as many children as they want typically report that their abortion was motivated by a desire to prevent adding to an already large family. In many cases, these women also mention other related reasons, such as being unable to afford another child. On the other hand, unmarried women are more likely than married women to say they chose an abortion because they are too young to have a child, because a baby would have a negative impact on their education or work, or because they fear the reactions of parents or others if they carry to term.

The scarcity of research on why women have abortions points to the need for more work in this area. A comparison of results from studies that obtained only the most important reason for abortion and from others that allowed women to give multiple responses shows that the decision is likely to be motivated by more than one factor.

Thus, studies should obtain both types of information, because they are clearly complementary. While information on the most important reason for the decision is valuable, probing to further clarify women's answers and to request other reasons that may have contributed to the decision is essential. More in-depth questioning may succeed, for example, in obtaining the specific reasons that underlie the general response that an abortion was sought to space births or to control family size. It may also be useful to conduct studies at longer intervals since the abortion, to evaluate whether women change their reasons over time as a way of coping with their decision. Research on the degree or intensity of fertility preferences would also help to better understand why women choose to have an abortion rather than carry to term.

Evidence from one U.S. study shows that the reasons why women do not want a child at the moment are remarkably similar to the reasons women give for obtaining an abortion.²¹ Studies of all women who have an unplanned pregnancy that examine the reasons why it was unplanned and why women opt for either of the available alternatives—birth or abortion—would greatly enhance our understanding of the personal and structural factors that shape women's decisions about whether and when to have a child. Such cohort studies following women who choose different paths are rarely undertaken and should be encouraged.

Despite the scarcity of studies on reasons why women obtain abortions and the limitations of the existing research, the evidence presented here points to the usefulness of such exploratory research. More investigations and improved research approaches are crucial to better understand the complex situations and processes that lead to unintended pregnancies and to women's decision to end them through abortion. Such an understanding would increase the chance that policymakers and providers respond humanely and effectively to the varied situations and needs that lead to the decision to resolve unwanted pregnancy through abortion.

At one extreme, in countries where abortion is illegal, this understanding could motivate better treatment of women who seek medical care for complications from unsafe abortions. At the other extreme, in all settings (including those where legal abortion is safe and accessible), a greater appreciation of the roles that partners and other family members play may convince policymakers and counselors of the need to stress social and family support for women at risk of unintended pregnancy.

The research reviewed here supports the conclusion that improved contraceptive practice is an important means of reducing abortion. However, it also suggests that some unplanned pregnancies and abortions are difficult to prevent, because of limits to individuals' ability to determine and control the circumstances of their lives.

Appendix: Sources
The following are the 32 studies from which data on reasons for abortion were obtained. Listed are the country and year of data collection, the type of study, the data collection approach, the sample size and the marital-status composition of the sample.

Africa

•Benin, 1993: subnational hospital/clinic-based survey; one-year prospective study of abortion patients, with or without complications, hospitalized in three hospitals; N=380; all marital statuses (71% married). Source: Alihonou E, Goufodji S and Capo-Chichi V, Morbidity and mortality related to induced abortions, African Journal of Fertility, Sexuality and Reproductive Health, 1996, 1(1):58-65.

•Kenya, 1991: subnational fertility survey; cross-sectional knowledge, attitudes and practices study of nurses; N=218; all marital statuses (77% married). Source: Kidula N, A survey of knowledge, attitude and practice of induced abortion among nurses in Kisii District, in Rogo K, Leonard A and Muia E, see reference 16.

•Zambia, 1985-1986: subnational hospital/clinic-based survey; 10-month prospective study—medical record review and interviews of women requesting legal terminations, and women presenting with complications; N=264; all marital statuses (60% single). Source: Likwa RN and Whittaker M, The characteristics of women presenting for abortion and complications of illegal abortions at the University Teaching Hospital, Lusaka, Zambia: an explorative study, African Journal of Fertility, Sexuality and Reproductive Health, 1996, 1(1):43-49.

Asia

•Bangladesh, 1989-1990: subnational hospital/clinic-based survey; medical record excerpts and interviews with patients admitted with a diagnosis of abortion; N=1,301; all marital statuses (more than 90% currently married). Source: Begum SF et al., Hospital-Based Descriptive Study of Illegally Induced Abortion—Related Mortality and Morbidity, and Its Cost on Health Services, Dhaka, Bangladesh: BAPSA, 1991.

•Bangladesh, 1995-1996: subnational hospital/clinic-based survey; case histories collected from women admitted with abortion-related complications; N=53; all marital statuses (98% married). Source: Bangladesh Association for Prevention of Septic Abortion (BAPSA), Septic abortion: results from an anthropological study, MR Newsletter, 1997, 13(1):1-5.

•India, 1977-1978: official government statistics; records of women coming for legal abortion in West Bengal; N=13,511; all marital statuses (88% married). Source: Dutta R, Abortion in India, with particular reference to West Bengal, Journal of Biosocial Science, 1980, 12(1):191-200.

•India, 1990: subnational hospital/clinic-based survey; interviews conducted with abortion patients prior to consultation with doctor; N=1,197; married women only. Source: Khan ME et al., Abortion acceptors in India: observations from a prospective study, in International Population Conference, Montreal, 1993, Volume 1, Liége, Belgium: International Union for the Scientific Study of Population (IUSSP), 1993, pp. 253-267.

•Indonesia, 1987-1988: subnational hospital/clinic-based survey; data collected in family planning clinics from women seeking menstrual regulation services (97% aged 15-25); N=200; unmarried women only. Source: Widyantoro N, Enhancing the quality of women's reproductive health care: an experimental approach in Indonesia, paper presented at the Christopher Tietze International Symposium, Rio de Janeiro, Brazil, Oct. 29-30, 1988.

•Malaysia, 1981: subnational hospital/clinic-based survey; interviews of abortion patients; N=148; all marital statuses (91% married). Source: Foong CS, A study of characteristics of women seeking induced abortion, Medical Journal of Malaysia, 1982, 37(4):318-321.

•Pakistan, 1994: subnational survey; interviews of key informants; N=30; ever-married women only. Source: see reference 6.

•Philippines, 1979: subnational hospital/clinic-based survey; interviews of selected providers and women who had had an abortion; N=392; all marital statuses (75% married). Source: Gallen M, Abortion in the Philippines: a study of clients and practitioners, Studies in Family Planning, 1982, 13(2):35-44.

•Singapore, 1984: subnational hospital/clinic-based survey; interviews of abortion patients; N=400; all marital statuses (84% married). Source: Tsoi WF, Tay GE and Ratnam SS, Psychosocial characteristics of repeat aborters in Singapore, Biology and Society, 1987, 4(2):78-84.

•Singapore, 1985: official government statistics; data collected on women obtaining legal induced abortions (reasons given for married women only); N=23,512; all marital statuses. Source: Saw S, Seventeen years of legalized abortion in Singapore, Biology and Society, 1988, 5(2):63-72.

•South Korea, 1994: national fertility survey (National Fertility and Family Health Survey); interviews with representative sample of women; N=5,183; married women only. Source: Hong MS et al., 1994 National Fertility and Family Health Survey Report, Seoul, Korea: Korea Institute for Health and Social Affairs, 1994.

•Sri Lanka, 1988-1990: subnational survey; interviews with married women who reported having had an induced abortion; N=548; married women only. Source: Pereta DC and Rajapaksa LC, A community-based study on socio-demographic characteristics of women reporting an induced abortion, Ceylon Journal of Medical Science, 1991, 34(2):63-74.

•Taiwan, 1980-1981: national fertility survey; interviews with mothers of children in elementary schools; N=2,176; all marital statuses (majority married). Source: Wang JF, Contraception, psychological responses, social support and coping after abortion in Taiwan, Journal of Nursing Science, 1995, 1(1-2):1-15.

•Thailand, 1983-1984: national survey of facilities; records of abortion patients from sampled hospitals (reasons only for legal induced abortions); N=5,701; all marital statuses (91% married). Source: Koetsawang A et al., Report of the Nation-Wide Study on Health Hazard of Illegally Induced Abortion, Bangkok, Thailand: Mahidol University, 1987.

•Turkey, 1993: national fertility survey; (Turkish Demographic and Health Survey); interviews with representative sample of women; N=6,519; ever-married women (96% currently married). Source: Akin A and Bertan M, Contraception, Abortion, and Maternal Health Services in Turkey: Results of Further Analysis of the 1993 Turkish Demographic and Health Survey, Ankara, Turkey: Ministry of Health; Ankara, Turkey: Hacettepe University; and Calverton, MD, USA: Macro International, 1996. (Micro data files from this study were also used for original analyses.)

Latin America

•Chile, 1988: subnational hospital/clinic-based survey; interviews of women admitted for complications of induced abortion in three hospitals; N=357; all marital statuses. Source: Weisner M, Induced abortion in Chile, with references to Latin American and Caribbean countries, paper presented at the annual meeting of the Population Association of America, Toronto, Canada, May 3-5, 1990.

•Colombia, 1990-1991: subnational hospital/clinic-based survey; interviews of a sample of clinic clients in Bogotá requiring treatment for incomplete abortion; N=602; all marital statuses. Source: Meja JV and Téllez MM, Embarazo Indeseado y Aborto, Bogotá, Colombia: Editorial Presencia, 1992.

•Honduras, 1992-1993: subnational hospital/clinic-based survey; interviews of abortion patients aged 15-35; N=30; all marital statuses (63% cohabiting or in partnership, 27% married). Source: Kennedy M, El aborto: enfoque psicosocial y de salud pública, paper presented at the Meeting of Researchers on Induced Abortion in Latin America and the Caribbean, Bogotá, Colombia, Nov. 15-18, 1994.

•Mexico, 1967-1971: subnational survey; interviews of women hospitalized for abortion complications; N=3,714; all marital statuses. Source: Ordóñez BR, Induced abortion in Mexico City: summary conclusions from two studies conducted by the Mexican Social Security Institute, in Pan American Health Organization (PAHO) and Transnational Family Research Institute, in The Epidemiology of Abortion and Practices of Fertility Regulation in Latin America, Washington, DC, USA: PAHO, 1975.

•Mexico, 1988: subnational survey; interviews with the interviewers' personal acquaintances (who had had an abortion) and with additional abortion patients referred by these aquaintances; N=156; all marital statuses (35% married). Source: Pick de Weiss S and David HP, Illegal abortion in Mexico: client perspectives, American Journal of Public Health, 1990, 80(6):715-716.

Developed Countries

•Australia, 1992: subnational hospital/clinic-based survey; self-administered questionnaire distributed by staff to abortion patients at 11 clinics; N=2,249; all marital statuses (33% married or in de facto unions). Source: Adelson PL, Frommer MS and Weisberg E, A survey of women seeking termination of pregnancy in New South Wales, Medical Journal of Australia, 1995, No. 163, pp. 419-422.

•Czech Republic, 1993: national fertility survey; nationally representative sample of women aged 15-44; N=2,249; all marital statuses (64% married, 3% consensual union). Source: Czech Statistical Office et al., see reference 2. (Micro data files from this study were also used for original analyses.)

•Finland, 1993: official government statistics; women obtaining legal induced abortions; N=10,342; all marital statuses (27% married). Source: Hämäläinen H, Rasimus A and Ritamo M, Tilastotiedote Statistikmeddelande: Aborttitlasto 1993, Helsinki, Finland: National Research and Development Centre for Welfare and Health (STAKES), 1995, No. 14.

•Netherlands, 1983-1987: subnational hospital/clinic-based survey; clinic admission statistics of women of Caribbean descent who had had an abortion; N=230; all marital statuses (12% formally married, 16% common-law marriages). Source: Lamur HE, Characteristics of Caribbean-born women having abortions in an Amsterdam clinic, Genus, 1993, IL(3-4):135-145.

•Romania, 1993: national fertility survey; national household sample of women aged 15-44; N=4,772; all marital statuses (63% married, 4% consensual unions). Source: Romanian Ministry of Health, see reference 2.

•United States, 1987-1988: national survey of facilities; questionnaire distributed to abortion patients; N=1,900; all marital statuses. Source: see reference 8. (Micro data files from this study were also used for orginal analyses.)