The current trend toward liberalization of abortion laws in many countries throughout the world has generated a proliferation of statistical analysis of the incidence of induced abortion, characteristics of women having abortions, procedures, and morbidity and mortality associated with the operation. As a preface to data to be presented in this report on this subject, the first four sections review briefly the terminology that has been established in the field; describe the sources of data, including methods of estimation: summarize the methods of statistical analysis that appear in the report; and discuss the correlation between abortion experience and contraceptive experience.

DEFINITIONS

As used by the medical profession, the term "abortion" denotes the termination of a pregnancy before the fetus has attained viability, that is, before it has become capable of independent extrauterine life. Viability is defined in terms of the duration of the pregnancy and/or fetal weight. According to tradition, viability is attained after 28 weeks of gestation, counting from the first day of the last normal menstrual period, corresponding to a fetal weight of approximately 1,000 gram. This definition was based on the observation that infants below this weight had little chance of survival, whereas the mortality of infants above 1,000 gram declined rapidly with increasing weight.

In recent years, many medical textbooks, especially in the United States, have lowered the upper limits of abortion to 20 weeks' gestation or a fetal weight of 500 gram, because some of these tiny infants have, in fact, survived. The term "immature birth" is used to describe the delivery of a fetus weighing between 500 and 1,000 gram. From a clinical point of view, a further distinction in the timing of abortion can be made by separating those at 12

weeks or less, or in the first trimester of pregnancy, from those at 13 weeks or more, or in the second trimester.

The two major categories of abortions are known as induced and spontaneous. Induced abortions are those initiated voluntarily with the intention to terminate pregnancy; all other abortions are considered spontaneous, even if an external cause is involved, such as a trauma or a high fever. In this report, unless otherwise noted, the term abortion refers to induced abortion, including legal and illegal abortions, as defined by the laws of each country. Where abortions are illegal, they are frequently performed by unqualified persons under un

sanitary conditions, resulting in an increased risk of infection and other complications. This higher risk also applies to self-induced abortions, regardless of their legal status.

Diagnosis of illegal or of selfinduced abortion in most cases requires information from the woman involved, members of her family, or the abortionist. In the absence of such information or of evidence of manipulation, such as injury to the cervix or perforation of the uterus, it is rarely possible, either clinically or at autopsy, to differentiate between spontaneous and induced abortion. Most septic abortions are thought to be induced, but many induced abortions show no signs or symptoms of infection, whereas some spontaneous abortions do.

Such diagnostic categories as imminent, inevitable, and incomplete abortion describe stages in the process of abortion and are significant only in relation to a specific point in time. After an abortion has been initiated, a woman may seek medical care at any stage of the process, although a more experienced woman may wait until she feels sure that the pregnancy can no longer be saved against her wishes. Incomplete abortion is the most common diagnosis when women are admitted to hospitals for

aftercare, usually for postabortal bleeding caused by retention of placental tissue.

SOURCES OF DATA

Most of the tables in this report are based on official statistics of legal abortions from countries where liberalized abortion laws have been enacted at various times since the late 1930s. Most of these laws provide for the reporting to health authorities of all abortions performed under the provisions of the statute, and in most of the countries, but not in all, nationwide statistics are published periodically. The completeness of reporting and the quantity and quality of the tabulations vary among countries.

Official statistics of legal abortions have been supplemented, in some cases, by the results of studies sponsored by private organizations or conducted by individual scholars. The most comprehensive of these studies, used extensively in this report, is the Joint Program for the Study of Abortion (JPSA) in the United States, sponsored by the Population Council (115, 118). JPSA extended over a period of 12 months, from July 1970 through June 1971, and obtained medical records of 73,000 abortions from the Ob/Gyn Departments of 60 teaching hospitals and from six free-standing clinics in 12 states and the District of Columbia, about oneseventh of all legal abortions in the United States.

Although information on contraceptive practices has been successfully obtained in many countries by interviewing samples of women (usually married women) and occasionally of men, drawn from the general population, efforts to estimate the incidence of illegal abortion by such surveys have been far less successful. It is quite apparent that many respondents who are willing to reveal their contraceptive practices to an interviewer will not report experience

with illegal or even legal abortion. The classic example is the Hungarian Fertility and Family Planning Study of 1966, conducted more than a decade after the legalization of abortion in that country (56). In that study, the numbers of abortions reported by the respondents for the years 1960-1965 corresponded to 50-60 percent of the number actually performed, according to reports from hospitals.

Some data based on surveys of samples of general populations have nevertheless been included in this report (Tables 2, 3, and 15), because they throw additional light on such aspects of abortion as regional differences and the percentage of women with abortion experience. These data were selected because the reported abortion rates and ratios were high for at least one segment of the population and, therefore, more "plausible" than the low rates and ratios obtained in other surveys. The possibility cannot be excluded, of course, that the selected surveys do not represent populations with a particularly high incidence of abortion, but rather women more willing than others to reveal their abortion experience.

Some investigators have tried to estimate the incidence of abortion on the basis of interviews conducted in various medical settings: with obstetrical patients, gynecological patients, or with women attending birth control clinics. Although rapport in these medical situations may be better, as a rule, than in door-to-door interviewing, the results may be severely distorted by a number of selective factors (109, 141). Hence, no data of this type are shown in this report.

A major step forward in the methodology of abortion research involves the use of the "randomized response technique" (RRT), which permits respondents to give information on highly sensitive subjects without revealing the facts to the interviewer (138).

In its simplest form, RRT in

volves the use of a box containing a number of small beads in two colors, for example, 35 red beads and 15 blue beads. The box has a large window, allowing a view of its contents, and a smaller window through which only a single bead can be seen. Two alternate questions are provided: one sensitive and the other nonsensitive. For example, a red bead visible through the small window may mean "Have you had an abortion during the past 12 months?" whereas a blue bead means "Were you born in the month of March?" The respondent is asked to shake the box and to answer the question indicated by the color of the bead visible in the small window with either "yes" or "no." The interviewer records the answer but does not know which question is being answered.

The percentage of women who had an abortion during the past year (P) can be estimated from the percentage of respondents answering "yes" (P') according to the following formula:

requires larger samples than direct questioning.

RRT has been applied to abortion in the United States and in Taiwan (1, 19). In the latter country, one-half of a sample of 2,200 women was asked directly whether they had ever had an induced abortion. RRT was administered to the other half. Among the group directly questioned 12.7 percent admitted experience with abortion, whereas for those interviewed by RRT the corresponding estimate was 28.2 percent, or more than twice as high. However when RRT was administered to a small group of women known to have had abortions, the estimated proportion with abortion experience was only 40 percent, suggesting a considerable amount of prevarication, even with RRT (20).

STATISTICAL ANALYSIS

Two distinct approaches to the assessment of the incidence of abortion involve relating the number of abortions (1) to the number of women at risk of having them and (2) to the number of births or pregnancies. The types of information obtained by these two approaches are different but complementary. This report follows the convention that measures relating abortions to women (or the total population) are referred to as rates whereas measures relating abortions to births or pregnancies are ratios (141). Since this convention is not universally accepted, the relevant denominator is identified whenever necessary. Specifically, the following abortion rates and ratios appear in this report: 1. Abortion rates per 1,000 population, corresponding to crude birth rates (Table 1).

2. Abortion rates per 1,000 women of reproductive age, defined as 15-44 years, corresponding to general fertility rates (Table 1).

3. Age-specific abortion rates per 1,000 women in five-year age groups, corresponding to agespecific fertility rates (Table 5). Age

specific abortion rates for women under 20 years of age are computed per 1,000 women aged 15-19, and those for women 40 or over, per 1,000 women aged 40-44 years.

4. Total abortion rates per woman (Table 1), corresponding to total fertility rates. These rates are computed by adding the age-specific abortion rates for quinquennial age groups, multiplying the sum by five, and dividing the product by 1,000.

5. Abortion ratios per 1,000 live births (Table 1). In the computation of these and all other abortion ratios in this report, the number of live births in the denominator is the number occurring during a 12month period starting six months later than the period during which the abortions occurred. This shift of six months is necessary if the abortions and births are to represent a reasonable approximation of the same cohort of conceptions.

6. Abortion ratios per 1,000 abortions plus live births (Table 1). This type of abortion ratio is a substitute for the ratio per 1,000 pregnancies, ignoring those ending in spontaneous abortions or stillbirths and, in most cases, illegal abortions as well. An abortion ratio per 1,000 live births (R) can be converted into a ratio per 1,000 abortions plus live births (R) by the formula: R' R. (1,000+R).

==

7. Age-specific abortion ratios per 1,000 live births (Table 6). In the computation of these ratios, abortions were "de-aged" by three months and births (six months later) by nine months in order to approximate distributions by age at conception. The procedure is described in Appendix A.

8. Abortion ratios per 1,000 live births specific for parity (Table 8) and for marital status (Table 10).

All rates and ratios shown in Tables 1, 5, 6, 8, and 10, are period rates, not cohort rates. We have been able to locate only a single set of data from which it was possible to compute age-specific abortion

The outstanding difference between the period rates and the cohort rates is the much higher level of the period rates among women under age 35 years. The total abortion rate per woman was 2.72 for the period 1961-1965 and 3.47 for 1966-1970, compared with 1.85 for the birth cohorts 1926-1930.

Tables 2 and 3 show three indicators of lifetime experience with abortions, customarily obtained in population surveys. For all age groups combined, these indicators are neither period rates nor cohort rates. For the oldest age group at the time of the interview, representing women who have passed their fertile period, the mean number of abortions per woman is the equivalent, in cohort terms, of the total abortion rate shown in Table 1, and the abortion ratio per 1,000 live births is also a measure for the entire life span of the cohort. The third indicator, the percentage of women with abortion experience, can also be defined as a cumulative rate of first abortions; it is rarely possible to compute the corresponding rate from period data.

The importance of making a sharp distinction between abortion rates and abortion ratios is illustrated by the following example: Assume two countries, each with a population of 1 million, one with 45,000 live births and 9,000 abortions, the other with 15,000 live births and 6,000 abortions. The abortion rates are 9.0 and 6.0, respectively, per 1,000 population,

whereas the abortion ratios are 200 and 400, respectively, per 1,000 live births, that is, one of these two countries has the higher abortion rate and the other has the higher abortion ratio. A given abortion rate corresponds to a higher abortion ratio when the birth rate is low than when the birth rate is high.

ABORTION AND
CONTRACEPTION

Because induced abortion and contraception share the prevention of unwanted births as a common objective, a high correlation between abortion experience and contraceptive experience can be expected in populations in which some couples have attempted to regulate the number and spacing of their children while others have not wanted to do so. In such populations, women who have used contraception are more likely to have had abortions than those who have not used contraception, and the contraceptors will also have higher abortion rates and ratios than the noncontraceptors.

This apparent paradox is immediately resolved if the comparison is limited to couples all of whom are determined to avoid or to defer childbirth but not all of whom practice contraception. In such a population the abortion ratio will be very high because most unwanted pregnancies will be aborted. However, those women who (or whose sexual partners) practice contraception consistently and effectively will experience fewer unwanted pregnancies and therefore a lower abortion rate, over a period of time, than women who use contraception ineffectively or not at all.

Another aspect of the complex relationship between abortion and contraception concerns the number of abortions required to replace one live birth. In reference to a given pregnancy the answer to this question is of course "one" since a preg

nancy can be terminated only once, but a different answer is required when one considers either the life of a woman over a period of years or an entire population.

The interval between two successive conceptions has three components: (a) the pregnancy itself; (b) an anovulatory period following the pregnancy during which conception cannot occur; and (c) an ovulatory period during which the monthly probability of conception is more than zero but less than one.

Induced abortion reduces components (a) and (b) because the duration of pregnancy is shortened from about nine months for a live birth to about three months for an abortion, and the subsequent anovulatory period is also shortened following an abortion compared with a live birth, the magnitude of the difference depending on the extent and average duration of breastfeeding. The average time required for conception during the ovulatory period (c) is presumably not affected. With breastfeeding of moderate to long duration and without contraception, component (b) is comparatively long and component (c) is short, the net effect being that at least two abortions are required to avert one live birth. If contraception is practiced widely with even moderate effectiveness, (c) is extended relative to (a) and (b) with the result that only slightly more than one abortion is required to avert

one birth (89, 141). Although the model is crude, it makes the point that abortion alone is an inefficient method of fertility regulation, becoming progressively more efficient as the expanding use of contraception relegates its role to a backstop

measure.

By way of illustration, the uncontrolled fertility of a human population may be conservatively represented by a total fertility rate of 7.0 live births per woman during her life time. To reduce this rate to the level needed for the maintenance of a stationary population (2.2) by the use of abortion alone, without contraception, a total abortion rate of at least 9.6 per woman would be required. If, on the other hand, this hypothetical population practices contraception universally with a fairly high degree of effectiveness (95 percent), using abortion to terminate pregnancies resulting from contraceptive failure,

exclusive reliance on abortion to universal use of contraception, effective at a 95 percent level, with abortion as a backstop in all cases of failure, is also assumed to follow a logistic path.

The time intervals indicated be low are not necessarily of equal duration, and each could extend over years or even decades. In the course of the transition, the total abortion rate per woman and the total pregnancy rate (live births plus induced abortions, excluding stillbirths and spontaneous abortions) both increase, then decline.

Since World War II, total fertility rates in most of the developed countries have ranged from 2.0 to 3.0 per woman. According to the assumptions of the model, total fertility rates of this magnitude imply total abortion rates ranging from 0.7 to about 1.8 per woman, corresponding to annual abortion rates of about 30 to 50 per 1,000 women aged 15-44 years and abortion ratios on the order of 300 to 600 per 1,000 live births. Rates and ratios of legal abor

tions of this order of magnitude, and in some cases higher, have been reported from a number of countries; for countries with restrictive legislation and with fertility rates on the same order of magnitude as in countries with liberal abortion laws, equally high rates and ratios of illegal abortion can be postulated.

A lower incidence of abortion than indicated above implies one or more of the following conditions: (1) a low level of reproductive health with a high prevalence of sterility and subfecundity and/or a high incidence of spontaneous abortions and stillbirths; (2) widespread deferment and avoidance of marriage combined with little or no coital activity among unmarried women; (3) universal practice of contraception at a level of effectiveness exceeding 95 percent, including widespread acceptance of surgical sterilization by women and/or men; and (4) desired or intended family size substantially below the level achieved, with acceptance of some or all unwanted pregnancies.

ABORTION INCIDENCE AND
POLICIES

The number of pregnancies terminated each year by induced abortion throughout the world is not known. The widely used figure of 30 million, corresponding to an abortion rate of almost 40 per 1,000 women of reproductive age and to an abortion ratio of about 240 per 1,000 live births, is a highly speculative estimate, not supported by hard data. Legal abortions reported in 17 countries with reasonably complete statistics, shown in Table 1, totaled less than 1,700,000 according to the latest available statistics, generally for 1971 or 1972. However, this total excludes the three countries with the presumably largest numbers of legal abortions: China, the USSR, and Japan. No reliable method has yet been developed to estimate the numbers of illegal abortions.

Among countries, the legal status of induced abortion ranges from

complete prohibition to elective abortion at the request of the pregnant woman. Appendix B (pp. 5152) presents the most recent available information. The situation can be summarized as follows: About 7 percent of the world's population live in countries where abortion is prohibited without exception and 12 percent in countries where abortion is permitted only to save the life of the pregnant woman. About 15 percent live under statutes authorizing abortion on broader medical grounds, that is, to avert a threat to the woman's health, rather than to her life, and sometimes on eugenic and/or juridical grounds (rape, etc.) as well. Countries where social factors may be taken into consideration to justify termination of pregnancy account for 22 percent of the world's population and those allowing elective abortion for at least some categories of women, for 36 percent. No information is available for the remaining 8 percent; it would appear, however, that most of these people live in areas with restrictive abortion laws.

It should be noted that the abortion statutes of many countries are not strictly enforced and some abortions on medical grounds are probably tolerated in most places. It is well known that in some countries with very restrictive laws abortions can be obtained from physicians openly and without interference from the authorities. Conversely, legal authorization of elective abortion does not guarantee that abortion on request is actually available to all women who may want their pregnancies terminated. Lack of medical personnel and facilities or conservative attitudes among physicians and hospital administrators may effectively curtail access to abortion, especially for economically or socially deprived women.

Because the amount and kind of information available for each country or region are closely related to the legal status of abortion, the discussions of incidence will be preceded, in each instance, by a short

exposition of the applicable statutes and policies. The numerical data are shown in Tables 1-3.

Most of the statistics shown in Table 1 are based on official reports showing the numbers of abortions performed according to the laws of the countries concerned. Some of the data are estimated from official statistics with allowance for underreporting. A few of the estimates are based on surveys of hospitals and still others represent the total number of induced abortions, legal and illegal. Tables 2 and 3 show three indicators of lifetime experience with induced abortion, mostly illegal, for a few selected countries.

THE AMERICAS

United States

In the United States, legislation on abortion has traditionally been the responsibility of the several states. Restrictive legislation, enacted in the 19th century, remained in force during the first two-thirds of the 20th century. The laws of most of the states stipulated a threat to the life of the pregnant woman as the sole legal ground on which abortion could be performed, and in a few states, a serious threat to the womIan's health as well.

A more liberal type of legislation was proposed by the American Law Institute (ALI) in 1955 (3). The relevant paragraph of its Model Penal Code would have permitted abortion if a licensed physician "believes there is substantial risk that continuance of the pregnancy would gravely impair the physical or mental health of the mother or that the child would be born with grave physical or mental defect" or if the pregnancy resulted from rape, forcible or statutory, or incest. During the 1960s, liberalization of the laws regulating abortion became a controversial issue in many states, and, beginning with Colorado in 1967, about a dozen states enacted legislation based on the ALI Model Penal Code.