system exists in the United States, comparable data were obtained from special surveys of hospitals.

In a few instances, the statistics on mortality following legal abortion in Table 24 are based on the general system of vital registration and cover only those deaths which are attributed to abortion on the death certificate. Generally speaking the categories of "attributed deaths" and "associated deaths" overlap to a considerable extent; however, attributed deaths, by defition, exclude deaths assigned to causes other than abortion, and associated deaths may exclude delayed complications with fatal outcomes. By matching individual records of attributed and associated deaths a maximum estimate of mortality can be obtained. This type of estimate was made for England and Wales for the period 1968-1971.

The earliest data on mortality following legal abortion originated in Sweden and Denmark, where the frst steps toward the liberalization of abortion laws were taken in the 1930s. The longest continuous series, that for Sweden, indicates a dramatic decline in mortality over the last quarter of a century (from 252 to 8 per 100,000 legal abortions).

A comparable series of statistics for Denmark, extending over a longer period but with two breaks in continuity, also shows a marked decline in mortality from 195 to 33 per 100,000. (The rate for Sweden in 1960-1961 was 40 per 100,000 legal abortions.)

Nationwide statistics for Japan, available for a period of 22 years, should be interpreted with great caution. On the one hand, the numbers of deaths are those attributed to legal abortion in the general cause-of-death statistics, representing a narrower definition than deaths associated with abortion available for most other countries. On the other hand, the reporting of legal abortions in Japan is incomplete, according to qualified Japanese scholars, which tends to inflate mortality ratios computed on the basis of reported abortions. It is worth noting, however, that the low mortality ratio based on national statistics for Japan for the early 1950s is identical with a ratio based on a special survey of gynecologists made at about the same period of time.

In eastern Europe, Czechoslovakia and Hungary provide the most reliable national statistics on mor

tality following legal abortion. Levels of mortality in these countries have been far below those reported from northern Europe, but even these low levels have shown declines in recent years, reaching 2 per 100,000 in Czechoslovakia and 1 per 100,000 in Hungary, for the latest available period.

Mortality associated with legal abortions in England and Wales in 1968-1969 (30 per 100,000 according to the maximum estimate) was almost as high as in Sweden and in Denmark in the late 1950s. By 1972, it had declined by more than onehalf to 13 per 100,000.

In the United States, high levels of mortality from legal abortion prevailed during the period of restrictive legislation when a significant proportion of the women undergoing abortion suffered from preexisting complications which made them poor risks for any type of surgery. In recent years, with abortion more easily available in several states, mortality ratios have declined to about 5 per 100,000 in California and in New York City. Reports for 1972 based on deaths attributed to abortion suggest a ratio of 3 per 100,000 for the entire country.

Total patients

Local patients with FU

13 weeks or more

3rd 1970 4th 1970 1st 1971

2nd 1971

12 weeks or less

3rd 1970

4th 1970

1st 1971 2nd 1971

13 weeks or more 3rd 1970

4th 1970 1st 1971 2nd 1971

mortality. The much higher rates for both first and second trimester abortions reported from England and Wales, compared with New York City, also reflect the large number of hysterotomies and hysterectomies in that country (Table 17).

However, New York City had a higher level of mortality associated with hysterotomy compared with England and Wales, at least in part because in New York instillation of saline is the preferred method for the second trimester and hysterotomy is used for patients considered unsuitable for abortion by saline, because of such preexisting conditions as diseases of the cardiovascular system or of the kidneys. In England where hysterotomy is far more widely used than in New York, this adverse selection does not apply. The higher proportion of first trimester abortions among the hysterotomies in England, compared to New York City, may also contribute to the lower level of mortality with that procedure. Conversely, the higher proportion of second trimester abortions by suction may be a contributory factor in the higher mortality associated with this procedure in England compared with New York City.

As shown in Table 24, mortality was six times higher in England and Wales for abortions with sterilization than for abortions without sterilization. However, since hysterectomy is always a sterilizing procedure and hysterotomy is usually combined with sterilization, it is not possible, on the basis of the available data, to dissociate the effect of sterilization from that of major surgery as the method of abortion. It is worth noting, however, that only four deaths occurred in England and Wales among 107,500 first trimester abortions without sterilization, most of which were accomplished by either suction or D & C. The mortality ratio for this group does not differ significantly from the corresponding ratios reported from eastern European countries and,

more recently, from New York City,

Area, years.

and characteristics

Number of legal abortions

Number of deaths

England and Wales (cont.)

dentified by confidential

incones into maternal deaths1o

968

31.300

369

49.800

12

52

Max mum estimate11

968

31,300*

49.800

18

76.000

15

58544

where vaginal evacuation is rarely combined with tubal sterilization. With or without sterilization, mortality associated with legal abortion in England and Wales increased with age; even though a larger proportion of younger than of older women were aborted in the second trimester.

For New York City, the lower mortality of nonresident women compared with residents reflects in part a more favorable distribution by period of gestation, but mainly the fact that the nonresidents were primarily white and middle-class women, whereas the local residents included women from all ethnic and socioeconomic segments of the metropolitan population. Furthermore, women with preexisting medical complications were more likely than healthy women to obtain legal abortions in their home states under more restrictive legislation.

on

Data maternal mortality from complications of pregnancy, childbirth, and the puerperium are shown for 14 countries in Table 25. A comparison of these data with mortality following legal abortion shows clearly that the risk to life associated with first trimester abortion is far smaller than the risk incurred if the pregnancy is permitted to go to term, even in countries where maternal mortality is low. Mortality with abortion in the second trimester is comparable to maternal mortality.

Risks to life associated with abortion can also be compared with such risks associated with contraception. The senior author made this comparison a few years ago by means of a statistical model of 100,000 women of reproductive age in fertile unions and exposed to the risk of pregnancy, assuming the following rates of mortality, based on the experience then available in several developed countries (112):

1. Maternal mortality associated with pregnancy, childbirth, and the puerperium, excluding induced

TABLE 26 Hypothetical annual rates of pregnancies and of deaths associated with contraception, maternity, and legal abortion per 100,000 women of reproductive age in fertile unions

TABLE 27 Deaths attributed to abortion per 100,000 women aged 15-44: Selected areas, 1955-61 and 1965-71

0.4

New York City

1 29 (0.94)' 0.69 (0.34)

'One-year period ending 30 June of year shown in stub. Ref. 88.

abortion: 20 deaths per 100,000 pregnancies.

2. Mortality associated with legal abortion at an early stage of gestation: 3 deaths per 100,000 abortions, based on statistics from eastern Europe (113).

with

3. Mortality associated highly effective contraception: 3 deaths per 100,000 users per year, based on studies in Great Britain on excess mortality from thromboembolic disease attributable to the use of oral contraceptives (58).

In addition, the following assumptions were made in regard to the three components of the interval between two successive conceptions: average duration of pregnancy-nine months for all births and three months for spontaneous and induced abortions; anovulatory phase-four to 14 months after a birth, depending on the extent and average duration of breastfeeding and one month after an abortion; ovulatory phase-eight months in the absence of contraception.

Line 1 of Table 26 illustrates the reproductive behavior of 100,000 women neither using contraception nor having induced abortions. Under the assumptions of the model, this group would have between 40,000 and 60,000 pregnancies in the course of one year. Given the level of maternal mortality, excluding abortion, of 20 per 100,000 pregnancies assumed for the model, these women would experience between 8 and 12 deaths associated with the reproductive process.

Line 2 assumes a noncontracepting population terminating all pregnancies by early legal abortions. Because gestation and the postgestational anovulatory period are both substantially shorter than under the assumptions of the preceding line, the number of pregnancies rises to 100,000 resulting in 3 deaths.

Line 3 assumes a population not using abortion and consistently using a highly effective method of contraception-in this case, oral contra