Legal induced abortion is one of the most frequently performed surgical procedures in the United States. With approximately 1.2 million legal induced abortions performed in 1997,1 minimizing risk for women who choose to terminate their pregnancies is of clear public health importance.
Pregnancy-related deaths are deaths that occur among women within 1 year of pregnancy from complications of the pregnancy or delivery; deaths associated with complications of induced abortion2 (ie, abortion-related deaths) also are considered pregnancy related. Previous reports on abortion-related mortality for 1972–1987 have informed abortion policy and practice and improved safety for women. In addition, data on the lower risk of death with certain procedures and anesthetics have guided practice, substantially reducing the number of abortions conducted with methods found to be associated with increased risk.3–8 However, the medical practice and provision of abortion services continues to change. For example, since the mid-1990s, medical (ie, nonsurgical) regimens using abortifacients within the first 7 weeks of pregnancy have been used to terminate pregnancies.9 This report provides information on risk factors for abortion-related deaths among women who had abortions in recent years that will help inform and update policymakers and practitioners about abortion-related maternal mortality.
MATERIALS AND METHODS
Data for these analyses were derived from 2 data sets from the Centers for Disease Control and Prevention (CDC). Numerator data were obtained from the Abortion Mortality Surveillance System, now a part of the Pregnancy Mortality Surveillance System, which attempts to identify all deaths in the United States caused by pregnancy, including those ending in induced abortion. For abortion mortality rate denominators, we used data from CDC's Induced Abortion Surveillance System, compiled since 1969. From 1973 through 1997, data were received from state health departments or estimated for 52 reporting areas, including 50 states, the District of Columbia, and New York City. Legal induced abortion was defined as “a procedure, performed by a licensed physician or someone acting under the supervision of a licensed physician, that was intended to terminate a suspected or known intrauterine pregnancy and to produce a nonviable fetus at any gestational age.” The total number of legal induced abortions was available or estimated from all reporting areas; however, not all of these areas collected information regarding some or all of the characteristics of women who obtained abortions.1
The Abortion Mortality Surveillance System defines an abortion-related death is a death resulting from 1) a direct complication of an abortion, 2) an indirect complication caused by the chain of events initiated by the abortion, or 3) an aggravation of a preexisting condition by the physiologic or psychologic effects of the abortion, regardless of the amount of time between the abortion and the death.10 The inclusion of abortion-related deaths in this surveillance system, regardless of the amount of time between the abortion procedure and the death, is unique and differs from the temporal limit for other pregnancy outcomes in the Pregnancy Mortality Surveillance System. Legal induced abortion–related mortality rate is defined as the number of deaths from legal induced abortion per 100,000 legal induced abortions.
Multiple sources are used in the Abortion Mortality Surveillance System to identify potential cases of abortion-related mortality, including national and state vital records, reports from maternal mortality review committees, private citizens, health care providers, medical examiners, the media, and, more recently, a full-text newspaper database. For each suspected case identified, the Abortion Mortality Surveillance System requests death certificates, clinical records, and autopsy reports. Death certificates were obtained for all cases, but complete clinical records were not always available. Two medical epidemiologists reviewed the available records for each case to determine the cause of death and if it was abortion-related.
Gestational age was defined as the number of completed weeks elapsed from the start of the last menstrual period and was categorized as either 1) 8 weeks or less, 9–10 weeks, 11–12 weeks, 13–15 weeks, 16–20 weeks, and 21 or more weeks or 2) first (12 weeks or less) or second trimester (13 weeks or more). Parity was defined as the number of previous live births and was categorized as 0, 1–2, and 3 or more. When calculating mortality rates specific to parity, gestational age, and marital status, we excluded cases for which the decedent's parity, gestational age, or marital status were unknown, unless specifically noted. Procedures were categorized as curettage, dilatation and evacuation (D&E), instillation, or other. Curettage includes suction or sharp curettage performed at or before 12 weeks of pregnancy. For cases in which the procedure was curettage but the gestational age was unknown, we assumed the procedure occurred at or before 12 weeks of gestation for those analyses that were stratified by trimester of gestation. For those analyses that were performed by weeks of gestation, cases with unknown gestational age were reported separately as unknown gestational age or were excluded. Similarly, when the procedure was unknown and gestational age was recorded as 12 weeks or less, we assumed that curettage was performed. D&E is a combination of suction and sharp curettage performed through a dilated cervix at or after 13 weeks; instillation involves prostaglandin or saline instillation; and “other” associated procedures include hysterectomy, hysterotomy, and use of prostaglandin vaginal suppositories. For the time period of this analysis (1988–1997), approximately 0.10% of legal induced abortions were performed with abortifacients in early pregnancy.11 No deaths associated with them were identified by the Abortion Mortality Surveillance System during the study period.
Causes of abortion-related deaths included direct causes (eg, vaginal and intraabdominal hemorrhage), infection (including endometritis, septicemia, and other infections), emboli (including thrombotic, amniotic fluid, and air emboli), complications of anesthesia, and indirect causes (categorized as “other”), mainly cardiac, and cerebral vascular events. Women were divided into 2 racial categories: 1) white and 2) black or other. Women who were of black or other races (eg, Asian/Pacific Islander, American Indian) were combined into 1 category because of the difficulty in separating races in the denominator before 1990 and because only 2 cases were reported for a nonwhite, nonblack woman during 1988–1997.
The crude (unadjusted) legal induced abortion–related mortality rates were calculated for each year from 1972 through 1997. In addition to calculating the crude mortality rate, we stratified the unadjusted mortality rates by various sociodemographic and medical factors, including the type of procedure; woman's race, age, and parity; and gestational age of the pregnancy that was terminated during 1988–1997, the 10 most recent years of data available from the Abortion Mortality Surveillance System. For all rates, the relative risks (RRs) with 95% confidence intervals (CIs) were calculated by using the Taylor series method in Epi-Info 6.04c.12
To understand the effect of differences in gestational age distribution on the RR of death for women of different ages and race, we calculated gestational age–adjusted, race-specific, and maternal age–specific mortality rates. For the race-specific analyses, we directly standardized the mortality rates to the gestational age distributions of white women and for the maternal age–specific rates, we used the gestational age distribution of older women as the standard. In these standardized analyses, deaths for which the gestational age at the time of abortion was unknown were assigned a gestational age in proportion to the gestational age distribution of the deceased women where the gestational age was known. To determine whether the shift toward earlier gestation abortions was primarily responsible for the decrease in abortion mortality over time, we calculated and compared gestational age–specific mortality rates over 3 time periods from 1972 through 1997. Because the risk of death with increasing gestational age does not follow a linear distribution, we fit exponential models to assess the relationship between mortality and increasing gestational age.
The project resulting in this manuscript was reviewed for human subjects issues and determined to be in compliance with CDC's guidelines. The analyses used data from the Pregnancy Mortality Surveillance System and Legal Induced Abortion Surveillance System, both housed in the Division of Reproductive Health at CDC.
During 1972–1997, a total of 337 deaths determined to be causally related to legal induced abortions was identified by the Abortion Mortality Surveillance System for an overall legal induced abortion–related mortality rate of 1.1 deaths per 100,000 legal induced abortions (Table 1). From 1972 through 1997, the annual number of legal induced abortion–related deaths decreased from 24 to 7, and the mortality rate decreased from 4.1 to 0.6. Most of the decline occurred early in this time period, from 1972 through 1976; after the legalization of abortion in January of 1973, the mortality rate fell from 4.1 to 1.1 deaths per 100,000 abortions, a reduction of 73% (P = .001). Women in the earlier time period (1972–1979) were 3 times (RR 3.1; 95% CI 2.4, 4.1) more likely to die of complications of an abortion than women in the most recent time period (1988–1997) (Table 2 and Figure 1).
We also calculated the gestational age–specific relative risks of dying comparing the earliest (1972–1979) and most recent (1988–1997) time periods using the most recent time period as the referent group. Although the risk of death declined at all gestational ages, the greatest proportion of the decline occurred at earlier gestational ages. Women who had abortions performed in the earlier time period were significantly more likely to die at each gestational age than women who had abortions in the most recent time period; women receiving abortions during 1972–1979 had RRs of 5 (at or before 8 weeks of gestation), 8.6 (at 9–10 weeks), 6.2 (at 13–15 weeks), and 4.1 (at 16–20 weeks), and 1.9 (at or after 21 weeks). These declines are all statistically significant, with the exception of the women who died of complications of abortion at 21 weeks or more of gestation; although their mortality decreased almost 50%, the decrease was not statistically significant. To examine risk factors among women receiving abortions in the most recent time period, we analyzed deaths that occurred during 1988–1997. Gestational age at the time of abortion was the strongest risk factor for abortion-related mortality (Table 2). The lowest rates were among women who had their abortions in the first trimester of pregnancy, particularly within the first 8 weeks of pregnancy. Women whose abortions were performed in the second trimester (at or after 13 weeks of gestation) had abortion-related mortality rates greater than women whose abortions were performed in the first 8 weeks of pregnancy (RR at 13–15 weeks, 14.7 [95% CI 6.2, 34.7]; RR at 16–20 weeks, 29.5 [95% CI 12.9, 67.4]; RR at or after 21 weeks, 76.6 [95% CI 32.5, 180.8]). If women who had abortions after 8 weeks of gestation had obtained abortions during the first 8 weeks of pregnancy, when risk is lowest, 87% of deaths likely could have been prevented.
In addition, we used the data to model the association between the mortality rate and gestational age (Figure 1). We found that for the most recent time period (1988–1997), the risk of death increased exponentially with increasing gestational age. According to this model, there is a 38% increase in risk of death for each additional week of gestation. This implies that the increase in the risk of death due to delaying the procedure by 1 week is much higher at later gestational ages than at earlier gestational ages. For example, applying this model, if an abortion is performed at 9 weeks rather than at 8 weeks of gestation, the estimated absolute increase in the mortality rate is 0.05 per 100,000 abortions (from 0.13 to 0.18 deaths per 100,000 abortions). However, if an abortion is performed at 18 weeks of gestation instead of at 17 weeks, the estimated absolute increase is 0.91 (from 2.4 to 3.3 per 100,000 abortions). Thus, the estimated increase in the risk of death due to delaying the procedure by 1 week at 17 weeks of gestation is 18 times greater than the estimated increase in the risk of death by delaying the procedure by 1 week at 8 weeks of gestation.
The second most significant risk factor for death overall was race. Women of black and other races were 2.4 times as likely as white women to die of complications of abortion (Table 2). At all gestational ages, women of black and other races had higher case mortality rates than white women. Because women of black and other races tend to have abortions at later gestational ages,1,11 we standardized the mortality rates for black women to the gestational age distribution of white women to assess the effect that gestational age may have had on the higher risk of death for women of black and other races. The ratio of the adjusted mortality rates for women of black and other races compared with white women decreased 20% to 1.9. However, this adjusted rate still differs significantly from the rate for white women. No statistically significant differences were observed between crude mortality rates for women of different age or parity. However, data from the Abortion Surveillance System indicate that women younger than 20 years of age had abortions later in gestation than did women aged 20–29 years, and women aged 30 years or older obtained abortions earlier in pregnancy than women in any other age group.1,11 To determine the impact of these differences on age-specific mortality, we standardized the maternal age–specific mortality rates for gestational age using the gestational age distribution of women aged 30 years or older as the standard. If women younger than 20 years of age who terminated their pregnancies had the same gestational age distribution as women aged 30 years or older, mortality among women younger than 20 years of age would decrease by 32%, and mortality among women aged 20–29 years would decrease by 17%.
The procedures that can be used to terminate a pregnancy are determined by the gestational age at the time of the procedure. For the years 1988–1997, more than 99% of abortions in the first trimester were performed by curettage. Therefore, we examined the relationship between abortion procedure and mortality in the second trimester. For women in the second trimester, the mortality rates for D&E were 2.5 times lower than those for instillation and other procedures. These differences were not significant; however, our analysis was limited by very small numbers in some categories and the large number of women who could not be included in this analysis because of unknown procedure or unknown gestational age. No deaths associated with early medical abortion procedures using abortifacients were reported during the study period.
Of abortion-related deaths, 85% were attributable to direct causes and 15% to indirect (ie, “other”) causes. Of the direct causes, hemorrhage and infection exceeded any other cause. Overall, each were responsible for approximately one fourth of abortion-related deaths, whereas embolism, anesthetic complications, and other causes were each responsible for about 15% of deaths (Table 3). Cause of death varied by gestational age and procedure type. For example, hemorrhage, a less frequent cause of death at or before 12 weeks of pregnancy, was the most frequent cause of death associated with D&E at 13 weeks or more of gestation.
Among women for whom the interval between the abortion procedure and death was known, 35% of the deaths occurred within 24 hours, and 85% died within 42 days of the procedure, the length of time considered the puerperal period.
In the 25 years following the legalization of abortion in 1973 (Roe v. Wade, 410 U.S. 113, 1973), the risk of death from legal abortion declined dramatically by 85%, from 4.1 to 0.6, with most of this decline occurring from 1973 through 1976. The number of illegal abortion–related deaths (induced abortions not performed by a licensed physician or a supervised assistant) also declined after legalization of abortion—only 5 deaths associated with illegal abortion were identified during 1988–1997.1 The initial decrease in legal abortion–related deaths can be largely attributed to an increase in the level of experience and skill of the providers,7,13 a factor that has reduced the risk of complications with other procedures.14 Further reductions in the number of deaths and risk of mortality can be attributed to changes in clinical practice—changes made in response to reports that identified procedures with an increased risk of complications. For example, in 1972, approximately 10% of abortions were performed by either saline or prostaglandin instillation procedures. Use of this higher-risk procedure declined through the 1970s to approximately 3% in 1980 and, concurrently, the proportion of providers using dilation and curettage (a procedure associated with lower risk of complications) increased. The heightened risk of death with the use of general anesthetics, in particular fast-acting barbiturates, was also identified in the 1980s; few abortions currently are performed using these substances.6 As the strong association between gestational age and the risk of complications became more widely known, an increased percentage of abortions were performed early in the first trimester; 34% of abortions were performed before 8 weeks of gestation in 1972 compared with almost 55% in 1997.1
The risk factor that continues to be most strongly associated with mortality from legal abortion is gestational age at the time of the abortion. The relationship between gestational age and risk of death has changed over time; currently, the risk of death increases exponentially at all gestational ages, whereas for women obtaining abortions in the earlier time period (1970–1979), the risk of death increased with increasing gestational age but leveled off at the highest gestational ages. The change in models for risk of death by gestational age likely results from the reduction in risk at earlier gestational ages as abortion policy and practice have changed; the risk of death at later gestational ages may be less amenable to reduction because of the inherently greater technical complexity of later abortions related to the anatomical and physiologic changes that occur as pregnancy advances. The increased amount of fetal and placental tissue requires a greater degree of cervical dilation, the increased blood flow predisposes to hemorrhage, and the relaxed myometrium is more subject to mechanical perforation. The technical challenges of the procedure during the second trimester are different from those present in the first trimester, and the inherently greater risk of complications may be less amenable to prevention. However, it is possible that other factors such as exacerbation of a preexisting disease may have also contributed to the greater risk of death for women obtaining abortions at later gestational age, but our ability to determine the potential contribution of other factors is limited because of limited information about the deceased women's medical or social history.
Almost half of abortions still occur after 8 weeks of gestation. Because access to abortions even 1 week earlier reduces the risk of death disproportionately as gestational age increases, addressing this risk factor by further reducing the gestational age at which women have abortions may help to further reduce the risk of death.
Our analysis suggests that almost one fifth of the excess abortion-related mortality among women of black and other races resulted from later gestational age at the time of the abortion. In addition, more than one third of the abortion-related mortality risk for women aged 19 years or younger was due to having an abortion at a later gestational age as compared with women aged 30 years or older.
Because gestational age at the time of abortion is such a strong risk factor for death, factors that can affect access to abortion services deserve examination. First, availability of services influences access to early abortion. Since 1982, the number of abortion providers has decreased by 20%; most of the decline has occurred among hospital-based providers and in nonmetropolitan areas, leading to decreased appointment availability and an increased average distance that women must travel to abortion facilities.15–17 In addition, many abortion facilities set a gestational age limit after which they will not perform abortions. Consequently, women seeking abortion services after the first trimester may have to travel longer distances, which may lead to even greater delay in obtaining services. Other factors that may also lead to abortions at later gestational ages include failure to recognize a pregnancy or miscalculation of the length of pregnancy; reluctance to tell a partner or parents about a pregnancy; time needed to decide how to resolve the pregnancy; and difficulty in finding a provider, making arrangements for the abortion, obtaining transportation, and being able to afford the procedure.18–20 In 2001, a total of 33 states required either parental notification or consent or a mandatory waiting period after a woman's initial visit to the abortion provider before the procedure could be performed.15,19 Both parental notification laws and mandatory waiting periods have been associated with an increase in second-trimester abortions.21,22 In 1998, only 16 states had Medicaid or other state-supported funding of abortions; thus women in most states must spend time seeking financial resources to pay for an abortion.15
Since the mid-1990s, methotrexate with misoprostol and more recently mifepristone have been used for nonsurgical termination of early pregnancies (ie, those up to 7 weeks of gestation).23 Mifepristone (commonly called RU-486) is approved for such use in most of Europe24 and has been used for more than a decade in France,24 Sweden, and Great Britain.25,26 Before the U.S. Food and Drug Administration approved the drug for use as a medical abortifacient in 2000, it was used in clinical trials in the United States.9 The CDC's Abortion Surveillance System began to collect data on medical terminations in 1997. In 1999, a total of 25 states reported that 6,278 of these early medical abortions using RU-486 had been performed, which likely is an underestimate.27 An early medical abortion requires more visits by the woman to her health care provider than are required for a surgical procedure, but acceptability among both providers and patients is reported as being high.28,29 No deaths determined to be related to use of medical abortifacients were reported in the United States during the study period.
The number or rate of abortions in European countries where mifepristone is used as an abortifacient has not increased, although the proportion of abortions performed at earlier gestational ages has risen.25 If the number of abortions remains constant in the United States, increased availability of mifepristone to U.S. women who choose to terminate their pregnancies may increase the proportion of abortions at earlier gestational ages and in turn decrease the risk of abortion-related mortality. Ongoing monitoring of both abortion procedures and abortion-related mortality will help to evaluate the effect of medical abortion regimens.
The United States continues to monitor the number of abortion procedures and abortion-related deaths nationally. Furthermore, CDC's Abortion Mortality Surveillance System uses multiple methods to identify cases of abortion-related mortality, thereby increasing the identification of potential deaths. Cases are confirmed through review of available hospital charts and coroners’ reports by clinically experienced epidemiologists. On average, the Abortion Mortality Surveillance System reports more than twice as many deaths related to legal induced abortion than are reported on routine death-certificate data. The completeness of death reporting is difficult to determine; however, an assessment that used multiple methods indicated that both reported numbers and rates of abortion-related deaths was consistent among multiple sources.30 Surveillance of abortion-related mortality continues to be essential in monitoring trends, evaluating risk factors, and identifying potential clusters of deaths.
Our analyses have several possible limitations. Although state health departments are asked to provide death certificates on all deaths associated with pregnancy and other sources are used to try to ascertain abortion-related deaths, some cases may not be identified. In addition, we were unable to obtain detailed clinical records for all cases, and therefore data on certain factors (eg, gestational age, type of abortion procedure, and other risk factors for death, such as preexisting diseases), were not available for all deaths. In addition, because of the data sources used for this study, we are unable to determine why some women obtain abortions later in their pregnancies. Some of these women may choose to terminate their pregnancies because of a preexisting medical condition or fetal indications (eg, severe fetal anomalies). Thus, our ability to understand all the barriers to early abortion is incomplete. Although determination of the cause of death and relatedness to the abortion procedure is a straightforward process, some misclassification may have occurred. Timeliness in reporting abortion-related deaths is affected by several factors, including delays of up to several years in death notification, difficulty in obtaining clinical information from providers and facilities, and the need to compile multiple years of data before release because of the small number of cases that occur annually and the need to maintain anonymity. In some stratified analyses, abortion-related mortality rates for the different strata may be underestimated, because cases with unknown values for the characteristic of interest could not be included. The aggregate nature of CDC's Abortion Surveillance System also served as a study limitation by preventing multivariable analyses of abortion mortality. Denominator data on abortion procedures is reported univariately, with a subset of states providing bivariate data. Thus, examining the affects of one risk factor while controlling for all other potential risk factors was not feasible.
Legal induced abortion–related deaths occur only rarely. Substantial reduction in the number and risk of deaths caused by complications of abortion can be affected by identification of risk factors for death and use of this evidence to inform policy and practice changes. Currently, gestational age at the time of the abortion is the strongest risk factor for death. If women who terminated their pregnancies after 8 weeks of gestation had accessed abortion services during the first 8 weeks of gestation, up to 87% of deaths might have been avoided. Reasons for delay in accessing services are likely multifactorial; to help guide prevention efforts to reduce mortality from complications of abortion, additional information is needed about the women who access abortion services later during pregnancy and the reasoning behind this decision. Primary prevention of unintended pregnancies is optimal. However, among women who choose to terminate their pregnancies, increased access to early abortion services (including emerging technologies such as early medical abortion regimens) may increase the proportion of abortions performed at the lower-risk, early gestational ages and help reduce maternal deaths.
1. Koonin LM, Strauss LT, Chrisman CE, Parker WY. Abortion surveillance—United States, 1997. In: CDC Surveillance Summaries, December 8, 2000. MMWR Morb Mortal Wkly Rep 2000;49:1–43.
2. Berg CJ, Chang J, Callaghan W, Whitehead S. Pregnancy-related mortality in the United States, 1991–1997. Obstet Gynecol 2003;101:289–96.
3. Cates WJr, Rochat R, Smith J, Tyler CWJr. Trends in national abortion mortality, United States, 1940–1974: implications for prevention of future abortion deaths. Adv Plan Parent 1976;11:106–13.
4. Centers for Disease Control and Prevention. Abortion surveillance 1976. Atlanta (GA): U.S. Department of Health and Human Services, Public Health Service, CDC; 1978.
5. Atrash H, MacKay T, Binkin N, Hogue C. Legal abortion mortality in the United States: 1972 to 1982. Am J Obstet Gynecol 1986;156:605–12.
6. Atrash H, Cheek T, Hogue C. Legal abortion mortality and general anesthesia. Am J Obstet Gynecol 1988;158:420–4.
7. Council on Scientific Affairs, American Medical Association. Induced termination of pregnancy before and after Roe v Wade. JAMA 1992;268:323–9.
8. Lawson H, Frye A, Atrash H, Smith J, Shulman H, Ramick M. Abortion mortality, United States, 1972 through 1987. Am J Obstet Gynecol 1994;171:1365–72.
9. American College of Obstetricians and Gynecologists. Medical management of abortion. ACOG Practice Bulletin 26; 2001.
10. Centers for Disease Control and Prevention. Abortion surveillance, 1978. Atlanta (GA): U.S. Department of Health, Education and Welfare, Public Health Service, CDC; 1980. p. 6–13.
11. Herndon J, Strauss LT, Whitehead S, Parker WY, Bartlett L, Zane S. Abortion surveillance—United States, 1998. In: CDC Surveillance Summaries, June 7, 2002. MMWR Morb Mortal Wkly Rep 2002;51:1–32.
12. Centers for Disease Control and Prevention. Epi Info version 6: a word processing, data base and statistics system for epidemiology on microcomputers. Available at: http://www.cdc.gov/epiinfo/
. Retrieved June 18, 2003.
13. Cates W, Grimes Schulz K. The public health impact of legal abortion: 30 years later. Perspect Sex Reprod Health 2003;35:25–8.
14. Hannan E, Siu A, Kumar D, Kilburn H, Chassin M. The decline in coronary artery bypass graft surgery mortality in New York State: the role of surgeon volume. JAMA 1995;273:209–13.
15. Henshaw SK. Factors hindering access to abortion services. Fam Plann Perspect 1995;27:54–9, 87.
16. Henshaw SK, Van Vort J. Abortion services in the United States, 1987 and 1988. Fam Plann Perspect 1990;22:102–9.
17. Henshaw SK, Van Vort J. Abortion services in the United States, 1991 and 1992. Fam Plann Perspect 1994;26:100–6.
19. Torres A. Forrest JD. Why do women have abortions? Fam Plann Perspect 1988;20:169–76.
20. Henshaw SK. Accessibility of abortion services in the United States. Fam Plann Perspect 1991;23:246–53.
21. Joyce T, Kaestner R. The impact of Mississippi's mandatory delay law on the timing of abortion. Fam Plann Perspect 2000;32:4–13.
22. Henshaw S. The impact of requirements for parental consent on minors’ abortions in Mississippi. Fam Plann Perspect 1995;27:120–2.
23. Creinin M, Edwards J. Early abortion: surgical and medical options. Curr Probl Obstet Gynecol Fertil 1997;20:1–32.
24. Silvestre L, Dubois C, Renault M, Rezvani Y, Baulieu E-E, Ulmann A. Voluntary interruption of pregnancy with mifepristone (RU 486) and a prostaglandin analogue: a large scale French experience. N Engl J Med 1990;322:645–8.
25. Jones RK, Henshaw SK. Mifepristone for early medical abortion: experiences in France, Great Britain and Sweden. Perspect Sex Reprod 2002;34:154–61.
26. UK Multicentre Trial. The efficacy and tolerance of mifepristone and prostaglandin in first-trimester termination of pregnancy. Br J Obstet Gyneaecol 1990;97:480–6.
27. Koonin LM, Strauss LT, Chrisman CE, Montalbano MA, Bartlett LA, Smith JC. Abortion Surveillance United States, 1996. In: CDC Surveillance Summaries, July 30, 1999. MMWR Morb Mortal Wkly Rep 1999;48:1–42.
28. Creinin MD, Burke AE. Methotrexate and misoprostol for early abortion: a multicenter trial—acceptability. Contraception 1996;54:19–22.
29. Winikoff B, Ellerston C, Elul B, Sivin I. Acceptability and feasibility of early pregnancy termination by mifepristone-misoprostol: results of large multicenter trial in the United States. Arch Family Med 1998;7:360–6.
30. Cates W, Smith J, Rochat R, Grimes D. Mortality from abortion and childbirth: are the statistics biased? JAMA 1982;248:192–6.