Large studies of mifepristone plus misoprostol, conducted in varied populations, have shown complete abortion rates of more than 90%, with rare serious side effects.1–4 Providers of medical abortion will be concerned with the possible need for curettage after initiating the mifepristone and misoprostol regimen. The need for curettage arises if patients have an incomplete abortion or a continuing pregnancy. Some patients do not tolerate the side effects of the treatment and request or require surgical termination to end the symptoms. Other patients simply change their mind midcourse and switch from a medical to surgical abortion. Of special concern are cases of acute heavy bleeding that may require emergency curettage, patients who require prolonged follow-up and treatment for retained tissue that causes bleeding for several weeks, and patients with a continuing pregnancy. Both patients and their physicians wish to avoid the possible congenital abnormalities associated with misoprostol; thus, continuing pregnancies must be identified and treated.5–7
We analyzed the frequency, timing, and indications for curettage using data from US clinical trials that used a highly effective regimen: mifepristone 200 mg orally followed by misoprostol 800 μg vaginally. The main objective of this analysis was to characterize the women who underwent curettage and to describe when curettage occurred and the reasons for the intervention. A secondary objective of this analysis was to categorize the indications for curettage according to a simple classification schema.
MATERIALS AND METHODS
These analyses used data from two multisite, randomized controlled trials sponsored by Abortion Rights Mobilization.8–10 Seventeen sites participated, including hospital abortion services, abortion clinics, and private family practice and gynecology offices. All sites had institutional review board approval. Participants were at least 18 years old, desired an abortion, and were healthy. In the first study, women pregnant less than 63 days received 200 mg mifepristone and 800 μg vaginal misoprostol to use 48 hours after the mifepristone. In the second study, women pregnant less than 56 days were randomly assigned to insert vaginal misoprostol at 1, 2, or 3 days after taking mifepristone. The inclusion and exclusion criteria and the method of transvaginal ultrasound dating and monitoring of the abortion followed protocols previously reported.8–10
At the first follow-up visit, 4–8 days after beginning treatment, if the gestational sac was absent by sonogram, the abortion was considered complete. Women with persistent gestational sacs received a second dose of 800 μg misoprostol and returned 1–10 days later, at their discretion, for reassessment. If interval growth had occurred, indicating an ongoing pregnancy, aspiration curettage was performed. If the gestational sac was present without growth, women were permitted to wait up to 1 month for a successful medical abortion. If the gestational sac was still present after 1 month, or if excessive bleeding or other severe symptoms occurred, or if the patient requested surgical termination, curettage was performed. Clinicians noted the date and indication for curettage on the case report forms.
We examined the case report forms and clinical case notes of all women who underwent curettage. We compared the demographic characteristics, obstetric history, and gestational age of the women who underwent curettage with women who had a successful complete abortion. We categorized treatment failures according to three different existing medical abortion failure classification schemas: 1) Winikoff et al11 developed a classification framework that divided failures into user choice, provider choice, and true drug failure, 2) Spitz et al1 classified failures in the Population Council Trials as medically indicated, patient request, ongoing pregnancy, and incomplete abortion, and 3) Schaff and colleagues in the Abortion Rights Mobilization trials grouped failures into six categories—continuing pregnancy, persistent growth sac, incomplete abortion, patient request, other medical indications, and acute heavy bleeding within 24 hours of misoprostol use. Each case was categorized independently by two of the authors (RA, LD) using each of these three schemas. A third evaluator (CW) assessed and resolved each initial disagreement between the first two. This process revealed two main difficulties: 1) a disagreement frequently occurred between the two investigators in assigning many of the cases to these existing classification schemas, and 2) the classification schemas did not agree with each other (data not shown). We then decided to create a composite classification schema that was simple and parsimonious. To create this system, we collapsed the six categories used by Schaff and colleagues into three. After this maneuver, we reevaluated each case and achieved complete agreement in assigning cases to these broader categories. Thus, the classifications presented in the Results section uses the new schema created by the authors.
Data were analyzed using the Scientific Package for Social Sciences (version 9.0, software; SPSS Inc., Chicago, IL), independent t tests, and χ2 tests as appropriate.
Seventeen sites enrolled 4393 women between November 1996 and June 1999, 2091 women under the first study protocol and 2302 under the second, respectively. In total, 59 women (1.3%) were lost to follow-up, 19 from the first protocol, and 40 from the second. Seven sites in the larger cities (Rochester, NY; San Francisco, CA; New York, NY; Cleveland, OH; Cherry Hill, NJ; Seattle, WA; and Atlanta, GA) contributed 85% of the women. The demographic characteristics of those enrolled are listed in Table 1. About two fifths of the women (39.3%) were between the ages of 18 and 25. The mean age was 28 years (range 18–52). Almost three quarters (73.2%) of the women were white. Fewer women had gestational ages between 56 and 63 days because only the first study accepted women in this category. Approximately one half (46.5%) the sample reported at least one prior live birth and one half (50.4%) reported at least one prior elective abortion.
One hundred sixteen women (2.6%, 95% confidence interval [CI] 2.1%, 3.1%) underwent curettage. The difference in the rates of curettage among the study sites was not statistically significant (range 0–8.6%, χ2 = 16.6, df = 16, P ≤ .42). The rate of curettage for the two study protocols was also similar (P ≤ .88). Table 2 shows only those characteristics (gestational age and prior elective abortion) that differed between those who underwent curettage and those who had a complete medical abortion. Age, race, gravidity, and parity were not related to the rate of curettage. Two women were enrolled with occult ectopic pregnancies, discovered when no bleeding response occurred after misoprostol. One was treated successfully with methotrexate, and the other underwent salpingectomy.
We obtained complete case report forms for all the women who underwent curettage in our study. We also reviewed the clinical case notes and contacted providers for more details to classify the indications for curettage. We divided the cases into three main categories: bleeding, continuing pregnancy, and patient request. Sixty-one women (53%, 95% CI 43.9%, 62.1%) underwent curettage for bleeding, including eight with onset of heavy bleeding within 24 hours of misoprostol use and 53 with prolonged or delayed bleeding. A single woman in this group received a blood transfusion. The prolonged or delayed bleeds were considered incomplete abortions by clinical impression, although a tissue examination was not performed in most cases. Seventeen of the women (15%, 95% CI 8.5%, 21.5%) had a continuing pregnancy, five of whom did not take the second dose of misoprostol that was allowed by the protocol.
Thirty-six women (31%, 95% CI 22.6%, 39.4%) underwent curettage for a spectrum of other reasons that we labeled patient request. These included 16 women who, despite counseling about the side effects and duration of treatment, changed their mind and opted for surgical termination before the medical abortion was complete. These women did not experience severe or unexpected side effects and were considered to be experiencing normal courses when the surgical termination was requested. Seven women requested curettage for persistent gestational sacs. These women were not experiencing side effects from the treatment and could have continued waiting for spontaneous expulsion. They chose surgical termination because they were tired of waiting or developed scheduling problems. Thirteen women requested curettage to alleviate unusually severe nausea, vomiting, and pain. These symptoms were more severe than the expected side effects of the treatment.
We were unable to classify two women (1.7%, 95% CI 1.67%, 1.75%) who underwent curettages by non-study physicians. In one case, the curettage was performed when the patient underwent tubal ligation 3 days after completing the medical abortion. We cannot tell whether the curettage was part of this physician's routine practice or for an indication related to the recent abortion. In the other case, a woman underwent a curettage procedure for presumed spontaneous abortion performed by her private gynecologist. This physician was unaware that the patient had recently used mifepristone and misoprostol. Other curettages were performed by non-study physicians; however, in all but these two cases, the indications were clear.
Figure 1 shows the timing of curettage broken down by indication. The distribution of timing was bimodal. One subset of women, 44 (38%, 95% CI 29.1%, 46.8%), underwent curettage in the first study week and another subset, 43 (37%, 95% CI 28.2%, 45.7%), during weeks 3–5. For bleeding, the mean interval between study day 1 and curettage was 25.4 days (range 1–71). Late curettage for bleeding was frequently performed because of a sudden increase in the amount of bleeding that occurred 3–5 weeks after beginning the treatment. Light but persistent prolonged bleeding was another common reason for late curettage. The timing of curettage for continuing pregnancy varied with the gestational age at enrollment. Patients with pregnancies of greater gestational age at enrollment underwent curettage immediately at the follow-up visit because they were not given the option of a second dose of misoprostol and additional observation. Patients with a pregnancy of lesser gestational age and a continuing pregnancy were given the option to wait or take the second dose of misoprostol as allowed by the protocol. Most women who requested curettage because of impatience or symptoms did so in the first week.
This study presents a detailed examination of the indications for curettage after the medical abortion regimen of 200 mg oral mifepristone and 800 μg vaginal misoprostol. This is one of several medical abortion regimens clinicians are likely to use; therefore, it is important to have detailed information on the treatment failures and the use of curettage. Understanding the number and nature of the treatment failures is relevant to implementing medical abortion into clinical practice.
Although gestational age and prior elective abortion were associated with the rate of curettage, the overall curettage rate was so low that these characteristics cannot be viewed as predictors of failure. Ashok and colleagues4 reached similar conclusions in a smaller study of the same regimen. This regimen has higher success rates than the US Food and Drug Administration-approved protocol using 600 mg mifepristone and 400 μg oral misoprostol, but the indications for curettage are similar in both approaches; therefore, our results may be broadly applicable.
We propose classifying the indications for curettage after medical abortion as bleeding, continuing pregnancy, and patient request. With this regimen, we found acute heavy bleeding to be very rare. Most of the bleeding that required curettage was a delayed increase in the amount of bleeding that occurred 3–5 weeks after the initiation of the treatment or light but persistent bleeding for several weeks. The timing of the delayed bleeding suggests a relation to the next menstrual period. Although the gestational sac was confirmed to have been expelled earlier, perhaps the decidual cast was not passed and was associated with a heavier next menstrual period.
Winikoff et al2,11 was concerned about providers intervening unnecessarily in the face of bleeding, but we found marked restraint among study providers when confronted with prolonged or delayed bleeding. According to a substudy of bleeding patterns after using this regimen, daily diaries indicated that many women experienced prolonged bleeding.12 Nonetheless, we documented a low rate of intervention by the study providers that may in part be a result of their experience with mifepristone. It is important that providers counsel patients on the bleeding patterns experienced after initiating this regimen and continue to be available in case curettage is needed several weeks later. Indeed, the US Food and Drug Administration has stated that the ability to provide surgical intervention in cases of incomplete abortion or heavy bleeding is a requirement for mifepristone provision.
Providers may also worry about continuing pregnancy among patients using medical abortion. We found that continuing pregnancy is reassuringly rare with this regimen (0.4%). Of course, a follow-up visit to identify the rare continuing pregnancy is mandatory. Two women with ectopic pregnancies were enrolled in these trials who should have been excluded. One would expect more ectopic pregnancies to occur among 4393 pregnancies. This indicates that study providers were able to successfully identify and exclude most women with early ectopic pregnancies by routinely using vaginal ultrasound scanning.
The patient request indication for curettage may indicate a need for better patient counseling and patient selection. Because one cannot predict who will experience unacceptable symptoms, all patients must be adequately counseled about what to expect. Some of the severe nausea and vomiting may have been a result of pre-existing pregnancy symptoms that were worsened by mifepristone and misoprostol. Aggressive treatment of concomitant symptoms, including pain, nausea, and vomiting, may also reduce the need for curettage.
We reviewed the published literature to determine the distribution of indications for curettage found by others in mifepristone trials. Using the GratefulMed search engine to search the MEDLINE database with the following MeSH search terms and connectors, “abortion, induced AND mifepristone,” we limited our search to first-trimester clinical trials in the past decade (1990–2000) with 800 or more participants to obtain sufficient numbers of failures to examine. Other limits included the English language and human participants. After text review, we found nine trials that met our criteria.
A failure in medical abortion is defined as the need for surgical intervention for any reason.11 Although failure rates ranged from 2% to 13%, all nine studies reported the proportion of curettages performed for incomplete abortions or bleeding to be about one half.1,2,4,13–18 This was true regardless of the dose, route, or type of prostaglandin used and agrees with our findings. Eight studies also reported continuing pregnancy rates. Rates similar to the Abortion Rights Mobilization trials (less than 1%) were found in studies using vaginal misoprostol and gemeprost as the prostaglandin.4,14,16,18 Studies using oral misoprostol reported greater rates of continuing pregnancy, particularly at later gestational ages.1,13,15 The reporting of other indications for curettage such as patient request and severe symptoms was incomplete and inconsistent among the studies. Only two studies described the proportion of curettages due to patient request and these could not be easily compared with the present results.1,2 The safety, efficacy, and acceptability of mifepristone abortion have been recently reviewed in detail elsewhere.19,20
Our study has several limitations. We do not have any record of whether the curettages occurred during regular office hours or were unscheduled. We realize, however, that clinicians would be most interested in the frequency of emergency procedures. Some women with continuing pregnancies received a second dose of misoprostol as allowed by the protocol. Because of the incomplete use of this treatment option, however, we were unable to assess whether a second dose is useful. In addition, 59 women were lost to follow-up in these trials. Some of these women may have undergone curettages outside the study. However, even if every one of these women had undergone a curettage procedure, the total curettage rate would still be only 4%. We think, however, that unrecorded curettage outside the study was rare because the study physicians were readily available to perform curettages at no additional cost to the woman. Finally, our results may not be generalizable to clinical practice. These randomized controlled trials followed strict protocols and were conducted by experienced investigators with a selected study population. The routine use of this medical abortion regimen in broader practice settings may result in higher curettage rates with more varied indications than are reported here.
1. Spitz I, Bardin W, Benton L, Robbins A. Early pregnancy termination with mifepristone in the United States. N Engl J Med 1998;338:1241–6.
2. Winikoff B, Sivin I, Coyaji K, Cabezas E, Bilian X, Sujuan G, et al. Safety, efficacy, and acceptability of medical abortion in China, Cuba, and India: A comparative trial of mifepristone-misoprostol versus surgical abortion. Am J Obstet Gynecol 1997;176:431–7.
3. Peyron R, Aubeny E, Targosz V, Silvestre L, Renault M, Elkik F, et al. Early termination of pregnancy with mifepristone (RU 486) and the orally active prostaglandin misoprostol. N Engl J Med 1993;328:1509–13.
4. Ashok PW, Penney GC, Flett GMM, Templeton A. An effective regimen for early medical abortion: A report of 2000 consecutive cases. Hum Reprod 1998;13:2962–5.
5. Gonzalez CH, Vargas FR, Perez AB, Kim CA, Brunoni D, Marques-Dias MJ, et al. Limb deficiency with or without Möbius' sequence in seven Brazilian children associated with misoprostol use in the first trimester of pregnancy. Am J Med Genet 1993;47:59–64.
6. Gonzalez CH, Marques-Dias MJ, Kim CA, Sugayama SM, Da Paz JA, Huson SM, et al. Congenital abnormalities in Brazilian children associated with misoprostol misuse in first trimester of pregnancy. Lancet 1998;351:1624–7.
7. Pastuszak AL, Schüler L, Speck-Martins CE, Coelho KE, Cordello SM, Vargas F, et al. Use of misoprostol during pregnancy and Möbius' syndrome in infants. N Engl J Med 1998;338:1881–5.
8. Schaff EA, Eisinger SH, Stadalius LS, Franks P, Gore BZ, Poppema S. Low-dose mifepristone 200 mg and vaginal misoprostol for abortion. Contraception 1999;59:1–6.
9. Schaff EA, Fielding SL, Eisinger SH, Stadalius LS, Fuller L. Low-dose mifepristone followed by vaginal misoprostol at 48 hours for abortion up to 63 days. Contraception 2000; 61:41–6.
10. Schaff EA, Fielding SL, Westhoff C, Ellertson C, Eisinger SH, Stadalius LS, et al. Randomized trial of vaginal misoprostol administered 1, 2, or 3 days after mifepristone for early medical abortion (≤ 56 days gestation): a randomized trial. JAMA 2000;284:1948–53.
11. Winikoff B, Ellertson C, Clark S. Analysis of failure in medical abortion. Contraception 1996;54:323–7.
12. Davis A, Westhoff C, De Nonno L. Bleeding patterns after early abortion with mifepristone and misoprostol or manual vacuum aspiration. J Am Med Wom Assoc 2000;55:141–4.
13. World Health Organization Task Force on Post-ovulatory Methods of Fertility Regulation. Comparison of two doses of mifepristone in combination with misoprostol for early medical abortion: a randomized trial. Br J Obstet Gynaecol 2000;107:524–30.
14. Urquhart DR, Templeton AA, Shinewi F, Chapman M, Hawkins K, McGarry J, et al. The efficacy and tolerance of mifepristone and prostaglandin in termination of pregnancy less than 63 days' gestation: UK Multicentre Study—Final results. Contraception 1997;55:1–5.
15. Aubeny E, Peyron R, Turpin CL, Renault M, Targosz V, Silvestre L, et al. Termination of early pregnancy (up to 63 days of amenorrhea) with mifepristone and increasing doses of misoprostol. Int J Fertil Menopausal Stud 1995; 40(suppl 2):85–91.
16. World Health Organization Task Force on Post-ovulatory Methods of Fertility Regulation. Termination of pregnancy with reduced doses of mifepristone. BMJ 1993;307:532–7.
17. Wu S, Gao J, Wu Y, Wu M, Fan H, Yao G, et al. Clinical trial on termination of early pregnancy with RU 486 in combination with prostaglandin. Contraception 1992;46:203–10.
18. Ulmann A, Silvestre L, Chemama L, Rezvani Y, Renault M, Aguillaume CJ, et al. Medical termination of early pregnancy with mifepristone (RU 486) followed by a prostaglandin analogue. Acta Obstet Gynecol Scand 1992; 71:278–83.
19. Christin-Maitre S, Bouchard P, Spitz IM. Medical termination of pregnancy. N Engl J Med 2000;342:946–56.
20. Kahn JG, Becker BJ, MacIsaa L, Amory JK, Neuhaus J, Olkin I, et al. The efficacy of medical abortion: A meta-analysis. Contraception 2000;61:29–40.