Materials and Methods
All healthy English- and Spanish-speaking women of 7–14 weeks' gestation requesting pregnancy termination at San Francisco General Hospital were invited to participate in the study from March 1997 to August 1997. The trial was approved by the Committee on Human Research, University of California, San Francisco. Inclusion criteria were age 16 years or older, a viable intrauterine pregnancy of 7–14 weeks' gestation confirmed by sonography, ability to give informed consent, and willingness to proceed with standard treatment if adequate dilation with the study drug failed to occur. Exclusion criteria included known intolerance or allergy to misoprostol and severe asthma or hypertension, defined as daily use of medication for either condition.
The subjects were assigned randomly to one of three groups: one medium laminaria, misoprostol 400 μg (two tablets) per vagina, or misoprostol 400 μg (two tablets) orally. Analgesia was not used for laminaria placement. Routine medical history, physical examination, and cervical cultures were done. Gestational age and viability were confirmed with transvaginal or transabdominal sonography. We used ultrasound to determine gestational age if menstrual dates and ultrasound determination differed by 7 days or more in the 7–12-week gestations or by 10 days or more in the 12–14-week gestations.
Subjects were exposed to the dilating agent for an average of 4 hours while being observed in the clinic. Each responded to a short, structured interview immediately after placement of the dilating agent, and another interview immediately before being called in for abortion. The first interview assessed side effects at placement, and the second interview assessed side effects during exposure to the dilating agent and overall acceptability. Additional complaints or side effects that were not part of the interview questionnaire, as well as requests for medication for pain, nausea, or diarrhea, were recorded by the research nurse.
Immediately before the abortion, one of the investigators removed any remaining dilating agent, after which the surgeon entered the room to perform the abortion. Cervical dilation was measured with Pratt dilators, using sequentially smaller ones until a dilator entered the internal os easily without pressure. The suction cannula chosen was the same diameter (mm) as the gestational age in weeks. If the suction cannula slid easily into the uterus, the cervix was considered adequately dilated and the abortion was completed without extra dilation. If further dilation was needed, it was achieved with successively larger Pratt dilators. The surgeon subjectively rated this dilation as easy, moderately difficult, or difficult, and completed the abortion. A research assistant recorded the duration of the procedure and milliliters of fluid collected.
The study was placebo-controlled and blinded to the degree possible, given the different natures of laminaria and misoprostol. Investigators administered the cervical dilating agent (or placebo vaginal tablets) during the routine preoperative examination. In all three groups, a gauze sponge was placed in the vagina after the dilating agent. Subjects in the laminaria and oral misoprostol groups had white placebo tablets placed in the vagina. Subjects in the laminaria and vaginal misoprostol groups received oral placebo tablets. Thus, the surgeon performing the abortion and measuring the outcomes was blinded to the treatment group; subjects in all three groups had dissolved white powder in the vagina, had swallowed white pills, and had the dilating agent and sponge removed by an investigator before the surgeon entered the room. The research nurse administering the questionnaires and monitoring side effects was also blinded to the treatment groups. In addition, the subjects were blinded to the treatment they received. Although a placebo for laminaria (or the tenaculum) was not used, all subjects had a speculum examination for placement of pills or laminaria, followed by a damp sponge. Therefore, single blinding on outcome measurements was complete: Both the surgeon measuring cervical dilation and the research nurse administering interviews on side effects were blinded to the treatment groups. Subject blinding between the laminaria and the misoprostol arms was compromised by the nature of administering two different types of dilation methods, but double blinding was complete in the two misoprostol arms. Blinding was unmasked before data analysis.
The outcomes included cervical dilation in millimeters, measured by Pratt dilators, proportion of subjects who required additional dilation, difficulty of additional dilation, duration of abortion, blood loss measured by fluid in the suction jar, side effects, subject rating of pain, requests for pain medications, and overall acceptability to the women. Mean values of cervical dilation were analyzed using one-way analysis of variance. Dunnett C test was selected for post hoc multiple comparisons because of unequal variances between groups. The normality assumption was confirmed with the Shapiro-Wilk test. Other variables failed to meet the normality assumption, so they were compared using the Kruskall-Wallis test, with Dunn test for post hoc multiple comparisons, as were the ordinal-scale data. Dichotomous and categoric data were analyzed with χ2 test. Subgroup analyses based on parity, gestational age, and body surface area were done to determine whether these three factors affected the ability of the cervix to respond differentially to misoprostol and laminaria. Statistical analyses were done with SPSS Version 7.5 for Windows (SPSS, Inc., Chicago, IL).
Sample size was based on the primary outcome of cervical dilation and was estimated to provide 80% power in two-sided tests with an alpha of 5%, adjusted using the Bonferroni method for three two-way comparisons, to detect an effect size of 15%. This effect size denoted 1.2 mm difference in mean dilation between the laminaria and misoprostol groups and 1.0 mm between the misoprostol groups. There is evidence of increased variability in cervical dilation with misoprostol as compared with laminaria,9,15 so the assumption of common variance was avoided; in these calculations, we assumed that dilation would have standard deviations (SDs) of 1.0, 1.5, and 1.5 mm in the laminaria, oral-misoprostol, and vaginal-misoprostol arms, respectively. In this relatively complex situation, total sample size was minimized by unequal assignment by treatment arms. We used LaGrange multipliers in conjunction with the Newton-Raphson algorithm to find the optimal sample sizes for each arm under the constraint of providing adequate power for each of the three comparisons. These power analyses resulted in sample sizes of 12, 48, and 48 study participants in the laminaria, oral-misoprostol, and vaginal-misoprostol arms, respectively. As a safeguard, an additional four women were recruited for the laminaria arm of the trial, resulting in a 1:3:3 ratio of subjects per treatment arm.
Randomization within two strata, first and second trimester, was done using a computer-generated, random-number-producing algorithm with randomly permuted fixed blocks of seven, the minimum consistent with treatment assignment in a ratio of 1:3:3. A member of the department not involved with any other aspect of the study prepared sequentially numbered, opaque, sealed envelopes for allocation concealment. An error occurred in the transmission of sample size calculation and randomization scheme to allocation concealment, resulting in a random loss of six envelopes during preparation; a series of 106 envelopes was prepared for subject enrollment.
We recruited 106 women for the study: 14, 45 and 47 into the laminaria, oral-misoprostol, and vaginal-misoprostol arms, respectively. Relevant demographic characteristics among the three treatment groups did not differ significantly, except for abortion history (which failed to yield statistical significance by Dunn test; Table 1). The average time exposed to each of the three dilating agents did not differ significantly (Table 2). There was no chance maldistribution of the treatment groups among the five surgeons performing the abortions.
Vaginal misoprostol produced significantly greater mean cervical dilation than oral misoprostol (Table 2). The mean dilation in the laminaria group was not significantly different from the mean dilation in the oral or vaginal misoprostol groups. The proportion of subjects requiring additional dilation, the amount of extra dilation needed, the difficulty of additional dilation, duration of the procedure, and amount of blood loss were not statistically different among the three groups. Two notable but nonsignificant trends were observed. Women who received vaginal misoprostol required less additional dilation, and the extra dilation was easier to achieve than in subjects in the other groups (Table 2).
Questionnaire responses showed that laminaria placement was more painful than oral or vaginal misoprostol placement (Table 3), but there were no significant differences in pain, degree of other side effects, or medication requests during the 4-hour waiting period. Gastrointestinal complaints were rare in all groups: Four women reported nausea, six reported vomiting, and two reported diarrhea. Overall acceptability was assessed by the question, “If you had to have an abortion again, would you choose the same dilation method as you received in this study?” Responses to this question were not significantly different among the groups, regardless of whether the subjects had previous abortion experience. At least 90% of the women in each group reported that they were content with the method they were given.
Heavy vaginal bleeding occurred in 4.7% of the subjects overall: three of 47 with vaginal misoprostol, two of 45 with oral misoprostol, and none of 14 receiving laminaria. The bleeding in these five women was similar to menses, and although it was not severe, they were brought to the operating room early for suction curettage. This decision was made by the research nurse after the subjects had soaked one pad. In these cases, the cervix required no further dilation and bleeding ceased during the procedure. There were no abortion complications in the 106 women.
Subgroup analysis showed that vaginal misoprostol was more effective than oral misoprostol for cervical dilation in later pregnancies (later than 11 6/7 weeks; P < .05). In earlier pregnancies, mean dilations were 25.7 and 23.6 mm, whereas in later pregnancies the means were 33.4 and 25.7 mm for vaginal and oral misoprostol, respectively. No significant effects were found for history of vaginal delivery (dichotomized to none versus one or more vaginal deliveries) or body surface area (dichotomized on the median).
To our knowledge, this is the first randomized controlled trial to directly compare misoprostol with laminaria for dilation of the cervix before surgical abortion. Previous studies found that preabortion cervical dilation with vaginal misoprostol was greater than with placebo10,11 and as effective as gemeprost.13 Our data demonstrate that vaginal misoprostol achieved greater cervical dilation than oral misoprostol and was not significantly different from laminaria. There was no difference in side effects noted with the three agents over the 4 hours of exposure, but laminaria caused significantly more discomfort during placement than either route of misoprostol.
Vaginal bleeding similar to menses occurred in five of 106 subjects, all in the misoprostol groups. Previous studies also reported vaginal bleeding and even incomplete abortions with either route of misoprostol used before surgical abortion.8–11 These data suggest that although misoprostol is easy to administer, women may require observation during treatment in case early intervention is necessary. Future research should focus on the ideal timing and dose of misoprostol to achieve the most dilation with the least amount of vaginal bleeding, and explore self-administration of vaginal misoprostol before surgical abortion.
Most of the limitations of our study are statistical. Because the research setting was a teaching hospital that provides abortion training, two attending physicians and three residents performed the abortions. The primary outcome of cervical dilation would not depend on experience of the operator, but duration of the procedure and blood loss might. This could have increased the variability in these measurements, making it more difficult to reach statistically significant differences in these two variables. Most of our data were not normally distributed, resulting in our choice of an analysis of ordinal data and thus a loss of power. The smaller sample size of the laminaria group resulted in difficulty in detecting statistically significant differences in comparisons with this group. Sample-size calculations were based on published reports that found increased SDs in cervical measurements of misoprostol-treated subjects compared with laminaria; we did not find such differences. A larger number of subjects in the laminaria group would have improved the power to detect statistically significant differences in cervical dilation in the laminaria comparisons.
There is evidence that cervical preparation should be used routinely in first-trimester abortions to reduce the risks associated with manual dilation.1–3 The choice of agent depends on the setting and organization of patient care. In our study, vaginal misoprostol was easy to use and inexpensive and achieved the greatest cervical dilation with few side effects.
1. Cates W, Jr, Schulz KF, Grimes DA. The risks associated with teenage abortion. N Engl J Med 1983;309:621–4.
2. Schulz KF, Grimes DA, Cates W. Measures to prevent cervical injury during suction curettage abortion. Lancet 1983;i:1182–4.
3. Grimes DA, Schulz KF, Cates WJ, Jr. Prevention of uterine perforation during curettage abortion. JAMA 1984;251:2108–11.
4. Munsick RA, Fineberg NS. Cervical dilation from multiple laminaria tents used for abortion. Obstet Gynecol 1996;87:726–9.
5. Spitz IM, Bardin CW, Benton L, Robbins A. Early pregnancy termination with mifepristone and misoprostol in the United States. N Engl J Med 1998;338:1241–7.
6. Jain JK, Mishell DR. A comparison of intravaginal misoprostol with prostaglandin E2 for termination of second-trimester pregnancy. N Engl J Med 1994;331:290–3.
7. Sanchez-Ramos L, Kaunitz AM, Wears RL, Delke I, Gaudier FL. Misoprostol for cervical ripening and labor induction: A meta-analysis. Obstet Gynecol 1997;89:633–42.
8. Lawrie A, Penney G, Templeton A. A randomised comparison of oral and vaginal misoprostol for cervical priming before suction termination of pregnancy. Br J Obstet Gynaecol 1996;103:1117–9.
9. Ngai SW, Tang OS, Lao T, Ho PC, Ma HK. Oral misoprostol versus placebo for cervical dilation before vacuum aspiration in first trimester pregnancy. Hum Reprod 1995;10:1220–2.
10. Bugalho A, Bique C, Almeida L, Bergstrom S. Application of vaginal misoprostol before cervical dilation to facilitate first-trimester pregnancy interruption. Obstet Gynecol 1994;83:729–31.
11. Ficicioglu C, Tasdemir M, Tasdemir S. Effect of vaginal misoprostol application for cervical softening in pregnancy interruption before ten weeks of gestation. Acta Obstet Gynecol Scand 1996;75:54–6.
12. Platz-Christensen JJ, Nielsen S, Hamberger L. Is misoprostol the drug of choice for induced cervical ripening in early pregnancy termination? Acta Obstet Gynecol Scand 1995;74:809–12.
13. el-Rafaey H, Calder L, Wheatley DN, Templeton A. Cervical priming with prostaglandin E1 analogues, misoprostol and gemeprost. Lancet 1994;343:1207–9.
14. Ngai SW, Yeung KCA, Lao T, Ho PC. Oral misoprostol versus vaginal gemeprost for cervical dilation prior to vacuum aspiration in women in the sixth to twelfth week of gestation. Contraception 1995;51:347–50.
15. Darney PD, Dorward K. Cervical dilation before first-trimester elective abortion: A controlled comparison of meteneprost, laminaria, and hypan. Obstet Gynecol 1987;70:397–400.