As a consequence of the widespread use of prenatal diagnosis, a need for second-trimester pregnancy termination has arisen when severe or lethal fetal anomalies are identified. There is frequently a requirement for autopsy examination of the fetus to define or refine the exact pathology diagnosis; hence, in the setting of fetal anomaly, medical termination of pregnancy is often performed. Prostaglandins (PGs) are the most common agents used to effect the medical termination of pregnancy in the second trimester, either alone or after pretreatment with mifepristone.
Since the first publication of our experience with misoprostol, 1 this PG has rapidly become the principal agent used for second-trimester pregnancy termination in our institution. The primary reasons for this prompt incorporation into standard practice include the low fiscal cost of misoprostol and the lack of stringent storage requirements. Misoprostol is a PGE1 analogue, designed and licensed for oral use as a therapy for gastric ulceration. The oral administration of misoprostol is associated with a rapid onset of action. 2,3 In obstetric practice misoprostol is well absorbed by the oral route, with peak plasma concentrations achieved earlier and higher than with vaginal administration, although the plasma concentrations are detectable for longer after vaginal administration. 3 Misoprostol is designed to be administered orally, and thus clinical studies using the manufacturer preparation by alternative routes have shown varying results. 2–4
In our anecdotal experience and that of others, 5 women have expressed a desire to have the misoprostol administered orally in preference to the vaginal route. We therefore conducted this clinical trial to assess the clinical efficacy of misoprostol in second-trimester pregnancy termination when administered vaginally compared with the oral route.
MATERIALS AND METHODS
This study was conducted as a randomized clinical trial. Misoprostol is not authorized for use in pregnancy in Australia, and permission to use the PG for this trial was obtained from the Australian Therapeutic Goods Administration under the Clinical Trials Notification scheme. The investigational protocol was approved by the Institutional Ethics Committee before commencement of the study.
Women admitted to King Edward Memorial Hospital for Women, Perth, Western Australia for second-trimester pregnancy interruption after a prenatal diagnosis of a fetal abnormality between March 2001 and July 2002 were invited to participate in the study. No restriction to recruitment was made in terms of the number of fetuses present or a history of previous uterine surgery. Gestational age was assigned on the basis of certain menstrual dates with confirmatory ultrasound, or ultrasound dating if the menstrual dates were uncertain or varied significantly from the menstrual dates. Once informed consent was obtained, the women were randomized to receive misoprostol in one of three dosage regimens from a series of opaque envelopes prepared by using random number tables. Because of the nature of the misoprostol administration, the women and staff were not blinded to the randomization allocation.
The misoprostol was administered by three therapeutic regimens. Women allocated to group 1 received 400 μg of misoprostol intravaginally at 6-hour intervals for 24 hours, repeated if delivery had not occurred to a maximum period of 48 hours. Women randomized to group 2 received 400 μg of misoprostol orally every 3 hours for a maximum time period of 48 hours. Those women randomized to group 3 received a loading dose of 600 μg of misoprostol intravaginally followed in 6 hours by an oral regimen of 200 μg of misoprostol at 3-hour intervals for a maximum total PG administration period of 48 hours. All women undelivered after 48 hours were considered protocol failures and received a transcervical Foley catheter with 2-hour extraamniotic PGF2α instillation or high-concentration intravenous oxytocin, depending on their cervical status and amniotic membrane integrity. No prophylactic medications were used to prevent the gastrointestinal side effects of misoprostol. Routine placental curettage was not employed in this trial. Spontaneous expulsion of the placenta within 60 minutes of delivery was awaited, with digital exploration of the uterine cavity and blunt curettage reserved for those cases where expulsion did not occur or was incomplete on the basis of clinical signs and symptoms.
Maternal pulse, blood pressure, and temperature were recorded at 3-hour intervals. A visual analogue assessment of pain was performed every 3 hours. The women performed the pain scale assessments with a visual analogue ruler scaled from 0 to 100, with 0 reflecting no perception of pain and 100 perceived as the most intense pain ever experienced. Analgesia was provided on patient request by intramuscular meperidine (2 mg/kg of maternal weight), with metoclopramide (10 mg intramuscularly) as the standard antiemetic medication. Within 24 hours of completion of the pregnancy termination process, the women completed a brief questionnaire using a visual analogue scale to ascertain their perceptions of the process in terms of pain, control, and overall opinion of the procedure.
The primary outcome measure was the delivery rate within 24 hours of PG commencement, and the rates of successful delivery were to be compared among the three study arms. It was expected from our previous studies 1,6 that at least 85% of patients would have successful deliveries in the vaginal administration group, and a delivery rate below 70% for the orally administered or combined groups would be regarded as clinically inferior. Maximal sample sizes of 75 in the vaginal group and 75 in each of the oral groups achieve 80% power at a 5% significance level, using a one-sided test for noninferiority for the proportion in each group to be 85% and the maximum allowable difference between these proportions that still results in equivalence to be 15%. However, because the possibility existed that the oral administration of misoprostol might result in delivery rates below 50% in 24 hours, the Institutional Ethics Committee requested a safety analysis at the point where 25 women had been recruited to each group. Group sample sizes of 25 achieve 80% power to detect a difference of 35% between the null hypothesis that both group proportions are 85% and the alternative hypothesis that the proportion in the oral misoprostol group is 50% using a one-sided Fisher exact test with a significance level of .05.
This analysis was carried out when there had been 28 women randomized to the vaginal misoprostol group and 29 to the oral misoprostol group. As the rates of delivery within 24 hours were 86% and 45%, respectively, the trial was closed. At this time, 27 women had been randomized to the vaginal loading with subsequent oral dosing misoprostol group with a delivery rate of 74%. No decision about noninferiority could be made for this group.
Data were analyzed using univariate and multivariable statistical techniques within Statistical Analysis System 8.02 (SAS Institute Inc., Cary, NC). Analyses were two tailed with the significance level set at .05 for all tests, and all were based on intention to treat. Numeric variables are presented as mean (standard deviation) or median (interquartile range) and were compared using analysis of variance or the Kruskal–Wallis χ2 test for variables displaying nonnormal behavior, tested using the Shapiro–Wilk statistic.
Group testing of categoric data is presented as n (%), and groups were compared with the Fisher exact test or χ2 test. Logistic regression analysis of variance was used to describe the probability of delivery within 24 hours in relation to the group effect as defined by intention-to-treat analysis, parity, and age of the woman; gestational age at delivery; and baseline variables of blood pressure and pulse. Kaplan–Meier survival analysis was used to calculate estimates of the median time to delivery and a corresponding 95% confidence interval, with groups being compared using the log rank test. The effects of parity and previous cesarean delivery on time to delivery were tested within this model using the log rank test. Mixed model analysis of variance was used to model group changes in the hemodynamic variables and temperature over the first 24 hours, with the individual woman included as a random effect to control for repeated measures. Occurrences of side effects within the first 12 hours are determined by calculating the area under the curve of the visual analogue scale scores taken at 3-hour intervals and compared using the Kruskal–Wallis test. Visual analogue scale scores for those women who had delivered in the first 12 hours were considered to be zero for the remainder of the time period.
Eighty-four women were recruited to the study during the 16-month period from March 2001 through July 2002. Twenty-eight women were randomized to group 1 and 29 to group 2, and 27 received the group 3 protocol. All pregnancy terminations were for fetal abnormality, and all fetuses were alive at the commencement of the termination process. No significant difference in maternal age, race, or prior uterine surgery among the three groups was present after randomization (Table 1). There were, however, significantly fewer women in group 1 who were nulliparous. For all groups the mean gestational age at recruitment was approximately 20 weeks (Table 1).
There was a significant difference in the median time to achieve delivery among the three groups, with those women randomized to the group 2 protocol having a significantly greater median duration from commencement to delivery than those in groups 1 and 3 (Table 2). Delivery was achieved in a median of 14.5 hours with the group 1 protocol, a median of 25.5 hours with the group 2 protocol, and 16.4 hours with the group 3 protocol (P = .042). The difference between group 1 and group 3 was not statistically significant with the reduced sample size (P = .083), but the median delivery interval was significantly different for group 2 versus group 1 (P < .001) and was close to the cutoff value for group 2 versus group 3 (P = .055).
There was no effect of parity (P = .573) or previous cesarean delivery (P = .622) on time to delivery. Delivery within 24 hours of PG commencement occurred in 85.7% of women in group 1, 44.8% of women in group 2, and 74.1% of those in group 3 (P = .003) (Table 2). Compared with group 2, the relative risk for delivery within 24 hours of PG commencement was 1.9 for group 1 and 1.65 for group 3. Transcervical Foley catheter insertion and extraamniotic PGF2α instillation were required in 20.7% of women in group 2 and 3.7% of women in group 3 who were undelivered after 48 hours of the allocated misoprostol regimen. No women in group 1 required further interventions to effect delivery (P = .011).
Using survival analysis techniques there was a significant difference in the induction to delivery interval between the three PG regimens (Figure 1). Group 2 demonstrated a significant increase in induction to delivery interval compared with groups 1 and 3.
Logistic regression analysis of the binary outcome variable, delivery within 24 hours, demonstrated no confounding effects of age (P = .423), parity (P = .542), gestational age (P = .680), or previous cesarean delivery (P = .122) but showed a highly significant group effect (P = .002).
Repeated-measures mixed model analysis of variance showed there was no significant difference among the three groups in terms of maternal hemodynamic parameters (systolic blood pressure P = .947, diastolic blood pressure P = .489, and pulse rate P = .630). For maternal temperature over the first 24 hours, there was a significant increase in level with time (0.01C per hour, P < .001). There was also an overall group effect (P < .001) with group 2 (0.3C) and group 3 (0.2C) being higher than group 1 (P < .001 and P = .005, respectively). There was no difference between the two vaginal groups (P = .429).
There was no difference in the area under the curve for the visual analogue scale scores for the first 12 hours for nausea (P = .616), but group 1 had significantly lower vomiting visual analogue scale scores (P = .026) than the other two groups (Table 3). There was a significant difference among the groups for diarrhea (P = .003), with the difference being a higher rate for group 2 than group 1 (P = .004) or group 3 (P = .007). There was no difference between the groups 1 and 3 (P = .625). Pain scores (P = .343) did not significantly differ among the three groups. Narcotic analgesic use and the incidence of placental retention were similar among the three groups.
We did not exclude women from the trial on the basis of previous uterine surgery. In group 1, 17.9% of women had experienced at least one prior cesarean delivery; in group 2, 13.8%; and in group 3, 22.2% (P = .71). No known case of uterine rupture occurred.
Within 24 hours of the termination process a simple four-question survey was given to the women to complete, and the responses are shown in Table 4. There was no significant difference in the responses provided regardless of the treatment allocation group; however, the women in group 2 (who received oral misoprostol alone) felt they had less control over their delivery process than those in the other groups (P = .049).
The abortifacient properties of misoprostol for second-trimester pregnancy termination have been reported in the medical literature since 1993. 7 Over the past decade several studies have demonstrated misoprostol to be an effective and cost-efficient agent for the medical interruption of the second-trimester pregnancy. 1,5–8 Misoprostol has been effectively used after priming with mifepristone in countries with access to this agent. 5,9 Our current study was designed to ascertain the effectiveness of misoprostol when administered as a sole agent orally compared with the vaginal route.
Misoprostol is manufactured to be administered by mouth, and peak concentrations of the active metabolite, misoprostol acid, are reached 30 minutes after oral administration, with a rapid decay. 2–4 In contrast, when the misoprostol tablets are administered vaginally, peak plasma concentrations of misoprostol acid are present after 80 minutes, with a slow decline thereafter. The maximum plasma concentration of misoprostol acid is higher after oral administration than vaginal, although in pharmacokinetic studies the reported incidence of side effects did not differ. 2 We selected a 3-hour dosing schedule for oral misoprostol based on the available published pharmacokinetic data. Because of the more prolonged effects of vaginally administered misoprostol and the published trial data, 1,6 we selected a dosing schedule of every 6 hours for the vaginal administration route. The combination of a vaginal and oral misoprostol regimen has been previously reported after pretreatment with mifepristone in second-trimester pregnancy termination 9 and would appear to have some potential benefits such as minimizing the number of vaginal examinations while maintaining the initial PG cervical priming influence. In this study we assessed the impact of such a combination approach of misoprostol administration routes in the absence of mifepristone priming.
This study reveals that vaginally administered misoprostol is superior in terms of induction to delivery interval to oral administration for second-trimester pregnancy termination. All women in the group assigned to vaginal administration of misoprostol delivered by 48 hours, with most delivering in less than 24 hours from PG commencement. In contrast, one fifth of women receiving the oral misoprostol regimen had not delivered after 48 hours. The addition of a vaginal misoprostol loading dose before commencement of oral administration reduced the delivery interval, reflecting the positive influence of vaginal misoprostol in pregnancy termination, but this difference was not statistically significant, perhaps because of the reduced sample size. Although the three groups were not identical in terms of parity, this difference did not impact significantly on the duration of the commencement to delivery interval. These marked temporal differences in the commencement to delivery intervals are clearly significant in terms of the health economics of hospitalization duration.
All fetuses in this study were anomalous, with the most common indication for pregnancy termination being an abnormal fetal karyotype (36%). All fetuses were alive at the commencement of the termination process, and no interventions to produce fetal demise were used. We have previously demonstrated that second-trimester pregnancy termination complicated by fetal death is usually more predictable, with a shorter induction to delivery interval, than that conducted when the fetus is alive. 1,6 A prolonged induction time may occur if the fetus is alive, and it is this group in which research efforts need to be directed. Thus, in this study we only recruited women in whom the fetus was alive at the commencement of the termination process.
The maternal PG side effect profile showed an increase in the incidence of objective gastrointestinal side effects (vomiting and diarrhea) for women allocated oral misoprostol. There were no significant differences in maternal perception of the termination process when analyzed by the misoprostol regimen received. Overall satisfaction scores were moderate (median 50–65), with no obvious preference for one regimen over the other. This is in contrast to a previous study 5 in which oral administration was viewed more positively by women than vaginal administration of misoprostol. The majority of women would recommend their allocated method of termination to a friend in a similar situation. This latter observation may reflect satisfaction with nursing care and support rather than the medical method of termination.
This randomized trial demonstrated that intravaginal misoprostol is more effective at achieving delivery within 24 hours than oral misoprostol in second-trimester pregnancy termination for fetal anomaly. No benefit in administering the first dose vaginally was observed relative to all doses orally. We conclude that intravaginal misoprostol at a dose of 400 μg at 6-hour intervals is a more superior dosing regimen than an oral regimen of 400 μg at 3-hour intervals for second-trimester pregnancy interruption.