le Roux, Paul A. MBChB; Olarogun, Jeremiah O. MBChB, FCOG; Penny, James MBChB, MRCOG; Anthony, John MBChB, FCOG
Misoprostol, a prostaglandin E1 analogue, is an inexpensive and potentially useful drug to women requiring induction of labor. Three meta‐analyses have concluded that misoprostol is an effective oxytocic, but have not recommended its routine use because of an increased incidence of tachysystole and possible adverse fetal outcome.1–3 Hofmeyer et al have concluded that “if sufficient numbers are studied, an unacceptably high number of serious adverse events including uterine rupture and asphyxial deaths may occur” (p. 798).3
Both efficacy and adverse events are likely to be dose related. Most studies assessed in meta‐analysis have used regimens of 50 μg (or more) every 4 hours. There are two possible methods of reducing the potency of the drug: either the dose of the drug can be decreased, or the dosage interval can be prolonged.
This study aimed to assess the oral and vaginal routes of misoprostol administration in a 6‐hour regimen, compared with the standard treatment using prostaglandin E2 gel, dinoprostone.
MATERIALS AND METHODS
From April 1999 to November 2000, all women with obstetric or medical indications for induction of labor were recruited to the trial. Half the patients were recruited at Groote Schuur Hospital, which is a university teaching hospital and tertiary referral center for women with severe maternal disease. The remainder were recruited at Mowbray Maternity Hospital, which is a district hospital receiving patients referred from primary level Midwife Obstetric Units.
The study was approved by the University of Cape Town Ethics Committee and the South African Medicines Control Council. Misoprostol was imported from the United States as 100‐μg tablets, which were divided in half to get 50‐μg doses. Written informed consent was taken from all women who participated in the trial.
A total of 573 women were recruited to the study. There were 93 “protocol violations” where women were excluded. The reasons for the protocol violations were: clerical errors (65 patients), exclusion criteria ignored (18 patients), patient withdrew (one patient), incorrect dosage given (five patients), underage (two patients), and cesarean section performed after randomization before prostaglandin was given (two patients). Computer randomization to three groups included: vaginal misoprostol, oral misoprostol, and a control group receiving dinoprostone. There were twice as many women randomized into the dinoprostone group. After exclusions, there were 480 patients suitable for analysis. The randomization also ensured that equal numbers of patients in each group were recruited at the two different hospitals. The instructions/questionnaires were placed in sealed opaque envelopes, and opened by the labor ward medical attendants when the patient was randomized. Demographic data including age, gravidity, parity, gestational age, and indication for induction were recorded.
Inclusion criteria were any women having induction of labor with a singleton cephalic fetus of 34 weeks or more in whom the membranes were unruptured, cardiotocography showed no fetal distress, and there were no painful contractions. The minimum age for recruitment was 18.
Exclusion criteria were a previous cesarean section, parity greater than 4, or intrauterine fetal anomaly or death. Patients who had a Bishop score greater than 7, where amniotomy was possible without the prior use of prostaglandins, were also excluded.
A Bishop score was established before randomization as well as a 20‐minute cardiotocograph recording. A vaginal examination was also performed at 6‐hour intervals. Once the patients were in active labor (ie, three strong contractions in 10 minutes and greater than 3 cm dilated), they were not given further prostaglandins. Intravenous oxytocin was not administered routinely, and artificial rupture of membrane was not performed as part of the induction process. However, once in the active phase of labor, if progress was judged to be slow on the partogram as a result of inadequate contractions, artificial rupture of membranes was then performed, and incremental oxytocin infusion administered (doses from 2 mU to 12 mU per minute).
In the misoprostol group, the dose of 50 μg was given either orally or inserted by the medical attendant into the posterior fornix every 6 hours, to a maximum of four doses. In the dinoprostone group, the dinoprostone gel (1 mg) was inserted into the posterior fornix and the dose repeated after 6 hours (maximum two doses). The patients were managed by the usual daily labor ward team consisting of an obstetric registrar and intern. The Bishop score and the number of doses of prostaglandin given were recorded.
Continuous external fetal monitoring and external tocodynamometry were used on all patients. The cardiotocograph traces were analyzed later by the first two authors to assess whether tachysystole or fetal distress was present. The authors were not blinded to the induction agents used, but evaluated the traces according to strict criteria. Tachysystole was defined as five or more uterine contractions in 10 minutes, present in two 10‐minute windows. Fetal distress was defined as a fetal heart rate tracing that justified immediate delivery.
Induction of labor was allowed to continue despite tachysystole, in the absence of fetal heart rate abnormalities. The next dose of prostaglandin was, however, withheld until the cardiotocograph normalized. Fetal distress was managed by the usual labor ward protocol. This included the administration of β‐sympathomimetic drugs (hexoprenaline). The analgesia available to the trial patients included opiate analgesia, entonox gas, and epidural anesthesia. This was documented.
The primary outcome studied was the incidence of vaginal delivery within 24 hours. This was chosen because it had been used in similar studies. If the patient was undelivered after 24 hours, further management occurred at the discretion of the consultant in charge. If the mother's condition deteriorated during the 24‐hour study period, a cesarean section was performed for maternal reasons. The mode of delivery (vaginal delivery or cesarean section) was recorded, as well as the time of delivery (including those after the first 24 hours). The indication for cesarean section was also recorded.
The presence of fetal complications was noted. These included: the presence of thick meconium‐stained amniotic fluid, low 5‐minute Apgar score (less than 7), admission to the neonatal intensive care unit, or the diagnosis of hypoxic ischemic encephalopathy by the pediatrician.
The null hypothesis was that fewer than 33% of patients treated with misoprostol would deliver within 24 hours. The alternate hypothesis stated that 33% or more patients would deliver vaginally over a 24‐hour period after administration of misoprostol. The α value was set at 0.05 and the power at 80%. The sample size calculation showed that 120 patients were needed in each misoprostol group, with group‐specific delivery rates of misoprostol, 45%, and dinoprostone, 30%.
The computer packages Microsoft Excel (Microsoft South Africa, Johannesburg, South Africa), Epiinfo 6 (Centers for Disease Control and Prevention, Atlanta, GA), and GraphPad Prism (GraphPad Software Inc., San Diego, CA) were used for data analysis.
Statistical analyses on continuous variables were performed using one‐way analysis of variance tests on normally distributed data and Kruskal‐Wallis tests on data that were not normally distributed, with appropriate posttests. Categorical variables were analyzed using χ2 tests or relative risk ratios. Where multiple comparisons were performed, relative risks were expressed with 99% confidence intervals.
The subjects were similar with respect to age, gravidity, parity, mean Bishop score, gestational age, and indication for induction in the misoprostol and dinoprostone groups (Tables 1 and 2).
The primary outcome was vaginal delivery in 24 hours (Table 3). In the vaginal misoprostol group, 57% of patients delivered vaginally in the first 24 hours compared with 54% in the dinoprostone group. The median induction to delivery time was 12.3 hours in the vaginal misoprostol group compared with 14.8 hours in the dinoprostone group (P < .01). In the oral misoprostol group, significantly fewer patients (39%) delivered vaginally in the first 24 hours with a longer median induction to delivery time of 22.6 hours (P < .001).
When comparing oral and vaginal misoprostol groups, significantly fewer patients were delivered vaginally within 24 hours in the oral group. The induction to delivery time was also significantly longer in the oral group.
The median number of doses of misoprostol in the vaginal misoprostol group was two compared with the three doses in the oral misoprostol group (P < .001). The median number of doses of dinoprostone was two.
There was no significant difference in the use of oxytocin between groups (6.6%, 16.6%, and 16.2% in the vaginal misoprostol, oral misoprostol, and dinoprostone groups, respectively). Significantly fewer patients in the vaginal misoprostol group required artificial rupture of membrane (23.3%) compared with the dinoprostone group (40.4%).
There were more patients with tachysystole in the vaginal misoprostol group (5.8%) compared with the dinoprostone group (0.8%), although this was not a statistically significant difference. There were no significant differences in the incidence of thick meconium, low 5‐minute Apgar scores (less than 7), admission to neonatal intensive care units, or hypoxic ischemic encephalopathy.
The overall cesarean section rate was 33% with no differences between the treatment and control groups. The indications for cesarean section were, however, significantly different in the vaginal misoprostol group compared with the dinoprostone group. There was a higher proportion of cesarean sections for fetal distress (27.5%) in the vaginal misoprostol group compared with the dinoprostone group (13.7%), and fewer cesarean sections for failed induction of labor after 24 hours in the vaginal misoprostol group (0.8%) compared with the dinoprostone group (9.1%).
At Groote Schuur Hospital (tertiary level care), the number of patients delivered vaginally in the first 24 hours was 45.8%, compared with 55.6% of patients at Mowbray Maternity Hospital (secondary level care) (P = .03). There was no difference in the cesarean section rate between the two hospitals. The cesarean section rate was 33% and 34% at Groote Schuur Hospital and Mowbray Maternity Hospital, respectively.
In the regimen used, misoprostol did not result in more vaginal deliveries in 24 hours than dinoprostone. In this study, the dosage interval was extended to 6 hours to assess efficacy and complications of both oral and vaginal routes of administration.
In the vaginal group, efficacy was maintained with the 6‐hour regimen. However, the increased incidence of tachysystole observed in other studies was again demonstrated in this study with vaginal misoprostol administration, even with a 6‐hour dosage interval. Wing and Paul showed that if the dose was reduced to 25 μg at 6‐hour intervals, there would be a loss of efficacy with vaginal misoprostol.4
There was a decrease in efficacy with oral misoprostol, which may still be judged as an acceptable compromise because there was no increase in tachysystole or fetal distress with this regimen. The 6‐hour regimen also resulted in fewer intrusive vaginal examinations for women, and fewer assessments by busy labor ward medical attendants. In the oral misoprostol group, the benefit of a 6‐hour dosage interval was not shown, and a 4‐hour dosage interval may be better still.
The route of administration of misoprostol clearly has an effect on efficacy. This is supported by pharmacodynamic studies. In 1997, Zeiman et al, with large doses (400 μg) showed that the oral route had a a steep rise and then fell at 120 minutes, compared with the vaginal route where there was a slow rise, and 61% of the peak level of drug was still present at 4 hours.5 However, increasing efficacy is associated with a trend towards increased tachysystole and an increase in cesarean sections for fetal distress within the first 24 hours. These differential effects do not mean that only one route of administration should be adopted as the “correct” route. Urgent delivery may require a shorter induction time, and vaginal administration may be the route of choice. Conversely, where the indication for delivery is not urgent, oral administration of misoprostol would be the most suitable method. Other factors such as parity, gestational age, and cervical status may also be relevant when choosing the route of administration.
There were no differences noted in terms of fetal complications, but this study did not have the power to detect significant differences. However, the lack of any serious adverse events was noteworthy because this study was carried out among a group of women with high‐risk pregnancies. Underlying placental insufficiency characterized many of the women studied, and this would highlight differences in drug effects if they existed. The data, however, remain inconclusive, and a sample size of 4000 women would be needed to address this issue. Although the published meta‐analyses have fetal complications as secondary outcomes, the drawbacks of assessing many studies with different dosage regimes and obstetric populations makes the assumptions inconclusive.
The reduced need for artificial rupture of membranes with the vaginal misoprostol regimen is important in hospital settings with a high human immunodeficiency virus carrier rate because this would lower transmission rates. The oral misoprostol route would also be of benefit because fewer vaginal examinations would be performed, which would be associated with less postpartum infectious morbidity in immunocompromised women.
The conclusion that misoprostol was not more effective should not necessarily be seen as a negative result. There was bioequivalence (or similarity between the groups) when assessing reasonable limits for the treatment outcomes. The concept of bioequivalence when comparing treatment effects is well described in the literature by Phillips,6 and is accepted by the Food and Drug Administration and other research groups evaluating drugs. The bioequivalence model supports the conclusion that although there was not a statistical difference found (which was the intention of the trial design), there may have been benefits in the bioequivalence between the groups. After a finding of similar efficacy, the other advantages and disadvantages of the drugs need to be considered, before concluding which treatment regimen is ultimately the best one. The cost benefit of misoprostol over dinoprostone is a strong argument for using misoprostol for induction of labor.