A controlled, randomized, triple-blind clinical trial was conducted with 200 pregnant women with an indication for induction of labor at the Instituto de Medicina Integral Prof. Fernando Figueira (IMIP), Recife, Pernambuco, and at the Assis Chateaubriand Maternity Teaching Hospital (MEAC) of the Federal University of Ceará, Fortaleza, Ceará, both in northeastern Brazil, between July 2014 and November 2016. The study was registered at clinicaltrials.gov under reference NCT01406392.
Sample size was calculated using the Statcalc tool of Epi Info, version 3.5.2 for Windows (Centers for Disease Control and Prevention [CDC], Atlanta, GA, USA). Predicting a frequency of tachysystole of 6.7% in the group of women using 12.5 µg misoprostol sublingually  and of 21.6% in the group using 25 µg of misoprostol vaginally , for a 95% confidence level and a power of 80%, 98 women would have to be recruited to each group. To compensate for any possible losses, the total sample size was increased to 200 participants, 100 in each group.
The inclusion criteria were: indication for induction of labor, gestational age ≥ 37 weeks, single live fetus, vertex presentation, Bishop score ≤ 6, estimated fetal weight < 4000 grams, amniotic fluid index > 5 and good fetal well-being. The exclusion criteria were: previous Caesarean section, previous uterine scar resulting from uterine surgery, genital bleeding of unknown origin, fetal abnormalities, chorioamnionitis, tumors, malformations and/or ulceration in the vulvar or perineal region or birth canal that could be harmful to mother or child during the expulsion stage of labor, and HIV positivity. Eligibility was determined by performing digital vaginal examination to evaluate the Bishop score ; rapid HIV testing; and ultrasonography to estimate fetal weight and measure amniotic fluid volume. In addition, fetal well-being was assessed using Doppler flow velocity and/or cardiotocography and/or fetal biophysical profile and/or fetal vibroacoustic stimulation, depending on what was available in the maternity hospitals.
The vaginal and sublingual tablets of misoprostol and placebo were manufactured by Laboratório Hebron S.A. Indústrias Químicas e Farmacêuticas (Caruarú, Pernambuco, Brazil). Misoprostol for vaginal use consists of misoprostol together with lactose, microcrystalline cellulose, aerosil, explocel and sorbitol, and is available commercially under the trade name Prostokos® (25 µg). For this study, the pharmaceutical company also prepared the drug for sublingual use at the dose of 12.5 µg of misoprostol together with spray-dried lactose, croscarmellose sodium, crospovidone and magnesium stearate.
Randomization was performed using a single block of sequential numbers from 1 to 200 and the letters A and B. A statistician not otherwise involved in the study and who was unaware of what A and B represented prepared the list using the Random Allocation software program, version 1.0 (Isfahan, Iran). This list was sent to Laboratório Hebron where the pharmacist responsible for preparing the medication defined the meaning of A and B (sublingual or vaginal administration), without the investigators or the statistician having access to this information. Standardized, sequentially numbered, identical opaque packages were prepared in accordance with the randomization list. Each package contained either 8 active sublingual tablets of misoprostol (12.5 µg) and 8 placebo vaginal tablets or 8 placebo sublingual tablets and 8 active vaginal tablets of misoprostol (25 µg). Information on the contents of each package remained concealed until data analysis was complete. The vaginal and sublingual placebo tablets were identical in shape, size, color, smell, taste and weight to the tablets containing the active substance and were specially prepared for this study. Therefore, the triple-blind procedures were assured, since neither the investigators nor the patients nor the statistician were aware of the contents of each package.
To induce labor, the attending physician administered the sublingual tablets (misoprostol or placebo) and the vaginal tablets (misoprostol or placebo) at the same time, every six hours, except when the mother was asleep, up to a maximum of eight tablets. Whenever possible, the induction process began at 6 am and was interrupted at 10 pm if labor had not begun in the first 12 hours. The sublingual tablets were placed under the tongue and the vaginal tablets were inserted into the posterior fornix, with the patient then being instructed to lie on her left side for one hour to allow the tablet to dissolve spontaneously.
Patient monitoring and care occurred with no interference whatsoever from the investigators. The patients were examined every 30 minutes to evaluate fetal heart rate (FHR) and uterine dynamics. Digital vaginal examination was only performed to re-evaluate the Bishop score when the vaginal medication was given or when labor began or six hours after administration of the final tablet in order to diagnose failed induction of labor.
If any change in uterine contractility such as tachysystole, hypertonus or uterine hyperstimulation was identified during monitoring, the patient was placed in the left lateral decubitus position and hydrated with 1,000 ml of Ringer’s lactate solution in 30 minutes, and oxygen therapy and tocolysis with oral nifedipine 10 mg were prescribed . If the pattern of contractions and/or FHR did not return to normal, Caesarean section was indicated. Induction of labor was considered to have failed if labor had not been triggered six hours after administration of the final tablet, and a Caesarean section was then performed.
The primary outcome was the frequency of tachysystole. Secondary outcomes were: changes in the cervix at 12 and 24 hours; failure to achieve vaginal delivery within 12 and 24 hours; the mother’s preferred route of administration; time between the first dose and the onset of labor and delivery; duration of labor, need for oxytocin; failed induction of labor; Caesarean section and its indications; uterine hyperstimulation, need for epidural anesthesia; maternal side effects (nausea, vomiting, diarrhea, postpartum hemorrhage, fever); severe maternal morbidity (uterine rupture, sepsis, admission to intensive care unit) or maternal death; meconium in the amniotic fluid; non-reassuring FHR; one- and five-minute Apgar scores < 7, admission of the newborn to a neonatal intensive care unit; need for neonatal resuscitation; and severe neonatal morbidity (convulsions and neonatal asphyxiation) or perinatal death.
The control variables were mother’s age; gestational age at admission; amniotic fluid index; estimated fetal weight; parity; Bishop score; and the indications for induction of labor.
The abnormalities in uterine contractility evaluated were tachysystole (the presence of ≥ 6 uterine contractions for two consecutive 10-minute periods) [14, 15]; uterine hypertonus (a single contraction lasting 2 minutes or longer); and uterine hyperstimulation, with the presence of tachysystole or uterine hypertonus associated with non-reassuring FHR [14–16]. Non-reassuring FHR was defined as the persistence of FHR < 110 bpm or late decelerations in FHR (reduction in FHR following a uterine contraction - type 2 dip) and/or fetal tachycardia, persistent FHR > 160 bpm .
Statistical analysis was conducted using Epi-Info, version 3.5.3 (CDC, Atlanta, GA, USA). The categorical variables were compared using the chi-square test of association or Fisher’s exact test, as appropriate. Student’s t-test was used to compare the continuous variables with normal distribution and variances, while the non-parametric Mann-Whitney test was used for the discrete, ordinal or continuous variables for which distribution was not normal. P-values were two-tailed for all tests and the significance level adopted was 5%.
Risk ratios (RR) and their 95% confidence intervals (95%CI) were calculated as a measure of relative risk. The number needed to harm (NNH) was calculated for the primary endpoint.