Study setting and approvals

The study was conducted from April 1, 2010, through March 31, 2011, in a rural teaching hospital in West Bengal, India, which serves as a tertiary referral hospital. During this period a total of 542 eclampsia cases were admitted, of which 340 (62.73%) were antepartum and intrapartum eclampsia and 202 (37.27) were postpartum eclampsia. The study protocol was approved by the institutional review board. Since written informed consent requires a clear mind and many patients were confused or unconscious, we sought written informed consent from next of kin, generally the spouse.

Study participants

The sample consisted of women with clinical diagnosis of antepartum or intrapartum eclampsia, presenting at gestational age ≥34 weeks, with singleton pregnancy, os <3 cm dilated, normal fetal heart rate (by auscultation), cephalic presentation, normal coagulation profile, and no other obstetric complications. We believed that once a patient is in active phase of labor, that is os >3 cm, she is likely to achieve vaginal delivery within a reasonably short period. Cesarean delivery in some of these women would have been inappropriate, although it might also have benefited some of them. The gestational age was determined on the basis of history, antenatal records, clinical examination and, where available, second trimester ultrasonography (USG) records. Women with known contraindication to labor and vaginal delivery (eg, placenta previa, cephalopelvic disproportion diagnosed clinically) or any medical complications such as heart disease, diabetes mellitus, or chronic renal disease or known lethal fetal congenital anomalies, were excluded. Large proportion of patients had USG reports from antenatal period, which allowed exclusion of cases with congenital anomalies. USG was not done at or after admission. We depended on clinical findings, such as grossly small-for-date uterus, excessive uterine tonicity, and fetal heart rate abnormalities on auscultation, to exclude women with compromised fetuses and abruption. HELLP syndrome cases with low platelet count were also excluded as many cases were to undergo cesarean section under spinal anesthesia.

Randomization

Eligible subjects whose next of kin consented to participate in the trial were randomly allocated to either planned cesarean delivery (group A) or to vaginal delivery (group B) arms. Randomization was done in blocks of 20 using computer-generated random number lists. The delivery mode was noted on cards that were then individually placed in opaque serially numbered envelopes and sealed. A telephone-based alternative to this traditional allocation concealment technique was not logistically feasible in our setting. The mode was revealed immediately prior to commencing the treatment for a subject. The randomization was performed by a statistician not otherwise involved in the conduct of the study.

Baseline characteristics

The baseline parameters recorded for the study were maternal age, parity, whether cared for by health care providers during the antenatal period, first convulsion to admission interval, number of convulsions before admission, blood pressure, and consciousness level at the time of admission. The birthweight was recorded for every baby delivered, even if stillborn.

Treatment protocols

All women received magnesium sulfate as anticonvulsant as per the institutional protocol: 3 g (20%) was given intravenously and 2.5 g (50%) was given intramuscularly in each buttock on admission (total 8 g). Subsequently 2.5 g (50%) was given intramuscularly to alternate buttock every 4 hours. If there was recurrence of convulsions, another 2 g (20%) was administered intravenously. Magnesium sulfate was continued for 24 hours postpartum, with clinical monitoring, that is periodic assessment of respiratory rate, knee jerk, and urine output. Serum magnesium levels were not monitored. Antihypertensive was given if systolic blood pressure was >160 mm Hg or diastolic blood pressure was >110 mm Hg, targeting systolic blood pressure between 140-160 mm Hg and diastolic blood pressure between 90-100 mm Hg. Labetalol was the antihypertensive agent used: 20-80 mg was given intravenously in incremental doses every 20 minutes as needed. After delivery, oral labetalol was used as soon as the patient was able to take drugs orally, and continued in both groups. The blood pressure was controlled, although not normalized in all cases, by use of labetalol. Intravenous fluids were restricted to a maximum of 85 mL/h.

For women allocated to vaginal delivery group, a pre-agreed protocol for management of labor was followed. Induction was achieved with misoprostol 25 μg vaginally every 4 hours for a maximum of 5 doses. Fetal heart rate and condition of cervix were assessed before giving each dose of misoprostol. As soon as the woman went into labor or the Bishop score was >5, misoprostol was stopped. The membranes were ruptured when os was at least 2-3 cm dilated, cervical effacement was >80%, and head engaged. Labor was augmented with intravenous oxytocin if there were ineffective uterine contractions. Oxytocin was administered at least 4 hours after the last dose of misoprostol. Adequate progress of labor in the first stage was defined as cervix dilating at a rate of at least 0.5 cm per hour after the onset of active labor. The fetal heart rate was monitored intermittently by stethoscope (every 15 minutes in the first stage and every 5 minutes in the second stage of labor). If labor failed to progress satisfactorily even after augmentation with oxytocin or if fetal heart rate abnormality occurred, cesarean delivery was undertaken; otherwise labor was allowed to progress and baby was delivered vaginally, either spontaneously or with the help of instruments. Epidural analgesia was not used in labor as this is not used in our institution. Instead, tramadol 100 mg was administered intravenously. The vaginal deliveries were conducted by residents under supervision of senior obstetricians.

Women allocated to cesarean delivery group underwent cesarean section as early as possible after initial stabilization and necessary assessment that included full blood cell count (including platelets) and clotting time (clot observation test) examination. Cesarean delivery was performed by consultants either under spinal anesthesia or under general anesthesia as decided by the anesthetist in charge of the patient. At birth, the neonate was assessed by the pediatrician on call. Mothers and their babies were followed up for 7 days.

Outcome measures

The ideal primary outcome measures for such a study are maternal and perinatal deaths. However, since these are relatively rare events, we decided to use 2 composite end points–one pertaining to the mother, the other to the newborn–as our primary outcome measures. The “maternal composite event” included maternal death and severe maternal morbidity indicated by ≥1 among the following: respiratory depression, need for ventilator support, pulmonary edema, pneumonia, renal failure, hepatic failure, coagulopathy, cerebrovascular accident, and admission to intensive care unit. Respiratory depression was diagnosed by clinical findings: respiratory rate <12/min and oxygen saturation <92%. Pulmonary edema was diagnosed by clinical findings, pulse oximetry, and chest x-ray. Pneumonia was diagnosed by clinical findings, chest x-ray, and leukocytosis. Renal failure was diagnosed clinically by oliguria along with raised serum creatinine level. Hepatic failure was diagnosed by clinical findings, serum bilirubin level, and liver enzyme values. Coagulopathy was diagnosed from bleeding manifestations and laboratory coagulation profile. For neonates, the composite event included perinatal death and severe neonatal morbidity indicated by Apgar score <7 at 5 minutes, delivery room intubation, or admission to special care baby unit for >7 days. In addition, some other events related to labor and mode of delivery were recorded as secondary outcome measures. These included need for blood transfusion, wound infection (infected abdominal or episiotomy wound), puerperal febrile morbidity (temperature >38°C on ≥2 occasions in any 48-hour period, excluding the first 24 hours after delivery), evacuation of hematoma (rectus sheath hematoma or vulval hematoma), and intraoperative trauma (including uterine artery laceration; laceration of the bladder, bowel, or ureter; or extension of the uterine incision in cesarean delivery cases and cervical tear or third- or fourth-degree perineal tear in instrumental delivery cases). The length of hospital stay, total magnesium sulfate dose, and recurrence of convulsions after admission were also treated as secondary outcome measures.

Sample size and statistical analysis

The primary measure of interest for this study was taken as the composite adverse maternal and adverse neonatal outcomes. Survey of our hospital records for the past 2 years showed that the rate for the combined composite outcome was around 30% for vaginal delivery. Logically, a reduction in the event rate by 10% in case of cesarean delivery would be sufficient grounds for considering cesarean section to be the preferred mode of delivery in eclampsia cases. It was estimated that 294 women would be required in each group to demonstrate a 10% difference in composite outcome with 80% power and 5% probability of type 1 error. For detecting 5% reduction in the event rate, the sample size went up to 1251 women in each arm. Since a marked difference in composite event rate was not likely for the 2 modes of delivery in eclamptic mothers, this situation was suited for a noninferiority trial. Assuming a noninferiority margin of 5% it was estimated that 986 women would be required in each arm to establish noninferiority of vaginal delivery in comparison to cesarean delivery with 80% power and 5% probability of type 1 error.

Since we were constrained by the fact that long-term availability of the investigator team was not assured due to administrative reasons, and recruitment of so many subjects was not feasible in a short time in our single-center setting, we decided a priori to include 100 subjects in each group, over a period of 1 year, to assess actual event rates through randomized intervention. Therefore, we viewed this trial as exploratory, allowing informed estimates for a larger definitive trial or as a first study for potential inclusion in a metaanalysis of similar studies. The experience gained from this study, in terms of barriers to recruitment, compliance with treatment protocols, and data collection teamwork, would also allow better conduct of a later definitive trial.

The results were analyzed according to intention-to-treat principle, and all women who were randomized were included in the analysis. Numerical variables were compared between groups by Student unpaired t test, and categorical variables by Fisher exact test or χ ² test as appropriate. A 2-sided P value < .05 was considered statistically significant. Event rates associated with cesarean delivery were compared with those in vaginal delivery to calculate relative risk (RR) with 95% confidence interval (CI). Software (Statistica, version 6; Stat Soft Inc, Tulsa, OK; and Graph Pad Prism, version 4; Graph Pad Software Inc, San Diego, CA) was used for statistical analysis.