Statistical analysis
The primary outcome was CS, analysed according to the intention-to-treat
principle. The primary outcome measure was evaluated using logistic
regression models, initially unadjusted and then after adjustment for
pre-determined important potential confounding variables and covariates.
Effect sizes are presented as the odds ratio in CS delivery rates
between the two study treatment arms with its 95% confidence interval.
Secondary outcome measures were also evaluated unadjusted and then,
where possible, adjusted for the same pre-determined variables and
covariates using logistic regression, ordinal (ordered) logistic
regression, Poisson regression, Cox proportional hazards and standard
regression models according to data type. Stata was used for all
analyses.
A formal interim analysis was performed by an independent data and
safety monitoring committee (IDSMC) after 214 women were randomised,
whilst safety data was reported bi-annually. Stopping rules for the
interim analysis were in accordance with O’Brien and
Fleming.13 The IDSMC had the authority to request
further interim analyses if indicated but this was not exercised.
The study was sponsored by the University of Liverpool, registered with
ClinicalTrials.gov (NCT03749902) and the Clinical Trial Registry, India
(CTRI/2019/04/018827). Consumer
representatives reviewed the protocol and participant-facing
documentation at all stages, and a representative sat on the Trial
Steering Committee.
RESULTS
Overall, 1033 women were recruited to participate in the MOLI study
between 6th of January 2020 and the 14th July 2022
when the sample size was reached (Figure 1). Two of the participating
government hospitals, located in Nagpur, India, recruited participants
from the launch of the study in 2020 till the trial ended. Recruitment
was temporarily halted on 19th March 2020 due to the
COVID-19 pandemic, before restarting again from 1stOctober 2020 with additional precautions against infection for
participants and staff in line with Indian government recommendations. A
third site, Mahatma Gandhi Institute of Medical Science in Sevagram, was
added and began enrolling patients in February 2021.
Of the 1033 women, 520 required ongoing induction after membrane rupture
following cervical preparation with
LDOM. These women were randomised
to receive either continued LDOM (n = 260) or oxytocin infusion (n =
260). There was no missing data or loss to follow-up. Prior to
randomisation, the participants had received a mean (s.d.) of 2.9 (1.7)
and 3.0 (1.7) doses of LDOM respectively for cervical preparation.
Membrane rupture was spontaneous in 65 (25.0%) and 60 (23.1%)
respectively; the remainder underwent artificial membrane rupture.
The two randomised groups were well matched by age, parity, and severity
of disease (Table 1). Most women had not given birth previously, were
close to their ultrasound-estimated delivery date and mild
non-proteinuric hypertension.
The primary outcome of caesarean section (CS) was similar in the two
arms: 84 (32.3%) for women who had ongoing induction with LDOM and 71
(27.3%) for those who received oxytocin infusion (adjusted odds ratio
1.226 (95% confidence interval 0.814 - 1.847); p = 0.329 (Table 2)).
The most common reasons for CS were progress failure in the
1st stage of labour and fetal heart rate
abnormalities; no statistically significant differences were detected
between the two study arms. In an exploratory subgroup analysis, the
preterm gestations showed a statistically significant increase in CS in
the misoprostol group (Figure 2).
Rates of fetal heart rate abnormality and meconium-stained liquor were
similar in each group. There were no cases of uterine hyperstimulation
(contractions >5 in 10 minutes), no differences in the
rates of placental abruption, postpartum haemorrhage, manual removal of
placenta, receipt of blood products, or hypertensive complications. No
woman in either group experienced a uterine rupture, admission to
intensive care or death.
Women in the LDOM group took a statistically longer time to give birth
(geometric mean time from randomisation to birth 225 versus 194 minutes
in the oxytocin group; Figure 4), with the difference driven by those
who had normal vaginal births. There was no statistically significant
difference for those who underwent CS.
Fewer babies allocated to the LDOM group were admitted to the special
care baby unit (Table 3); while these babies tended to spend more time
there, this difference was not statistically significant (p=0.510).
Furthermore, there was no statistically significant difference between
the babies receiving ventilation, resuscitation, or intubation. All
other neonatal outcomes were similar between the two arms. Three babies
died neonatally and all were in the oxytocin arm. The causes of death
were septicaemia (a 2.5kg baby), asphyxia (a 1.7kg baby with severe
hypoxic-ischaemic encephalopathy with pulmonary bleed and with
cardiovascular arrest), and severe growth restriction (1.1kg baby with a
severe pulmonary bleed). None were thought to be related to the study
medications.
After giving birth, participants in both groups showed high levels of
acceptability with their augmentation method (Figure 3, Table 4). Only
16 (6.2%) of the women in each group would not be happy to have the
same method used again for future inductions if needed. There was no
difference between groups in their acceptability ratings on the time
taken to give birth, the amount of pain during the induction and birth,
or their anxiety.
DISCUSSION