Intramuscular versus intravenous oxytocin to prevent postpartum haemorrhage at vaginal delivery: randomised controlled trial

Study design and participants

This was a pragmatic double blind, placebo controlled randomised trial carried out in a large Irish maternity unit. Women were eligible for the trial if they were 18 years or older, were at 37 weeks or more gestation, had a singleton pregnancy, and were aiming for a vaginal birth with an actively managed third stage. We excluded women whose caregiver had previously decided to use an additional oxytocin infusion because of an increased risk of PPH, including those with a history of PPH owing to atonic uterus, known fibroids, a history of coagulopathy and receiving anticoagulant treatment, and thrombocytopenia. We excluded women with pre-existing cardiovascular disease and women who did not understand English.

Pregnant women were informed of the study in the outpatient clinics and antenatal classes and provided with an information leaflet that had been approved by the research ethics committees. Recruitment and written consent took place when women attended the assessment unit with symptoms and signs of early labour or when they were admitted for induction. They were eligible to be randomised when the second stage of labour was diagnosed or delivery appeared imminent. Diagnosis was based on the finding of full cervical dilation on vaginal examination, and in parous women, an urge to push.

Randomisation and masking

The Nottingham Clinical Trials Unit, registered with the UK Clinical Research Collaboration, randomised women to the allocated intervention (allocation ratio 1:1) using a secure automated web based randomisation system. Allocation was stratified by parity (nulliparous or multiparous) and blocked using random permuted blocks of varying size. Women were randomised to receive either an intramuscular bolus of oxytocin (10 IU in 1 mL) and a placebo intravenous injection (1 mL 0.9% saline given slowly over one minute) or an intravenous bolus of oxytocin (10 IU in 1 mL given slowly over one minute) and a placebo intramuscular injection (1 mL 0.9% saline).

A research fellow (NA, RC-T, CMcC) randomised women on diagnosis of the second stage of labour and prepared the trial syringes according to the trial allocation. Oxytocin and placebo were drawn up into two separate unlabelled sterile 2 mL syringes in the labour room and checked by the midwife. This was in keeping with the hospital policy for drawing up drugs and the stipulation of the research ethics committee. The labelling of the syringes—Drug 1-IM or Drug 2-IV injection—then took place outside the labour room according to the randomised allocation. The syringes were returned to the midwife in the labour room. This ensured blinding of everyone except the research fellow, who then played no subsequent role in patient management or outcome assessment.

Procedures

Women who were confirmed to be in labour or ready to have their labour induced, were brought into a room in the delivery suite. A sample was taken for a full blood count at the time of siting a venous cannula. The first and second stages of labour were managed as usual. To overcome potential variation in the actual practice of active management of the third stage of labour, a working definition was specified. Immediately after delivery and preferably within one minute of birth Drug 1-IM was administered intramuscularly into the thigh muscle followed by Drug 2-IV administered intravenously over one minute through a cannula. The baby could be positioned on the mother’s chest or abdomen after the birth while the cord was intact and immediate newborn care could be commenced. Owing to the effects of gravity on placental transfusion, babies should not be raised above this level with the cord still intact 18 At about one to three minutes after the birth the cord was clamped and cut in keeping with the recommendations of WHO and RCOG guidelines.37 However, early cord clamping was indicated in emergency situations, such as in the presence of a tight nuchal cord or if immediate resuscitation was required for asphyxia. The placenta was delivered by controlled cord traction once signs of separation were apparent (uterine contraction, show of blood, and apparent cord lengthening) while applying counter traction to the uterus with the other hand. To ensure the membranes do not tear, the placenta should be held in both hands and rotated carefully.

It is widely reported that clinicians underestimate rather than overestimate blood loss and for this study we formally measured blood loss rather than relying on subjective estimates.19 We measured blood loss after vaginal deliveries by direct collection of blood and the gravimetric method. After a normal vaginal birth, a fresh delivery sheet was placed under the mother’s buttocks as soon as the baby was placed on her chest or abdomen. Once there was no concern of further active bleeding, the blood loss was estimated by weighing all the soiled materials (swabs, pads, disposable sheets) on a scales and subtracting the known weights of these materials. Women remained on the delivery suite for a minimum of one hour after the birth and were monitored throughout that period. They were then transferred to the postnatal ward if there were no apparent complications.

For women who had an operative vaginal delivery, a surgical drape (Barrier Under Buttocks Drape; Mölnlyke Healthcare, Sweden) with a pouch for collecting vaginal blood loss, was placed under the woman’s buttocks in preparation for birth and before delivery of the placenta. The bag is transparent, allowing continuous monitoring of blood loss. The bag was left in situ until the birth attendant was no longer concerned about blood loss, such as when a sanitary towel was applied to the vulva and the legs were removed from the lithotomy position. The blood collected in the surgical drape was weighed. Blood soaked swabs and additional drapes were also weighed and the known dry weight of the swabs and drapes subtracted. This volume was added to the measured blood volume from the surgical drape.

After spontaneous vaginal births, the women were monitored on the labour ward for one hour.

The standard practice for recording blood loss and reporting PPH is based on the midwives’ subjective estimate of blood loss from the soaked pads, incontinence sheets, and drapes. We used this record of PPH for the intention to treat analyses for the small number of women whose blood loss was not formally measured.

Maternal vital signs were recorded after delivery and before transfer to the postnatal ward. We used an observation form to record side effects during the first hour after birth. Deviations from the standard procedure were recorded. If the uterus remained atonic despite the trial intervention, the midwife, obstetrician, or anaesthetist could use any additional uterotonic agents according to the hospital PPH protocol. Any major blood loss within 24 hours of delivery was also recorded. A full blood count was performed at day 1 (24 hours) after delivery to assess haemoglobin levels and haematocrit. Clinical follow-up of the mother continued until hospital discharge, and those who were readmitted with a maternal postnatal complication up to six weeks after the birth were flagged up.

Outcomes

PPH is defined by WHO as a blood loss of 500 mL or more within 24 hours after birth, and severe PPH is defined as a blood loss of 1000 mL or more within the same timeframe.3 Our primary outcome measure was PPH of 500 mL or more. We chose a measured blood loss of 500 mL or more as a clinically relevant outcome and to allow direct comparisons with other studies. Secondary outcomes addressed important measures of clinically relevant PPH, including severe PPH (measured blood loss ≥1000 mL), need for blood transfusion, mean total estimated blood loss, and admission to a high dependency unit. Additional secondary outcomes were the complications and consequences of PPH, including the need for additional uterotonic agents or additional surgical procedures, change in haemoglobin levels and haematocrit, severe anaemia (decrease in haemoglobin levels by ≥20%) 24 hours after delivery, and length of postnatal stay in hospital. The incidence of side effects after oxytocin treatment was recorded, including nausea, vomiting, hypotension, and tachycardia. We defined hypotension as a fall in blood pressure of more than 30% below the predelivery blood pressure or use of ephedrine, or both, and tachycardia as a heart rate persistently greater than 100 beats per minute. We recorded breastfeeding initiation within one hour of birth and at the time of discharge and readmissions for maternal postnatal complications up until six weeks post partum.

Statistical analysis

We estimated from local and national reports that 6-10% of study participants would have a blood loss of 500 mL or more.620 These estimates were based on observed recordings of PPH and it was therefore likely that the rate of PPH based on measured blood loss would be higher, consistent with our findings on the Elective caesarean Section Syntocinon Infusion Trial.21 With 500 patients in each group for analysis, the study would have 80% power at the 5% two sided α level to detect differences in the proportion of patients with a measured blood loss of 500 mL or more of 10% versus 5%, or 15% versus 9%, equivalent to odds ratios of 0.50 and 0.55, respectively. We inflated the final sample size to allow for women who subsequently delivered by caesarean section, and for missing outcome data or deviations from the trial protocol.

Data reporting was according to CONSORT guidelines for randomised controlled trials, and the trial statistician and research team conducted analyses blinded to group status.22 The statistical analysis plan was agreed before trial completion as described in the study protocol.23 As we aimed to evaluate PPH at vaginal delivery, we excluded women who delivered by caesarean section after randomisation but accounted for them in the CONSORT diagram.

For the first stage of analysis we used descriptive statistics to describe recruited women and to investigate comparability of the trial arms at baseline. The primary analyses involved intention to treat comparisons between the two groups for the primary and secondary outcomes adjusted for operative vaginal delivery and the stratification variable (parity). Secondary analyses investigated the effects of further adjustment for any variables displaying imbalance between the arms at baseline. All analyses used logistic or linear regression models, with results presented as point estimates (odds ratios or difference in means), 95% confidence intervals, and P values. We calculated the number needed to treat (NNT) with 95% confidence intervals for primary and secondary PPH outcomes. The NNT is the reciprocal of the absolute risk difference, or 1/adjusted risk ratio.24

Planned subgroup analyses evaluated the effects of parity and whether the onset of labour was spontaneous or induced. The analyses were replicated for women who completed the trial as per protocol, excluding those who were randomised but did not receive the trial intervention or who did not have blood loss measured (see supplementary tables).

The trial steering committee met quarterly. An independent data monitoring committee was established and reviewed recruitment and safety data after 700 women had been recruited. The women provided written informed consent.

Patient and public involvement

An initial pilot phase was completed, which helped inform the design of the main trial and the recruitment procedures. Recruited women were asked to provide feedback on how they were approached, the timing of recruitment and the consent procedures, blood testing and cannula insertion, and the presence of the researcher in the second stage of labour to prepare the trial drug syringes. Additional insights were provided by the lay members of the institutional research ethics committee, particularly with drafting of the patient information leaflet.