Intended for healthcare professionals

Clinical Review Evidence based case report

Use of prostaglandins to induce labour in women with a caesarean section scar

BMJ 1999; 318 doi: (Published 17 April 1999) Cite this as: BMJ 1999;318:1056

This article has a correction. Please see:

  1. Sarah Vause, specialist registrar,
  2. Mary Macintosh, consultant.
  1. Department of Obstetrics and Gynaecology, Leeds General Infirmary, Leeds LS1 3EK
  1. Correspondence to: Dr S Vause, Saint Mary's Hospital, Manchester M13 0JH
  • Accepted 12 November 1998

Mrs B, a 26 year old woman, was 10 days post term in her second pregnancy. Her first child had been delivered by elective caesarean section at 38 weeks' gestation because of a breech presentation. Two doses of prostaglandin E2 vaginal gel (1 mg), administered six hours apart, were used to induce labour. Eight hours after the second dose, Mrs B's cervix was found to be soft,fully effaced, and dilated to 3 cm, but her uterine contractions were mild and irregular. Amniotomy was performed and an infusion of oxytocin was started. Mrs B made good progress in labour with epidural analgesia.

Initially the fetal cardiotocograph was normal. However, four hours after the oxytocin infusion was started, fetal tachycardia with recurrent early decelerations was noted. At that time Mrs B's cervix was dilated to 9 cm. While preparations were being made for fetal scalp pH sampling, there was a prolonged deceleration in the fetal heart rate and fresh vaginal bleeding. Uterine rupture was suspected. Mrs B was transferred immediately to theatre, where this was confirmed. The baby showed no signs of life at delivery and could not be resuscitated. MrsB required a blood transfusion at the time of operation but made a good recovery afterwards.

When the case was discussed at the monthly perinatal meeting, it became apparent that the obstetricians had different views on the advisability of using prostaglandins to induce labour in women with a scarred uterus. Some doctors believed that the main benefit of induction over repeat elective caesarean section was that some women would achieve a vaginal delivery and the mode of delivery in subsequent pregnancies would not therefore be compromised. Others expressed concern that the risk of scar rupture after a previous lower segment caesarean section might be higher with prostaglandins than with other methods of induction. They suggested that the prostaglandin has the same “softening” effect on the fibrous tissue of the scar as it did on the cervix.1 Hypertonic uterine activity is more common with prostaglandin than with other methods of induction,2and it has been suggested that this may also increase the risk of scar rupture.

Searching for evidence

We decided to try to resolve the debate by searching the published reports. Initially, we looked for randomised controlled trials comparing labour induced by prostaglandins with awaiting spontaneous labour and those comparing prostaglandins with oxytocin for inducing labour in women with a previous caesarean section

We searched the Cochrane Database3 and Medline using the terms “vaginal birth after cesarean section and prostaglandin” and “uterine rupture and prostaglandin.” We identified only two randomised controlled trials and therefore widened the search to include the following types of observational studies in women with a previous caesarean section:

  • Induction of labour by prostaglandins compared with awaiting spontaneous labour

  • Prostaglandins compared with oxytocin for induction of labour

  • Case-control studies of scar rupture.

Case reports were not included. The CINAHL and EMED databases were also searched; this search was restricted to papers published in English

The main outcome measures were the vaginal delivery rate and the incidence of scar rupture. Where necessary, we contacted the authors of the studies to obtain further information.

Evidenceor the lack of it

We were unable to find any studies in which the use of prostaglandins to induce labour was compared with an active decision to await spontaneous labour in managing women with a uterine scar from a previous section. However, there were two randomised controlled trials comparing the induction methods of prostaglandin with oxytocin plus amniotomy in these women.45 Personal communication established that the larger trial (n=42)4 was an extension of the smaller one (n=36),5 and we therefore considered only the former. In this study,4 a predetermined code in a sealed envelope was used to randomise women to vaginal prostaglandin followed by amniotomy or to amniotomy and intravenous oxytocin. Because the trial was small, it lacked power. This was only sufficient(α=0.05, β=0.80) to show a halving of vaginal delivery rates, from80% to 40%. The trial showed that more women in the oxytocin group had a repeat caesarean section because they failed to establish labour.In the prostaglandin group, all women became established in labour.There was one case of uterine rupture, in a woman in the prostaglandins group who was also given oxytocin.

Non-randomised cohort studies

We did not find any non-randomised cohort studies comparing the use of prostaglandins with alternative management in women with a uterine scar. In particular, neither the active decision to await spontaneous labour nor alternative methods of induction (oxytocin) had been formally compared with the use of prostaglandins. There were many (over 20) observational studies which included the use of prostaglandins in women with a uterine scar,628 but only six studies in which this was the main focus.611 These six studies were reviewed in greater detail (table). There were no comparison groups in two studies, 89and two had fewer than 30 cases.67The remaining studies, based on around 100 cases, were retrospective, and the controls were nullipara.1011The vaginal delivery rate in the prostaglandin groups in these six studies ranged from50.4%11 to 84%.7In the largest study it was 75% (95% confidence interval 70.9% to79.1%).8

Study designs and methods, and number of women achieving vaginal delivery

View this table:

Case-control studies of uterine rupture

Uterine rupture is rare —it occurs in between 0.3% to 0.7% of labours. 2930 The terms “rupture” and “dehiscence” are not used consistently, which hinders further assessment of these events. Assuming a background risk of 0.5%,prospective studies would have to include 10 000 women in order to show a twofold increased relative risk. It is unlikely that a such a large comparative study would be feasible. A case-control design is appropriate for studying any association between prostaglandins and an infrequent event such as rupture of a uterine scar. However, we did not identify any case-control studies and could not estimate the relative risk of scar rupture associated with prostaglandins. If the absolute risk (rather than the relative risk) of uterine rupture is calculated from the largest study (n=439),8 the rate of uterine rupture is 0.2% (0 to 0.6%) where symptomatic rupture is considered,but 1.1% (0.1 to 2.1) where cases of asymptomatic dehiscence are included.


There is a dearth of evidence from which to assess the risks and benefits of using prostaglandins to induce labour in women with a scar from a previous lower segment caesarean section. This is surprising considering how often prostaglandins are used for inducing labour and how frequently doctors have to make a decision about inducing labour in a woman who has had a previous caesarean section

Our systematic search did not identify the largest observational study.8 This was found by chance as a reference in a book.31 We also had difficulty in obtaining a copy of the results as these were not published in a peer reviewed medical journal,but in conference proceedings from 10 years ago. It could be argued that these data are not suitable reference material for an evidence based case report. However, this was by far the largest series of cases, and should therefore have provided the best estimate of the absolute risk of uterine rupture. Personal communication with the author established that the case series (n=439)8was an extension of an earlier case series (n=143) reported in a peer reviewed journal.32 Faced with the lack of other evidence, we felt justified in including the study in this review. It does, however,illustrate the difficulty in deciding which evidence is the most useful and appropriate to use.

While it is difficult for clinicians to find and assimilate the evidence, it should be remembered that the woman is also a participant in the decision making process. For some women the shorter recovery time and the feeling of satisfaction if a vaginal delivery is achieved would be important factors in favour of inducing labour. 3334 For others, the convenience and feeling of being in control, apprehension at the possibility of a failed trial of labour followed by an emergency section,35 or the possibility of perineal damage even if a vaginal delivery is achieved may make an elective caesarean section preferable to induction. It has been suggested that statistical information on medical risks may be less critical than social issues for women in this decision making process.36

Further research in this area is required. Randomised controlled trials are needed to assess aspects such as vaginal delivery rates, duration of labour, number of failed inductions, requirements for analgesia, and patient satisfaction. To detect a 10% difference in vaginal delivery rates, a randomised controlled trial would have to include 580 participants (α=0.05, β=0.80). However, uterine scar rupture is such an infrequent event that an impractically large randomised controlled trial would be needed to address this issue directly. An alternative approach would be to perform a case-control study.

Until studies have been performed, doctors have limited evidence to guide them. If women with a previous lower segment caesarean scar are induced with prostaglandins using the preparation and regimen suggested by MacKenzie,8 the risk of symptomatic scar rupture seems to be no greater than 0.6%, and the vaginal delivery rate is approximately 75%—similar to rates quoted for spontaneous labour in women with a caesarean scar. At present, faced with the lack of comparative evidence, clinicians can only provide women with the best estimate of risk based on uncontrolled observational data.

Evidence into practice?

Once we had finished reviewing the evidence, we presented our findings to the obstetricians and midwives at the monthly audit meeting. Most clinicians agreed that there was insufficient evidence to support the development of a guideline on whether women with a previous caesarean section scar should or should not be offered induction with prostaglandins. Diversity of practice has continued in the unit—but possibly within a more tolerant, evidence based culture.


Since this review was carried out, a further cohort study has been published.37 The findings of this study do not alter our conclusions.

Contributors: SV performed the literature search, participated in discussing and assimilating the evidence, presented the evidence to the clinicians at the audit meeting, and participated in writing the paper. MM supervised the literature search, participted in discussing and assimilating the evidence, and contributed to writing the paper. Mr M Glass, consultant obstetrician, participated in discussing the evidence and reporting this case. Mr J Thornton identified the topic as suitable for evidence based review.


  • Funding None.

  • Competing interests None declared.


View Abstract