Misoprostol in the treatment of postpartum haemorrhage in addition to routine management: a placebo randomised controlled trial


Dr G. Walraven, Aga Khan Health Services, Secrétariat de Son Altesse l'Aga Khan, Aiglemont, 60270 Gouvieux, France.


Postpartum haemorrhage remains a leading cause of maternal mortality, despite treatment with conventional methods. In this randomised controlled trial, we compared misoprostol 600 μg (200 μg orally and 400 μg sublingually) with placebo in the treatment of postpartum haemorrhage in addition to routine treatment. One hundred and sixty consenting women who delivered vaginally with measured blood loss ≥500 mL and for whom inadequate uterine contraction was thought to be a possible factor were given either misoprostol or placebo in addition to normal treatment and after routine active management with uterotonics. Blood loss was measured by collection in a special plastic bedpan and side effects of treatment were recorded. Measured average additional blood loss was 325 mL (95% confidence interval [CI] 265 to 384 mL) with misoprostol and 410 mL (95% CI 323 to 498 mL) with placebo. No severe side effects were noted in the use of misoprostol.


Haemorrhage is the largest single medical cause of maternal death, accounting for about 25% of the global total and claiming an estimated 150,000 lives annually.1 Most of these deaths are due to postpartum haemorrhage, primarily due to atonic uterus. When postpartum haemorrhage occurs, a number of medical and surgical interventions are used to control bleeding.2 One component in the treatment of postpartum haemorrhage is uterotonic therapy, most commonly with oxytocin and/or ergometrine. There is good evidence that their routine prophylactic use as part of the active management of the third stage of labour is effective in reducing the risk of postpartum haemorrhage.3 Routine use of misoprostol for preventing postpartum haemorrhage has been found to be less effective than conventional uterotonics.4

Beyond preventive use of misoprostol, its therapeutic use for the treatment of postpartum haemorrhage has been promoted, particularly in resource-poor situations, where its non-parenteral administration, stability without refrigeration and low cost make it highly attractive. Unfortunately, misoprostol is beginning to enter clinical use without systematic research to document its optimal dosage, effectiveness or risks. In six ‘uncontrolled’ studies, 41 women with postpartum haemorrhage were treated with misoprostol using different doses and administration routes.5 All but two of the women (who received 1000 μg misoprostol rectally) appeared to respond promptly to treatment. It is difficult to be sure whether these observed responses were due to misoprostol or other simultaneous treatments such as conventional uterotonics or bimanual compression of the uterus. A small randomised trial with 64 study subjects involving the use of misoprostol (800 μg) rectally showed promising results.6

The evidence for the effectiveness of misoprostol for treatment of postpartum haemorrhage is thus promising but unsubstantiated. A dose of 600 μg appears safe in the prevention of postpartum haemorrhage, with an acceptable level of side effects and good increase in uterine activity. Prior to this study, we recommended a misoprostol dosage of 600 μg (200 μg orally and 400 μg sublingually) for the management of postpartum haemorrhage based on available knowledge of pharmacokinetic, physiological and clinical studies of misoprostol (Hofmeyr, unpublished report for WHO). The combined route of oral and sublingual administration was used as these two routes combined together have shown the most optimum pharmacokinetic profile (rapid onset, highest peak and long duration).

The objective of this trial was to evaluate whether misoprostol administered 200 μg orally and 400 μg sublingually is safe and confers additional benefits when combined with conventional management for women with postpartum haemorrhage (measured blood loss ≥500 mL).


Women admitted to the labour room of the Farafenni APRC Hospital, Gambia were given information about the trial in an appropriate local language by a midwife and invited to sign or thumb print consent to participate.

Management of the third stage of labour was by routine active management with oxytocin 10 units or syntometrine one ampoule (5 mL). Women diagnosed with postpartum haemorrhage (measured blood loss ≥500 mL within 1 hour) and for whom inadequate uterine contraction was thought to be a possible factor, were given standard routine treatment for postpartum haemorrhage (rubbing the uterus if it was found to be poorly contracted, commencing an intravenous infusion, administrating oxytocics, delivering the placenta if undelivered and emptying the bladder).7 In addition, they were enrolled by opening the next in a series of randomised treatment packs in opaque envelopes containing either misoprostol three 200 μg or placebo tablets, each administered as one tablet orally and two tablets sublingually. The tablets were similar in size and colour but not in shape. Efforts to obtain identical placebo tablets were unsuccessful. Although there was no account where the midwife caught sight of the tablet, this is not sufficient guarantee of adequate blinding.

Immediately after administration of the study medication, the woman was positioned on a clean bedpan on top of a fresh large perineal pad with plastic backing, for a further period of at least 1 hour. The hour was measured using a standard timer. However, the blood collection ended only when the midwife considered active bleeding to have stopped. The blood collected in the bedpan was then transferred to a measuring jar. The measuring jar and all gauzes and pads used were put in a standard plastic bag and the total difference between the dry and wet weights was calculated. A 100-g increase in weight was considered to be equivalent to 100 mL of blood (i.e. specific gravity of blood was assumed to be 1 g mL−1). Under-estimation of blood loss was unavoidable as some blood can be expected to spatter on the drapes and on the gowns of the attendant although attempts were made to minimise such losses. Furthermore, maternal blood contained in placental interstices was also not measured. As these under-estimations, as well as any over-estimations (e.g. due to amniotic fluid) were likely to be distributed equally between the randomised study groups, there was no need for such corrections.

Side effects were recorded by the attending midwife as they were observed or reported in response to direct questioning. Haemoglobin level was estimated between 12 and 24 hours after delivery. Time of delivery and taking of the haemoglobin estimation were recorded on the data form. Haemoglobin was measured by taking a fingerprick and by using a Heamocue Haemoglobin machine and cuvette (HemoCue, Ängelholm, Sweden). Demographic and obstetric characteristics of the women were recorded to assess comparability of the study groups at trial entry and data were collected on outcomes and possible confounding factors.

Exclusion criteria were the lack of consent for participation, women who had a caesarean section, blood loss <500 mL in the first hour after delivery, if the delivery occurred at <28 weeks of gestation and if inadequate uterine contraction was not thought to be a possible causative factor for the postpartum haemorrhage. The primary outcome measure was average additional blood loss after enrolment. Secondary outcome measures were frequency and severity of side effects, assessed and recorded using standardised data capture sheets, additional blood loss ≥500 mL after enrolment, clinical complications (need for blood transfusion, hysterectomy) and haemoglobin 12–24 hours after delivery.

Sample size was calculated using the primary outcome measure. It was assumed that, to detect a 100-mL difference in additional average blood loss after enrolment (assuming mean additional blood loss of 400 mL in the placebo group and 300 mL in the misoprostol group, and a standard deviation of 200 mL in both groups) at 5% significance level with 80% power, 63 women per group would be required. Data were doubly entered, checked and cleaned using Epi-Info 6.0 and analysed using Stata 7.0 (Stata, Texas, USA). The randomisation code was broken only after entry and checking of data. Comparisons were by the χ2 or Fisher's exact probability test and relative risks with 95% confidence intervals (95% CI) for categorical data, while for continuous data the t test was used and 95% confidence intervals were calculated. Ethics approval was obtained from the ethics committee of the Gambia Government and MRC Laboratories. An independent data monitor (Dr A. M. Gülmezoglu) reviewed the data collected from the initial 80 women and recommended that the study to continue until complete recruitment.


During the study period from November 2002 to October 2003 there were 1094 vaginal deliveries at Farafenni AFPRC hospital. In 1048 women (95.8%), consent was obtained and blood loss was measured, and in 174 of these women (16.6%) blood loss was measured to be ≥500 mL. In 160 women, inadequate uterine contractions were thought to be a possible causative factor for the postpartum haemorrhage.

There were no withdrawals after enrolment, and all outcomes were analysed according to the allocated study group. The two groups were similar with regard to baseline characteristics (Table 1). There was no difference between the two groups in the amount, route and type(s) of other uterotonics used as prophylactic management and in the initial treatment of the postpartum haemorrhage. There was no problem recorded with the sublingual administration and acceptability of study medication. The study outcome variables are summarised in Table 2. Measured average additional blood loss was 325 mL (95% CI 265 to 384 mL) after misoprostol and 410 mL (95% CI 323 to 498 mL) after placebo, while the medians were 231 and 277 mL, respectively. Shivering and mild fever (temperature ≥37.5°C) were more common in the misoprostol group. There were only two women with a temperature higher than 39°C, and none of the participants had a temperature above 40°C. There was no report of severe side effects or other adverse events in the misoprostol group. With regard to the other secondary outcomes, namely, additional blood loss ≥500 mL (16.5%vs 28.4%) and ≥1000 mL (2.5%vs 6.2%) after enrolment, and use of additional uterotonics treatment for postpartum haemorrhage (3.8%vs 6.2%) was lower in the misoprostol than in the placebo group but not statistically significant. There was no difference in low postpartum haemoglobin or the proportion of patients requiring a blood transfusion between the two groups.

Table 1.  Comparison of baseline variables between women allocated to receive misoprostol or placebo in the treatment of postpartum haemorrhage in addition to normal treatment. Values are shown as mean [SD] or n (%).
 Misoprostol (n= 79)Placebo (n= 81)
Age (years)24.6 [6.4]24.1 [5.4]
Primiparous23 (29.1)26 (32.1)
Parity ≥ 612 (15.2)12 (14.8)
Fundal height (weeks)24.7 [3.6]24.6 [2.9]
Twin delivery8 (10.1)5 (6.2)
Singleton stillbirth9 (11.4)13 (16.1)
Birthweight for singletons (g)2891 [739]2909 [692]
Blood loss prior to randomisation (mL)702 [191]757 [338]
Episiotomy/tear18 (22.8)16 (19.8)
Manual removal of placenta3 (3.8)3 (3.7)
Use of therapeutic uterotonics22 (27.9)28 (34.6)
Table 2.  Comparison of outcome variables between women allocated to receive misoprostol or placebo in the treatment of postpartum haemorrhage in addition to normal treatment. Values are shown as mean [SD] or n (%).
 Misoprostol (n= 79)Placebo (n= 81)Effect size* (95% CI)
  • *

    Relative risk except for average blood loss after taking study medication for which the difference between means employed instead.

Primary outcome
Average blood loss after taking study medication (mL)325 [264]410 [397]−85 (−191 to +20.5)
Secondary outcomes
Shivering23 (29.1)8 (9.9)2.95 (1.40 to 6.19)
Nausea3 (3.8)5 (6.2)0.62 (0.15 to 2.49)
Headache7 (8.9)11 (13.6)0.65 (0.27 to 1.60)
Temperature ≥ 37.5°C16 (20.3)8 (9.9)2.05 (0.93 to 4.52)
Blood loss ≥ 500 mL13 (16.5)23 (28.4)0.58 (0.32 to 1.06)
Blood loss ≥ 1000 mL2 (2.5)5 (6.2)0.41 (0.09 to 1.77)
Postpartum Hb < 6 g dL−1 or blood transfusion12 (15.2)12 (14.8)1.02 (0.49 to 2.15)
Hysterectomy02 (2.5)
Use of additional uterotonics35 (6.2)0.62 (0.17 to 2.25)


The purpose of this trial was to explore any additional effect to conventional methods, and to establish the safety of 200 μg oral and 400 μg sublingual misoprostol in addition to routine uterotonics in the treatment of postpartum haemorrhage. Side effects were mild to moderate and comparable with findings from previous randomised controlled trials of misoprostol to prevent postpartum haemorrhage.8 The results of this study showed that the average additional blood loss, additional blood loss ≥500 mL and ≥1000 mL after enrolment, and use of additional uterotonics were lower in the misoprostol group but did not reach statistical significance.

In the large WHO collaborative trial of misoprostol in the management of the third stage of labour,4 blood loss ≥1000 mL and use of additional uterotonics (the primary outcomes in this trial) were more frequent with misoprostol than with oxytocin 10 units, indicating that oxytocin should remain the drug of choice for routine prophylaxis in those settings where it is possible to administer this drug. Given the advantages of misoprostol (stability at high temperature, oral administration, long shelf-life and low cost) and the fact that postpartum haemorrhage remains an important cause of maternal mortality especially in the developing world, it is important to assess the use of misoprostol in the treatment of postpartum haemorrhage.

In the WHO collaborative misoprostol prevention trial, blood transfusion was used less frequently in the misoprostol group. However, this was not a primary outcome for this study, and hence, it may have been a chance occurrence. There could be a synergistic pharmacological effect between misoprostol and conventional uterotonics. Many women in the misoprostol group thus received misoprostol as well as conventional uterotonics, whereas those in the oxytocin group received only conventional oxytocics. A pharmacokinetic study linked to the WHO trial reported that a longer time to peak level with misoprostol (20–30 minutes) compared with syntocinon (3 minutes) may account for more early blood loss with misoprostol. Physiological studies have also shown a more rapid onset of uterine contractions following syntometrine compared with misoprostol after delivery. This does not exclude the possibility that misoprostol may have an effect on more persistent bleeding.

In this study, the use of blood transfusion was more frequent and postpartum haemoglobin was lower in the misoprostol group compared with the placebo group. The seemingly positive results in the misoprostol group regarding additional blood loss and use of uterotonics should be interpreted cautiously as this is a small study with limited power to only detect large differences. Postpartum haemorrhage accounts for more maternal deaths in places where access to hospital delivery is limited and in situations where other interventions to treat bleeding are not available. To conduct a large-scale randomised trial of misoprostol for treatment of postpartum haemorrhage in a setting in which conventional uterotonics are not available poses logistical and ethical problems. If misoprostol is found to be effective over and above the effects of conventional treatment, it is a biologically reasonable assumption that it will be at least as effective or more so when conventional uterotonics are not available.


The authors would like to thank Sisawo Konteh, and staff at the maternity ward at Farafenni AFPRC hospital for their support and Dr Metin Gülmezoglu for his helpful discussions. The authors would particularly like to thank the women participating in this study.

Accepted 21 March 2004