Titrated oral misoprostol solution for induction of labour: a multi-centre, randomised trial

British Journal of Obstetrics and Gynaecology September 2001, Vol. 108, pp. 952±959

Titrated oral misoprostol solution for induction of labour: a multi-centre, randomised trial G.J. Hofmeyr a,*, Z. Al®revic b, B. Matonhodze a, P. Brocklehurst c, E. Campbell b, V.C. Nikodem a Objectives To determine the effects of titrated oral misoprostol solution, compared with vaginal dinoprostone. Study design Open, randomised clinical trial. Setting Academic hospitals in South Africa and Liverpool, UK. Methods Women undergoing induction of labour after 34 weeks of pregnancy were allocated by randomised, sealed opaque envelopes, to induction of labour with titrated oral misoprostol solution, or two doses of vaginal dinoprostone (2mg) administered six hours apart. Failure to deliver within 24 hours of randomisation was the primary outcome on which the sample size was based. The data were analysed by intention-to-treat. Results Six hundred and ninety-®ve women were randomly allocated: 346 to oral misoprostol and 349 to vaginal dinoprostone. There were no signi®cant differences in substantive outcomes. Vaginal delivery within 24 hours was not achieved in 38% of women in the oral misoprostol group and 36% in the vaginal dinoprostone group (RR 1.08; 95% CI 0.89-1.31). The caesarean section rates were 16% and 20%, respectively (RR 0.80; 95% CI 0.58-1.11). Hyperstimulation with fetal heart rate changes occurred in 4% of women in the oral misoprostol group and 3% after vaginal dinoprostone (RR 1.32, 95% CI 0.59±2.98). The response to induction of labour in women with unfavourable cervices was somewhat slower with misoprostol when membranes were intact, and with dinoprostone when membranes were ruptured. There were no differences in neonatal outcome between the two groups. Conclusions This new approach to oral misoprostol administration was successful in minimising the risk of uterine hyperstimulation, which has been a feature of misoprostol use for induction of labour, at the expense of a somewhat slower response in women with intact membranes and unfavourable cervices. Misoprostol is not registered for use in pregnant women, and further research is needed to con®rm optimal and safe dosages.

INTRODUCTION Induction of labour is a common intervention during pregnancy in both developing and developed countries. In the UK in 1994/1995, 19.5% of women had labour induced 1, which corresponds to approximately 130,000 women each year. Licensed pharmacological methods used for induction of labour, mostly vaginal prostaglandins, are expensive and require cold storage making them unsuitable for resource poor settings. Cheaper alternatives which are stable at room temperature may have the potential to produce substantial cost savings in devel-

a

Departments of Obstetrics and Gynaecology, Coronation/ Frere/Cecilia Makiwane Hospitals and Effective Care Research Unit, University of the Witwatersrand, South Africa b Department of Obstetrics and Gynaecology, Liverpool Women's Hospital, University of Liverpool, UK c National Perinatal Epidemiology Unit, Oxford, UK * Correspondence: Professor G. J. Hofmeyr, Frere and Cecilia Makiwane Hospitals and Effective Care Research Unit, University of the Witwatersrand, Private Bag X9047, East London 5200, South Africa. q RCOG 2001 British Journal of Obstetrics and Gynaecology PII: S03 06- 5456( 01) 0023 1-5

oped countries and allow safe induction of labour in those countries which currently cannot provide pharmacological induction of labour. Misoprostol is a unique prostaglandin E1 analogue which is rapidly absorbed orally. Tablets, marketed for anti-in¯ammatory drug-induced gastric ulceration, are stable and inexpensive. The use of misoprostol in pregnancy has been reviewed recently 2. Several randomised trials of induction of labour with misoprostol have been undertaken 4±10. In most trials, the vaginal route has been chosen, presumably because this route has been most successful for other prostaglandins and because misoprostol has a far longer half-life when administered vaginally than orally 3. However, the short half-life of oral misoprostol may be an advantage in induction of labour, given the varying responses of women to prostaglandins and the danger of uterine hyperstimulation. Previous randomised trials of oral misoprostol have used a ®xed dosage regimen administered four to six hourly 4±8, 100-200mg three-hourly 9, or 100mg vaginally followed by 100mg orally two-hourly 10. In November 1999 the American College of Obstetricians and Gynecologists provided guidelines to their members on the use of misoprostol for cervical ripening and induction of labour 11. The manufacturers have acknowledged the www.bjog-elsevier.com

TITRATED ORAL MISOPROSTOL FOR INDUCTION OF LABOUR 953

role of doctors in prescribing an approved pharmaceutical product for use outside its approved indication 12. The Cochrane Review 13 on the topic remains inconclusive because widely varying dosage regimens have been used and the studies to date have been far too small to address the issue of safety. We have emphasised the dangers of misoprostol, particularly in larger dosages, and the need for large studies to con®rm safety. Our review in this journal 14 has attracted criticism because of our view that misoprostol should not be recommended for widespread use before more controlled evidence of safety is available 15. In this study, in order to allow for individual sensitivity and the short half-life of oral misoprostol, we planned to use small doses at frequent intervals, increasing the dosage when necessary. Following completion of a small dose-®nding pilot study 16, we carried out a randomised controlled trial of the effectiveness of titrated oral misoprostol solution and dinoprostone for induction of labour in the third trimester of pregnancy. METHODS Women due for induction of labour with a singleton pregnancy and cephalic presentation were asked to participate. The exclusion criteria were: gestation less than 34 weeks, signi®cant antepartum haemorrhage, previous caesarean section, uncontrolled diabetes mellitus, severe pre-eclampsia or eclampsia and fetal distress. All eligible women were asked to sign a written consent form. The local research ethics committees approved the study protocol. To allocate treatment, the next in a series of sealed, consecutively numbered opaque envelopes was taken from one of four dispensers, depending on whether the membranes were ruptured and whether the cervix was favourable (modi®ed Bishop score .6) 17: 1. 2. 3. 4.

membranes intact/cervix unfavourable; membranes intact/cervix favourable; membranes ruptured/cervix unfavourable; membranes ruptured/cervix favourable.

The envelopes were prepared remote from the study sites by one author (P.B.), using a computer-generated list of allocations. Allocations were balanced between groups using random block sizes (between two and six). Each envelope was numbered sequentially and contained the appropriate management protocol. During the analysis the allocation recorded on the data form was cross checked with the lists held at Oxford. The woman's name was entered in a register with the number of the envelope before opening the envelope. The management protocol included in the envelope was followed unless clinical imperatives dictated otherwise. When labour q RCOG 2001 Br J Obstet Gynaecol 108, pp. 952±959

commenced, electronic fetal heart rate monitoring was used whenever feasible. The study was not placebo controlled, therefore women and clinicians were aware of the allocated treatment. The treatment protocols were as follows: 1. Misoprostol The smallest preparation of misoprostol available in trial countries was a 200mg tablet. For induction of labour a starting dose of 20mg was required. To overcome the problem of breaking the 200mg tablet into small fragments, we dissolved the tablet in 200ml water (1mg per mL), shaking the solution well before each administration. At the South African sites, tap water was used, and the same bottle used till ®nished (maximum 12 hours). At the Liverpool site, sterile water was used and the solution made up freshly for each administration. The solution was administered orally every two hours until adequate uterine contractions occurred (three per 10 minutes lasting 30 seconds) and then stopped. The initial dosage of 20mL (20mg) was increased to 40mg after two (Liverpool) or three doses (South African sites). The timing and strength of contractions was assessed by regular abdominal palpation. If the contractions became inadequate, augmentation of the active phase of labour was attempted with hourly titrated oral misoprostol (1mg/mL solution) escalating the dose from 5ml to 10ml and, 20ml (maximum). If uterine contractions were judged to be adequate, the next dose of misoprostol was omitted. 2. Conventional method Dinoprostone (Prandin, prostin gel) 2mg was inserted into the posterior vaginal fornix and repeated after six hours if the contractions were not adequate. If labour was not established after a further 12 hours, or contractions became inadequate, an oxytocin infusion was commenced with syntocinon at 2miu per minute, increasing if necessary to 4miu, 8miu, 16miu or 32miu per minute. At the South African sites, women with intact membranes were randomly allocated to three groups, the third being for induction of labour with an extra-amniotic Foley catheter bulb. The results of this comparison will be reported separately. Arti®cial rupture of the membranes was performed at the discretion of the person attending the woman. The South African centres followed a conservative policy because of the high prevalence of HIV infection. Tachysystole (.5 contractions per 10 minutes for at least 20 minutes) or hypersystole (a contraction lasting at least two minutes) were initially diagnosed by combination of abdominal palpation and cardiotocography. These conditions were managed conservatively with the woman being placed in the left lateral position with continuous

954 G. J. HOFMEYR ET AL.

Fig. 1. Trial pro®le.

fetal heart rate and contraction monitoring. Arti®cial rupture of the membranes was avoided if possible. Hyperstimulation syndrome (tachysystole or hypersystole with fetal heart rate abnormality) was managed by changing to the left lateral position, oxygen administration by face mask and if considered necessary, tocolysis (intravenous hexoprenaline or ritodrine).

subgroups for analysis were by membrane status, cervical status and country. Comparison by country was considered necessary because of anticipated differences in patient populations (more high risk women at the South African sites), and management (more conservative policy to membrane rupture at the South African sites), and because of minor differences in the misoprostol induction regimen (see above).

Statistical considerations The sample size calculation was based on the incidence of the primary outcome Ð failed vaginal delivery Ð within 24 hours, of 58% in the dinoprostone groups reported in randomised trials of vaginal misoprostol compared with dinoprostone 14. To detect a reduction to 45% with 95% certainty and 90% power would require 324 women in each group. Data were entered into Epi-info 6 or Excel at the recruiting sites and analysed independently using SPSS. Differences between groups were expressed as relative risks with 95% con®dence intervals. Pre-speci®ed

RESULTS The number of women randomised was 695. The trial Pro®le is shown in Fig. 1. Data were not kept on women excluded from participation. The characteristics of women at trial entry are shown in Table 1. Post-term pregnancy (41%), prelabour rupture of membranes (17%) and hypertension (16%) were the most common indications for induction of labour at all three sites. Most women had intact membranes and an unfavourable cervix at the time of trial entry (n ˆ 443, 64%) of whom 48%

Table 1. Characteristics of women at trial entry expressed as n (%), mean [SD] or median {interquartile range}. IOL ˆ induction of labour; POH ˆ past obstetric history; PROM ˆ premature rupture of the membranes. Denominators are given when data are missing. South Africa

No. randomised Age (years) Missing Gestation (weeks) Missing Previous pregnancy $24 weeks Primary indication for IOL Impaired growth Post-term Hypertension POH PROM Other Not known Cervical score (Bishops) ,7 Median Missing Membranes ruptured Not known

Liverpool

Combined

Misoprostol

Dinoprostone

Misoprostol

Dinoprostone

Misoprostol

Dinoprostone

211 27.7 [6.4] 1 39.4 [2.2] 1 132/210 (63)

220 27.1 [6.1] 3 39.2 [2.5] 3 134/217 (62)

135 26.4 [6.4]

129 27.3 [5.7]

40.2 [1.8]

40.2 [2.0]

45 (33)

48 (37)

346 27.2 [6.4] 1 39.7 [2.1] 1 177/345 (51)

349 27.2 [5.9] 3 39.6 [2.4] 3 182/346 (53)

4 55 14 4 20 38 0

10 (8) 48 (37) 9 (7) 3 (2) 22 (17) 37 (29) 0 (0)

21 (6) 143 (41) 62 (18) 16 (5) 56 (16) 47 (14) 1 (0)

32 (9) 145 (42) 48 (14) 13 (4) 63 (18) 46 (13) 2 (1)

270 (78) 5 {4-6} 1 60 (17) 4 (1)

269 (77) 5 {4-6} 3 65 (19) 5 (1)

17 (8) 88 (42) 48 (23) 12 (6) 36 (17) 9 (4) 1 (0)

22 (10) 97 (44) 39 (18) 10 (5) 41 (19) 9 (4) 2 (1)

145 (69) 5 {4-7} 1 36 (17) 3 (1)

146 (66) 5 {4-7} 3 41 (19) 5 (2)

(3) (41) (10) (3) (15) (28) (0)

125 (93) 4 {3-5}

123 (95) 4 {3-5}

24 (18) 1 (1)

24 (19) 0 (0)

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 952±959

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 952±959

a

These data collected in Liverpool only.

No. randomised Primary outcome Vaginal delivery not achieved ,24h Secondary outcomes 1. Labour: Cervical score unchanged after 12h a Randomisation to labour Interval (min) Missing Amniotomy Oxytocin augmentation Misoprostol augmentation Fetal blood sampling Vaginal bleeding Uterine activity Tachysystole Hypersystole Hyperstimulation with FHR changes Epidural or opioid 2. Delivery Mode of delivery CS Instrumental delivery Indication for operative delivery Delay Fetal distress Other Randomisation to delivery interval Missing

220 77/217 (35)

547 [470] 7 37/217 (17) 47/217 (22) 0 0 6/217 (3) 14/205 (7) 1/205 (0) 9/205 (4) 67/217 (31) 46/218 (21) 6/218 (3) 28/217 (13) 20/217 (9) 3/217 (1) 825 {450-1498} 3

78 (37)

819 [524] 4 44/210 (21) 13/210 (6) 22/210 (10) 2/210 (1) 5/210 (2) 14/193 (7) 1/193 (1) 10/193 (5) 57/210 (27) 29/210 (14) 5/210 (2) 14/210 (7) 15/210 (7) 5/210 (2) 980 {618-1634} 1

Dinoprostone

211

Misoprostol

South Africa

17 (13) 17 (13) 11 (8) 1125 {734-1680} 0

25 (19) 19 (14)

15 (12) 20 (16) 9 (7) 954 {589-1448} 0

22 (17) 22 (17)

4 (3) 0 (0) 1 (1) 113 (88)

76 (59) 67 (52) 1 (1) 17 (13) 16 (12)

59 (44) 37 (27) 36 (27) 15 (11) 14 (10) 9 (7) 1 (1) 3 (2) 117 (87)

782 [1599]

12/84 (14)

17/105 (16) 784 [533]

46 (36)

129

Dinoprostone

55 (41)

135

Misoprostol

Liverpool

Table 2. Outcomes expressed as n (%), mean [SD] or median {interquartile range}. FHR ˆ fetal heart rate; CS ˆ caesarean section.

31/345 (9) 32/345 (9) 16/345 (5) 1030 {673-1646} 1

54/345 (16) 24/345 (7)

23/328 (7) 2/328 (1) 13/328 (4) 174/345 (50)

805 [527] 4 103/345 (30) 50/345 (14) 58/345 (17) 17/345 (5) 19/345 (6)

17/105 (16)

133 (38)

346

Misoprostol

43/346 (12) 40/346 (12) 12/346 (3) 855 {505-1466} 3

68/347 (20) 28/347 (8)

18/334 (5) 1/334 (0) 10/334 (3) 180/346 (52)

636 [1053] 7 113/346 (33) 114/346 (33) 1/346 (0) 17/346 (5) 22/346 (6)

12/84 (14)

123/346 (36)

349

Dinoprostone

Combined

0.72 (0.47-1.12) 0.80 (0.52-1.25) 1.34 (0.64-2.78) 115 (-26.8-257)

0.80 (0.58-1.11) 0.86 (0.51-1.46)

1.30 (0.72-2.37) 2.04 (0.19±22.4) 1.32 (0.59-2.98) 0.97 (0.84-1.12)

0.91 (0.73-1.14) 0.44 (0.33-0.59) 58.2 (8.1-417.6) 1.00 (0.52-1.93) 0.87 (0.48-1.57)

169 (44-294)

1.13 (0.57-2.24)

1.08 (0.89-1.31)

Combined RR (95% CI) or Mean difference (95%CI)

TITRATED ORAL MISOPROSTOL FOR INDUCTION OF LABOUR 955

956 G. J. HOFMEYR ET AL. Table 3. Maternal side effects and complications expressed as n (%), mean [SD] or median {interquartile range}. South Africa

No. randomised Side effects Mild shivering Severe shivering Vomiting Any side effects Blood loss at delivery .500ml Maternal complications Ruptured uterus Sepsis Pyrexia .38C Retained placenta Other Hospital stay .5 days

Liverpool

Combined

Combined RR (95% CI) or

Misoprostol Dinoprostone Misoprostol Dinoprostone Misoprostol

Dinoprostone

211

346

349

220

135

129

Mean difference (95%CI)

76/210 (36) 2/210 (1) 35/210 (17) 94/210 (45) 51/210 (24)

74/217 (34) 3/217 (1) 32/217 (15) 97/217 (45) 51/218 (23)

4 (3) 0 (0) 7 (5) 20 (15) 18 (13)

0 (0) 1 (1) 2 (2) 4 (3) 20 (16)

80/345 (23) 2/345 (1) 42/345 (12) 114/345 (33) 69/345 (20)

74/346 (21) 4/346 (1) 34/346 (10) 101/346 (29) 71/347 (20)

1.08 (0.82±1.43) 0.50 (0.09±2.72) 1.24 (0.81±1.90) 1.13 (0.91±1.41) 0.98 (0.73±1.31)

0 0 1/210 (0) 1/210 (0) 6/210 (3) 2/210 (1)

0 0 2/218 (1) 2/218 (1) 9/218 (4) 1/218 (0)

0 0 7 (5) 5 (4) 8 (6) 16 (12)

0 0 1 (1) 0 4 (3) 13 (10)

0 0 8/345 (2) 6/345 (2) 14/345 (4) 18/345 (5)

0 0 3/347 (1) 2/347 (1) 13/347 (4) 14/347 (4)

2.68 (0.72±10.03) 3.02 (0.61±14.85) 1.08 (0.52±2.27) 1.29 (0.65±2.56)

trial. The main outcomes by membrane and cervical status are shown in Table 5. Kaplan Meier survival curves showing time from trial entry to delivery in the pre-speci®ed subgroups show a statistically signi®cant earlier delivery in the dinoprostone group for women with an unfavourable cervix and intact membranes (Fig. 2). For women with an unfavourable cervix and ruptured membranes, delivery occurred earlier in the misoprostol group (Fig. 3), although this may have been due to a greater number of caesarean sections (Table 5). When the cervix was favourable there was no statistically signi®cant difference between misoprostol and dinoprostone in time to delivery, regardless of membrane status (Figs. 4 and 5).

were primiparous. Eighteen percent of women had intact membranes and a favourable cervix (55 primiparous and 71 multiparous women); 9% of women had ruptured membranes and an unfavourable cervix (40 primiparous and 22 multiparous women) while 9% of women had ruptured membranes and a favourable cervix (24 primiparous and 38 multiparous women). All demographic and other characteristics at trial entry were balanced between the groups and between the sites. Of the women in the oral misoprostol group, 38% did not achieve vaginal delivery within 24 hour compared with 36% after vaginal dinoprostone (RR 1.08; 95% CI 0.89-1.31) (Table 2). There were fewer caesarean sections, a longer time to delivery and a greater incidence of uterine tachysystole with misoprostol, but differences were not statistically signi®cant. There were no differences in clinically important uterine hyperstimulation, maternal complications (Table 3) or neonatal outcomes (Table 4). In particular there was no increased incidence of shivering with the doses of misoprostol used in this

DISCUSSION Small, frequent, titrated oral doses of misoprostol were successful in avoiding the signi®cant increase in uterine

Table 4. Neonatal outcomes expressed as n (%), mean [SD]. South Africa

Liverpool

Combined

Misoprostol

Dinoprostone

Misoprostol

Dinoprostone

Misoprostol

Dinoprostone

No. randomised 211 Birthweight 3101 [499] Missing 1 Apgar ,7 at 5 mins 7 (3) Not known 1 (0) NICU admission 4/210 (2) Neonatal complications Perinatal death 1/210 (0) Seizures 1/210 (0) Sepsis 0 Others 8/210 (4)

220 3045 [547] 3 13 (6) 2 (1) 6/217 (3)

135 3470 [563]

129 3440 [659]

4 (3) 0 (0) 5 (4)

2 (2) 0 (0) 8 (6)

346 3246 [554] 1 11 (3) 1 (0) 9/345 (3)

349 3192 [620] 3 15 (4) 3 (1) 14/346 (4)

0 0 1 (1) 10 (7)

0 0 0 9 (7)

1/345 (0) 1/345 (0) 1/345 (0) 18/345 (5)

1/346 (0) 0 0 15/346 (4)

1/218 (0) 0 0 6/218 (3)

Combined RR (95% CI) or mean difference (95%CI)

54 (-33.8±142) 0.74 (0.34±1.58) 0.64 (0.28±1.47) 1.00 (0.06±15.97) 1.20 (0.62±2.35)

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 952±959

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 952±959

No. Randomised Vaginal delivery not achieved , 24h CS Uterine hyperstimulation with FHR changes

220 90/219 (41)

48 (22) 9/212 (4)

37 (17) 6/211 (3)

Dinoprostone

223 104 (47)

Misoprostol

0.76 (0.46±1.26) 0.67 (0.18±2.55)

1.13 (0.86±1.5)

RR (99% CI)

Membranes intact, cervix unfavourable

6 (10) 4/59 (7)

62 15 (24)

Misoprostol

14/64 (22) 0

66 17/64 (27)

Dinoprostone

0.44 (0.14±1.43)

0.91 (0.41±2.01)

RR (99% CI)

Membranes intact, cervix favourable

9 (31) 1 (3)

29 11 (38)

Misoprostol

3 (9) 0

33 11 (33)

Dinoprostone

3.41 (0.70±16.7)

1.14 (0.47±2.75)

RR (99% CI)

Membranes ruptured, cervix unfavourable

Table 5. Subgroup analysis by membrane and cervical status expressed as n (%) and relative risks (RR) and 99% CI.

2/31 (6) 2/29 (7)

32 3 (9)

Misoprostol

3 (10) 1/29 (3)

30 5 (17)

Dinoprostone

0.65 (0.07±6.19) 2.0 (0.09±43.8)

0.56 (0.10±3.29)

RR (99% CI)

Membranes ruptured, cervix favourable

TITRATED ORAL MISOPROSTOL FOR INDUCTION OF LABOUR 957

958 G. J. HOFMEYR ET AL.

Fig. 2. Kaplan-Meier plot of time to delivery: membranes intact, unfavourable cervix (log rank test, P ˆ 0.001).

Fig. 4. Kaplan-Meier plot of time to delivery: membranes intact, cervix favourable (log rank test, P ˆ 0.270).

hyperstimulation reported in previous studies, at the expense of a somewhat slower response in women with intact membranes and unfavourable cervices. A possible reason for the slower response in these women is the lack of a local effect on the cervix when misoprostol is administered orally rather than vaginally. However, there is a ®ne balance between the rapid response of vaginally administered misoprostol and the increased risk of uterine hyperstimulation in dosages exceeding 25mg fourhourly. More `aggressive' regimens will undoubtedly result in shorter labours. This may appeal to women and clinicians only if safety for the mother and the baby is not compromised. The external validity of our ®ndings is enhanced by the fact that sites with different patient pro®les were included and the results were similar between sites; and the fact that

the great majority of women who met the entry criteria agreed to participate (exact numbers not recorded). In our study women and clinicians were not blind to the allocated treatment. Therefore, events such as diagnosis of excessive uterine activity, decision to start syntocinon or to perform caesarean section were susceptible to bias. Such bias might operate in either direction, for example, early recourse to caesarean section because of anxiety about an experimental therapy, or delayed caesarean section because of enthusiasm for the new therapy. The possibility of uterine rupture as a rare complication of induction of labour with misoprostol must be considered. A case of uterine rupture in a nulliparous woman following a single 100mg vaginal dose of misoprostol has been reported 18. One trial of vaginal misoprostol for induction of labour in women with a

Fig. 3. Kaplan-Meier plot of time to delivery: ruptured membranes, unfavourable cervix (log rank test, P ˆ 0.015).

Fig. 5. Kaplan-Meier plot of time to delivery: ruptured membranes, favourable cervix (log rank test, P ˆ 0.217).

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 952±959

TITRATED ORAL MISOPROSTOL FOR INDUCTION OF LABOUR 959

previous caesarean section has been stopped prematurely because of dehiscence of the uterine incision in two of the ®rst 17 misoprostol treated women 19. The dosage of vaginal misoprostol used was conservative (25mg 6-hourly to a maximum of four doses). Two and three doses were used, respectively. A possible solution to the problem of hyperstimulation is the use of a single, small vaginal dose to achieve the local effect in women with unfavourable cervices, followed by frequent, small titrated oral doses to `®ne-tune' the uterine response and minimise the risk of hyperstimulation. The titrated oral regimen used in this study was tolerated well by women and staff. However, a requirement by pharmacists to provide fresh solution every two hours at the Liverpool site proved to be cumbersome and time-consuming. Currently unavailable small dose tablets would make such a regimen much more appealing. Whatever misoprostol regimen is tested in the future it should be assessed rigorously before it can be recommended for routine clinical practice. We believe that more randomised trials of various misoprostol regimens are needed. Ideally, they should be double-blinded. The real challenge ahead is to design suf®ciently large induction of labour trials which will be acceptable to women and clinicians and will have the power to address the issues of safety, as well as effectiveness, in an unbiased fashion. Acknowledgements The authors would like to thank the nursing staff, midwives and women of Coronation, Chris Hani Baragwanath and Liverpool Women's Hospitals, and Professor C. J. van Gelderen of the Chris Hani Baragwanath Hospital. Ms S. Ayers provided statistical support. Elizabeth Campbell was funded by University of Liverpool and Liverpool Women's Hospital NHS Trust. References 1. Department of Health. NHS Maternity Statistics, England: 1989-90 to 1994-5, London: HMSO, Statistics bulletin 1997/28, 1997:1±44. 2. Goldberg AB, Greenberg MB, Darney PD. Misoprostol and pregnancy. N Engl J Med 2001;344:38±47.

q RCOG 2001 Br J Obstet Gynaecol 108, pp. 952±959

3. Zieman M, Fong SK, Benowitz NL, Banskter D, Darney PD. Absorption kinetics of misoprostol with oral or vaginal administration. Obstet Gynecol 1997;90:88±92. 4. Tessier F, Dansereau J. A double-blind randomised controlled trial comparing oral misoprostol to vaginal prostaglandin E2 gel for the induction of labour at or near term. Am J Obstet Gynecol 1997;176:S111. 5. Windrim R, Bennett K, Mundle W, Young DC. Oral administration of misoprostol for labor induction: a randomized controlled trial. Obstet Gynecol 1997;89:392±397. 6. Adair CD, Weeks JW, Barrilleaux PS, Philibert L, Edwards MS, Lewis DF. Labor induction with oral versus vaginal misoprostol: a randomised, double-blind trial. Am J Obstet Gynecol 1998;178:S93. 7. Kwon JS, Davies GA, MacKenzie VP. A comparison of oral and vaginal misoprostol for induction of labour at term: a randomised trial. Br J Obstet Gynaecol 2001;108:23±26. 8. Wing DA, Ham D, Paul RH. A comparison of orally administered misoprostol with vaginally administered misoprostol for cervical ripening and labor induction. Am J Obstet Gynecol 1999;180:1155±1160. 9. Toppozada MK, Anwar MYM, Hassan HA, El-Gazaerly WS. Oral or vaginal misoprostol for induction of labour. Int J Gynecol Obstet 1997;56:135±139. 10. Kadanali S, Kucukozkan T, Zor N, Kumtepe Y. Comparison of labor induction with misoprostol vs. oxytocin/prostaglandin E2 in term pregnancy. Int J Gynaecol Obstet 1996;55:99±104. 11. Hale RW, Zinberg S. Misoprostol and pregnancy. N Engl J Med 2001;344:59±60. 12. Friedman MA. Manufacturer's warning regarding unapproved uses of misoprostol. N Engl J Med 2001;344:61. 13. Al®revic Z, Howarth G, Gausmann A. Oral misoprostol for induction of labour with a viable fetus, In: The Cochrane Library, Issue 3. Oxford: Update Software, 1999. 14. Hofmeyr GJ, Gulmezoglu AM, Al®revic Z. Misoprostol for induction of labour: a systematic review. Br J Obstet Gynaecol 1999;106:798± 803. 15. Day L, Tailor A, Howard R, Burrell S. Misoprostol for induction of labour at term; Hofmeyr GJ, Gulmezoglu AM, Al®revic Z. Author's reply. Br J Obstet Gynaecol 2000;107:576±577. 16. Hofmeyr GJ, Matonhodze BB, Al®revic Z, Campbell E, de Jager M, Nikodem C. Titrated oral misoprostol solution: a new method of induction of labour. S Afr Med J 2001. In press. 17. Bishop EH. Pelvic scoring for elective induction. Obstet Gynecol 1964;24:266±268. 18. Hofmeyr GJ, Gulmezoglu AM. Vaginal misoprostol for cervical ripening and induction of labour in late pregnancy, In The Cochrane Library, Issue 2. Oxford: Update Software, 2001. 19. Wing DA, Lovett K, Paul RH. Disruption of prior uterine incision following misoprostol for labor induction in women with previous cesarean delivery. Obstet Gynecol 1998;91:828±830. Accepted 6 June 2001