Intravaginal prostaglandins (PGE2 and PGF2a)
Our search identified one Cochrane systematic review of vaginal prostaglandin E2 (PGE2) or F2α (PGF2α) [14]. Within this review, 37 studies compared PGE2 with placebo. Of these, two trials with 384 women addressed the primary outcome of achieving vaginal delivery within 24 hours. These studies demonstrated that PGE2 reduced failure to achieve vaginal delivery within 24 hours compared with placebo (36/199 versus 183/185; Relative Risk [RR] 0.19, 95% Confidence Interval [CI] 0.14 to 0.25; NNT = 2). However, there was significant between-study heterogeneity in these two included studies, (P < 0.0001), possibly resulting from differences in baseline characteristics of included women and differences in dosing regimens studies [14].
Thirty-four trials with 6399 women compared rates of caesarean section and demonstrated similar rates between PGE2 and placebo groups. Fourteen trials including 1259 women reported that uterine hyperstimulation with FHR changes was increased with vaginal PGE2 compared with placebo (28/642 versus 3/617; RR 4.14, 95% CI 1.93 to 8.90; NNH = 65). Additionally, 13 trials with 3636 women demonstrated that hyperstimulation without FHR changes was also increased (26/1846 versus 7/1790; RR 2.48, 95% CI 1.17 to5.26; NNH = 174). Insufficient data prohibited any conclusions about serious maternal or neonatal morbidity or death [14]. Three trials with 387 women compared PGF2α with placebo. PGF2α reduced the need for oxytocin augmentation (2 trials, 122 women, 41/76 versus 41/46; RR 0.65, 95% CI 0.53 to 0.81, NNT = 4). PGF2α reduced the risk for instrumental vaginal delivery in 2 trials 355 women, 51 of 225 versus 48 of 130, RR 0.60, 95% CI 0.43 to 0.84, NNT = 7) and for epidural analgesia in 3 trials with 387 women (53/241 versus 47 of 146, RR 0.72, 95% CI 0.53 to 0.98, NNT = 17) compared with placebo. These trials did not demonstrate a difference in cesarean section rates or any other outcomes of interest between PGF2α and placebo [14].
Unfavorable cervix subgroup
The authors conducted subgroup analyses of trial participants who had cervices unfavorable for induction. There was no difference in the risk of caesarean delivery between vaginal PGE2 and placebo in the subgroup of women with unfavorable cervices (22 trials, 2173 women, 225/1093 versus 254/1080, RR 0.87, 95% CI 0.75 to 1.02) [14].
Summary: Compared with placebo, vaginal PGE2 increases vaginal delivery rates within 24 hours. However, overall risk of cesarean section was not changed. PGE2 increases uterine hyperstimulation with FHR changes.
Cervical PGE2
Our search identified one Cochrane systematic review of the use of intracervical prostaglandins for cervical ripening and induction of labour compared with placebo/no treatment [15]. This Cochrane review included 28 trials with 3764 women that compared intracervical PGE2 with placebo/no treatment. Four of the studies (n = 198) found that use of cervical PGE2 was superior to placebo in decreasing the number of women who did not achieve vaginal delivery within 24 hours (44/100 versus 71/98; RR 0.61, 95% CI 0.47 to 0.79; NNT = 4). In 27 trials including 3734 women, there was a non-significant trend toward decreased risk of caesarean section for women receiving cervical PGE2 (344/1941 versus 360/1793; RR 0.88; 95% CI 0.77 to 1.01). There were no significant increases in risk of hyperstimulation with FHR changes. However, 11 trials with 2531 women demonstrated significant increases in hyperstimulation without FHR changes (67/1344 versus 37/1187; RR 1.59, 95% CI 1.09 to 2.33; NNH = 55). Serious maternal and neonatal morbidity and mortality were infrequently reported and available data revealed similar findings in the PGE2 and placebo groups [15].
The authors identified 29 trials including 3881 women that compared cervical PGE2 with vaginal PGE2. Cervical PGE2 was less effective than vaginal PGE2 in achieving vaginal delivery within 24 hours (11 trials, 2200 women, 410/1122 versus 315/1078; RR 1.26; 95% CI 1.12 to1.41; NNH = 14). There was no difference in any other outcome of interest [15].
Subgroup with unfavorable cervix
Compared with placebo, there was a non-significant trend toward fewer caesarean sections among women receiving cervical PGE2 (27 studies, 3716 women, 343/1931 versus 359/1785, RR 0.88; 95% CI, 0.77 to 1.01). In 26 trials with 3586 women whose cervix were unfavorable for induction, there was no difference in caesarean deliveries between women receiving intracervical and intravaginal PGE2 [15].
Summary: Intracervical PGE2 appears more effective than placebo in achieving vaginal delivery within 24 hours.
Oxytocin
Our search identified one Cochrane systematic review which included 61 trials with 12,819 women and evaluated oxytocin for induction of labour [16]. Comparisons were made between intravenous (IV) oxytocin versus placebo/expectant management (25 trials, 6660 women), IV oxytocin versus vaginal prostaglandin (PGE2) (27 trials, 4564 women), IV oxytocin versus intracervical prostaglandins (PGE2) (14 trials, 1331 women), and IV oxytocin versus vaginal PGF2α (3 trials, 291 women) [16].
Three trials including 399 women reported that IV oxytocin, when compared with expectant management, reduced failure to achieve vaginal delivery within 24 hours (16/191 versus 112/208; RR 0.16, 95% CI 0.10 to 0.25; NNT = 3). Meta-analysis of 24 trials including 6620 women found a small but statistically significant increased rate of caesarean delivery for women in the oxytocin group (339/3267 versus 301/3353; RR 1.17, 95% CI 1.01 to1.35; NNH = 66). There was no significant difference in uterine hyperstimulation with or without FHR changes. Use of oxytocin significantly reduced chorioamnionitis (14 studies, 5514 women 144/2720 versus 213/2795; RR 0.69, 95% CI 0.57 to 0.85; NNT = 40); however there was significant heterogeneity among the included trials for this comparison, (I2 = 65%, P = 0.001) and the authors' analysis of the studies included in this comparison using the random effects method was not statistically significant. Likewise, NICU admissions were reduced by oxytocin compared to placebo or expectant management, (7 studies, 4387 women, 264/2196 versus 333/2191; RR 0.79, 95% CI 0.68 to 0.92, NNT = 32). However, there was significant between study heterogeneity for this comparison (I2 = 70%, P = 0.0003) and this result was no longer statistically significant when the random effects method was used for analysis. The majority of the studies included in these comparisons required ruptured membranes for entry, likely influencing this result. Data were insufficient to establish conclusions regarding neonatal and maternal mortality or serious morbidity [16].
Three trials including 260 women reported that oxytocin was associated with more failures to achieve vaginal delivery within 24 hours than vaginal PGE2 (73/132 versus 40/128; RR1.77, 95% CI 1.31 to 2.38; NNH = 5). When comparing oxytocin with vaginal PGE2, there was no significant difference in the rates of caesarean section (26 trials, 4514 women, 274/2259 versus 246/2255; RR 1.11, 95% CI 0.94 to 1.30). The incidence of uterine hyperstimulation with fetal heart rate (FHR) changes was very low and not different between groups. Fewer women receiving oxytocin developed chorioamnionitis than those receiving vaginal PGE2 (4 trials, 2742 women, 54/1381 versus 81/1361; RR 0.66, 95% CI 0.47 to 0.92, NNT = 50). Data were insufficient to draw conclusions regarding neonatal and maternal mortality or morbidity; based on limited data, there were no differences between groups [16].
Two studies that included 258 women comparing oxytocin with intracervical PGE2 found that oxytocin was associated with more failure to achieve vaginal deliveries within 24 hours (63/125 versus 46/133; RR 1.47, 95% CI 1.10 to 1.96; NNH = 7). Oxytocin was associated with more caesarean deliveries than intracervical PGE2 (14 studies, 1331 women, 123/643 versus 94/688; RR 1.37, 95% CI 1.08 to 1.74; NNH = 20). There was no significant difference in uterine hyperstimulation with FHR changes. There were not enough data to develop conclusions regarding neonatal and maternal mortality/morbidity [16].
There were only three trials with 291 women that compared oxytocin with PGF2α. None reported on the number of women failing to deliver vaginally within 24 hours. There were no significant differences in uterine hyperstimulation with FHR changes (one trial 23 women) or rates of caesarean delivery (3 trials 280 women). There were no cases of serious neonatal morbidity or perinatal deaths in the two studies that reported this outcome [16].
Unfavorable cervix subgroup
Compared with placebo or expectant management, there was no difference in caesarean deliveries among participants with unfavorable cervices (13 trials, 1366 women). Similarly, there was no difference in caesarean deliveries among 1041 women in 15 trials with unfavorable cervices who received oxytocin or vaginal PGE2. However, oxytocin use was more likely to result in caesarean delivery than intracervical PGE2 (10 trials, 1003 women, 107/477 versus 79/526, RR 1.44, 95% CI, 1.12 to 1.86, NNH = 16) [16].
Randomised controlled trials published after the search date of systematic reviews
Our search identified one study including 240 women that compared oxytocin with vaginal prostaglandin E2 for premature rupture of membranes at term [17]. In this study, oxytocin was associated with a significantly shorter time from induction to delivery (3.4 +/- 1.5 versus 9.6 +/- 4.7 hours; p = 0.02). There was no difference in the risk of caesarean section [17].
Summary: Oxytocin is more effective than expectant management or placebo but less effective than vaginal and cervical PGE2 in bringing about vaginal delivery within 24 hours. Oxytocin resulted in more caesarean deliveries than cervical PGE2.
Amniotomy
We identified one Cochrane systematic review of amniotomy for induction of labour [18]. This review included two studies with 310 total participants. One included study compared women receiving amniotomy with those receiving either oxytocin alone or no intervention. This study was underpowered to detect differences in any outcome of interest and the review concluded that no meaningful results could be drawn from these comparisons. The second included study compared amniotomy alone to a single dose of vaginal prostaglandins for women with a favourable cervix and found a significant increase in the need for oxytocin augmentation in the amniotomy alone group compared with the women receiving PGE2 (260 women, 57/130 versus 20/130; RR 2.85, 95% CI 1.82 to4.46; NNH = 3). There was no difference in caesarean deliveries [18].
Subgroup with unfavorable cervix
There were no studies that included participants with unfavorable cervices [18].
Summary: Compared with vaginal PGE2, amniotomy increases the need for oxytocin augmentation.
Oxytocin with amniotomy
Our search identified one Cochrane systematic review including 17 trials with 2566 women comparing IV oxytocin plus amniotomy with other methods for induction of labour [19]. This review compared amniotomy plus oxytocin (in varying doses), with placebo, vaginal prostaglandin E2 or F2α, or amniotomy alone. Oxytocin plus amniotomy resulted in fewer cases of meconium stained amniotic fluid than placebo or no treatment (one trial, 184 participants, 3/92 versus 13/92; RR 0.23, 95% CI 0.07 to 0.78; NNT = 9). There were no other significant differences in our outcomes of interest for this comparison [19].
When compared with vaginal prostaglandins, amniotomy plus IV oxytocin was associated with more postpartum hemorrhage (2 studies, 160 women, 11/80 versus 2/80; RR 5.5, CI 1.26 to 24.07; NNH = 9). One RCT of 100 subjects found that more women were dissatisfied with amniotomy and IV oxytocin than vaginal prostaglandins, (26/50 versus 0/50; RR 53, CI 3.32 to 846.51; NNH = 1). There were no other significant differences between oxytocin plus amniotomy and vaginal prostaglandins [19].
One study with 30 participants compared oxytocin plus amniotomy with cervical prostaglandins. This study was too small to detect any differences in outcomes of interest. Likewise, only two studies with 309 total participants compared oxytocin plus amniotomy with oxytocin alone. These studies were also underpowered to detect differences in any outcome of interest [19].
When compared with those who received amniotomy alone, fewer women who received amniotomy plus IV oxytocin were not delivered vaginally at 24 hours (2 studies, 296 participants, 3/148 versus 24/148; RR 0.13, 95% CI 0.04 to 0.41; NNT = 8). Amniotomy plus IV oxytocin also resulted in significantly fewer instrumental vaginal deliveries than amniotomy alone (2 studies, 510 participants, 57/255 versus 88/255; RR 0.65, CI 0.49 to 0.85; NNT = 995% CI 6 to 20) [19].
Unfavorable cervix subgroup
The review included two trials with 106 women who had cervices unfavorable for induction of labor. There was no difference in caesarean birth among women allocated to amniotomy and oxytocin versus vaginal prostaglandins. There were no other trials including women with cervices unfavorable for induction [19].
Summary: Oxytocin plus amniotomy is more effective than amniotomy alone in achieving vaginal delivery within 24 hours. Oxytocin plus amniotomy may be associated with more postpartum hemorrhage and less maternal satisfaction than vaginal prostaglandins.
Vaginal Misoprostol
The Cochrane review of vaginal misoprostol for labour induction included 121 trials [20]. There were no significant difference in vaginal deliveries not achieved with 24 hours among five trials with 769 women that compared vaginal misoprostol with placebo/no treatment. Likewise, in five trials with 777 women, there were no significant differences in hyperstimulation with FHR changes. Compared with placebo/no treatment, vaginal misoprostol was associated with more hyperstimulation without FHR changes (31/313 versus 10/481, 6 trials, 794 women, RR 3.52, 95% CI 1.78 to 6.99, NNH = 19) but with less meconium stained amniotic fluid (6 trials, 814 participants, 27/326 versus 83/488, RR 0.56, 95% CI 0.35 to 0.87, NNT = 14) Vaginal misoprostol reduced the number of participants with a cervix unfavorable or unchanged after 12 to 24 hours (2 studies 107 women, 4/56 versus 41/51, RR 0.10, 95% CI 0.01 to 0.64, NNT = 2). There was no difference in caesarean deliveries in 10 trials that included 1141 women.
Twenty-two trials with 5,229 participants compared vaginal misoprostol with other vaginal prostaglandins for the outcome of vaginal deliveries within 24 hours. Women receiving misoprostol were less likely to not be delivered within 24 hours (22 trials 5229 participants, 920/2550 versus 1179/2679, RR 0.77, 95% CI 0.66 to 0.89, NNT = 10) and were less likely to require oxytocin augmentation (38 trials, 7022 participants, 1355/3465 versus 1794/3557, RR 0.68, 95% CI 0.61 to 0.76, NNT = 7). Meconium-stained amniotic fluid was more common among subjects receiving misoprostol (18 trials, 3991 women, 246/1909 versus 190/2082, RR 1.35, 95% CI 1.13 to 1.61, NNH = 32). Misoprostol increased uterine hyperstimulation without FHR changes (26 trials 4804 women 381/2311 versus 199/2493, RR 1.99, 95% CI 1.41 to 2.79, NNH = 13), although hyperstimulation with FHR changes did not differ (31 trials 5830 women). Vaginal misoprostol reduced the need for oxytocin augmentation (38 trials, 7022 women, 1355/3465 versus 1794/3557, RR 0.68, 95% CI 0.61 to 0.76, NNT = 7) and epidural anesthesia (8 trials, 2141 women, 469/1063 versus 516/1078, RR 0.92 95% CI 0.85 to 0.99, NNH = 27). Caesarean section rates were not significantly different [20].
Compared with cervical PGE2, vaginal misoprostol reduced failure to achieve vaginal delivery within 24 hours (13 trials, 1627 women, 253/814 versus 402/813, RR 0.63, 95% CI 0.56 to 0.71, NNT = 6). Oxytocin augmentation was required less often with misoprostol based on 20 trials including 2316 women, (411/1177 versus 727/1139, RR 0.55, 95% CI, 0.48 to 0.64 NNT = 4) and women receiving misoprostol were less likely to have a cervix unfavorable for induction after 12-24 hours (1 trial, 155 women, 38/76 versus 58/79, RR 0.68, 95% CI 0.52 to0.88, NNT = 5). Women receiving misoprostol were less likely to require epidural anesthesia (2 trials, 321 women, 48/160 versus 75/161, RR 0.64, 95% CI 0.48 to 0.86, NNT = 6). Misoprostol resulted in more uterine hyperstimulation with FHR changes (20 trials, 2224 women, 98/1129 versus 39/1095, RR 2.32, 95% CI 1.64 to3.28, NNH = 22) and without FHR changes (17 trials 2178 women, 194/1097 versus 96/1081, RR 1.95, 95% CI 1.57 to 2.42, NNH = 12). Increased rates of meconium-stained amniotic fluid were observed with use of misoprostol (14 trials 2018 women, 161/1015 versus 123/1003, RR 1.29, 95% CI 1.04 to1.59, NNH = 29). There were no other statistically significant differences in perinatal or maternal outcomes [20].
Compared with oxytocin, vaginal misoprostol reduced the likelihood of participants not being delivered vaginally within 24 hours (10 trials, 1397 women, 135/690 versus 226/707, RR 0.65 95% CI.47 to 0.90, NNT = 9). Vaginal misoprostol was associated with increased uterine hyperstimulation with FHR changes (9 trials with 1419 women 49/690 versus 28/729, RR 1.87, 95% CI 1.20 to 2.91, NNH = 31) and without FHR changes (15 trials 2050 women, 218/1009 versus 102/1041, RR 2.24, 95% CI 1.82 to 2.77, NNH = 9). Women receiving misoprostol were more likely to experience gastrointestinal side effects than those receiving oxytocin (4 trials 334 women, 15/170 versus 2/164, RR 5.04, 95% CI 1.51 to 16.86, NNH = 21) Caesarean deliveries were less likely among women receiving vaginal misoprostol (25 trials 3074 women, 258/1527 versus 364/1547, RR 0.76 95% CI 0.60 to 0.96, NNT = 18) as were instrumental vaginal deliveries (13 trials, 1639 women, 69/810 versus 96/829, RR 0.74 95% CI 0.56 to 0.99, NNT = 34). Infants born to women receiving misoprostol were less likely to have Apgar score < 7 at 5 minutes of life (13 trials, 1906 women, 22/938 versus 41/968, RR 0.56, 95% CI 0.34 to 0.92, NNT = 54). There were no differences in other maternal or fetal outcomes of interest [20].
Unfavorable cervix subgroup
Compared with placebo, there was no difference in caesarean deliveries for women receiving vaginal misoprostol (7 trials, 862 women). There was no difference in the likelihood of caesarean deliveries among women receiving misoprostol or women receiving vaginal PGE2 (28 trials, 5832 women), cervical PGE2 (21 trials, 2499 women) or oxytocin (14 trials, 1598 women) [20].
Summary: Vaginal misoprostol is more likely to result in vaginal delivery within 24 hours than vaginal or cervical PGE2 or oxytocin but is associated with increased uterine hyperstimulation. Compared with IV oxytocin, vaginal misoprostol may reduce the likelihood of caesarean delivery.
Oral Misoprostol
The Cochrane review of oral misoprostol compared to other methods of labour induction included 56 RCTs with a total of 11,590 participants [21]. There were seven trials with 669 women that compared oral misoprostol to placebo. Women assigned to receive oral misoprostol were more likely to give birth vaginally within 24 hours (1 trial, 96 women, 3/47 versus 20/49; RR 0.16, CI 0.05 to 0.49; NNT = 3). Oral misoprostol was also associated with lower caesarean section rates than placebo (six trials, 629 women, 31/312 versus 51/317; RR 0.61, 95% CI 0.41 to 0.93; NNT = 16). Fewer women receiving oral misoprostol required oxytocin augmentation (6 trials, 535 subjects, 63/266 versus 181/269; RR 0.35, 95% CI 0.28 to 0.44; NNT = 3) [21].
Ten trials with a total of 3368 women compared oral misoprostol with vaginal PGE2. Fewer women assigned to receive oral misoprostol required caesarean delivery (10 trials, 3368 participants, 340/1599 versus 467/1769; RR 0.87, 95% CI 0.77 to 0.98; NNT = 30). In two trials including 930 subjects, more women assigned to oral misoprostol had an unfavourable cervix after 24 hours (74/470 versus 51/460; RR 1.41, 95% CI 1.01 to 1.96; NNH = 22). There were no statistically significant differences between the groups in any of the other outcomes, including hyperstimulation with and without FHR changes and the frequency of meconium-stained amniotic fluid. There was significant heterogeneity (P = 0.002) among studies comparing oral misoprostol and vaginal PGE2 for the outcome of uterine hyperstimulation without FHR changes, likely related to the differing doses of oral misoprostol used in the included studies [21].
Four trials with 681 women compared oral misoprostol to intracervical PGE2. Fewer women assigned to oral misoprostol group failed to deliver vaginally within 24 hours, although this finding was of borderline statistical significance (2 trials, 391 participants, 81/196 versus 100/195; RR 0.81, 95% CI 0.65--1.00; NNT = 11). Oral misoprostol was associated with more uterine hyperstimulation with FHR changes (3 trials, 490 women, 12/245 versus 3/245; RR 3.57, 95% CI 1.11 to 11.54; NNH = 32). There was a trend of borderline significance toward more hyperstimulation without FHR changes with oral misoprostol (1 trial, 190 women, 8/95 versus 0/95; RR 17.01, 95% CI 1.00 to 290.42). There were no differences in any other outcome of interest [21].
Eight trials including 1026 women compared oral misoprostol with IV oxytocin. Meconium staining of the amniotic fluid was seen more frequently in the misoprostol group (6 trials, 916 women, 47/477 versus 24/439; RR 1.72, 95% CI 1.08 to 2.74; NNH = 26), but oral misoprostol was not associated with any other differences in adverse fetal, neonatal or maternal outcomes [21].
Twenty-six trials with 5096 participants compared oral with vaginal misoprostol. The oral route of administration was associated with more frequent use of oxytocin (22 trials, 4557 women, 1301/2279 versus 1151/2278; RR 1.19, 95% CI 1.06 to 1.34; NNH = 11), but there was no difference in vaginal delivery within 24 hours. There were significantly lower rates of uterine hyperstimulation without FHR changes with oral regimens (9 trials, 1420 women, 85/698 versus 146/722; RR 0.58, 95% CI 0.35 to 0.96; NNT = 12), but the rate of hyperstimulation with FHR changes was not different between the two groups. Fewer babies born to mothers who received oral misoprostol had Apgar scores less than 7 at five minutes of life (14 trials, 3270 women 37/1638 versus 57/1632; RR 0.65, 95% CI 0.44 to 0.97; NNT = 82). There was no difference in caesarean deliveries in 25 trials with 5096 women [21].
Unfavorable cervix subgroup
There were no studies comparing oral misoprostol with placebo, vaginal PGE2, intracervical PGE2, or oxytocin that reported on caesarean deliveries among women with unfavorable cervices. Among women with unfavorable cervices who were randomized to receive oral misoprostol or vaginal misoprostol there were no differences in caesarean deliveries among primiparous (2 studies, 85 participants) or multiparous (1 study, 24 participants) subjects [21].
Randomised controlled trials published after the search date of systematic reviews
Subsequent to the search date of the Cochrane review, we identified one study with 87 participants that compared oral misoprostol with placebo [22]. This study found oral misoprostol superior to placebo in achieving delivery within 24 hours (19/43 versus 6/44, P = 0.024, NNT = 4). An additional study that compared oral misoprostol with PGE2 found that more women receiving misoprostol delivered vaginally within 12 hours, although vaginal deliveries within 24 hours did not differ [23]. More women receiving oral misoprostol were satisfied with their induction method [23].
Summary: Oral misoprostol reduced caesarean sections compared with vaginal PGE2 and placebo. Compared with vaginal misoprostol, oral misoprostol is associated with fewer contractile abnormalities, but more need for oxytocin augmentation.
Buccal or sublingual misoprostol
Our search uncovered three systematic reviews comparing sublingual or buccal misoprostol with other methods of labour induction [24–26]. The Cochrane review by Muzonzini[24] and colleagues and the review by Bartusevicius and colleagues[25] both included the same three studies with a total of 507 women and reached similar conclusions. Two of the studies with a total of 350 women compared buccal or sublingual misoprostol (50 μg) to oral misoprostol (50 or 100 μg) and one study with 157 participants compared buccal and vaginal misoprostol. Neither review found significant differences in any outcome of interest [24, 25].
Unfavorable cervix subgroup
The Cochrane reviewers did not conduct any subgroup analyses according to cervical status [24].
Other systematic reviews
In 2008 Souza and colleagues published a systematic review of studies comparing sublingual or buccal misoprostol with vaginal misoprostol for induction of labour. This review included five studies with 740 subjects [26]. The authors found no significant differences in the rates of vaginal delivery not achieved within 24 hours, hyperstimulation, or cesarean deliveries. There were more cases of uterine tachysystole, defined as more than five contractions in 10 minutes for at least 20 minutes, among women assigned to sublingual misoprostol (5 trials, 740 women, 42/368 versus 26/372; OR 1.70, 95% CI 1.02 to 2.83; NNH = 24, 95% CI 10 to 771), although there was significant heterogeneity among studies included in this comparison (P = 0.04) [26].
Randomised controlled trials published after the search date of systematic reviews
Our search identified two additional studies carried out subsequent to the search dates of these systematic reviews. One of these studies compared sublingual misoprostol 50 μg with amniotomy and oxytocin for induction of labour among 50 women at term with favourable cervices [27]. This study was terminated early when an interim analysis revealed that significantly fewer women allocated to sublingual misoprostol delivered within 24 hours (15/22 versus21/21; RR 0.68, 95% CI 0.51 to 0.91; NNH = 3). There were no differences in other maternal or fetal outcomes, although maternal satisfaction was significantly higher with sublingual misoprostol. The second study included 150 women and compared 50 μg misoprosol by oral, vaginal, or sublingual routes [28]. This study found that the induction to delivery interval was significantly decreased among women receiving sublingual misoprostol compared with the vaginal and oral routes (13.3 hours versus 16.1 hours [oral] versus 15.1 hours [vaginal]). Fewer babies born to mothers receiving sublingual misoprostol had Apgar scores less than seven at one minute (0/50 versus 6/100, P = 0.003, NNT = 17) [28].
Summary: Compared with vaginal misoprostol, administration of misoprostol by the buccal or sublingual route increases uterine tachysystole.
Mechanical methods
Our search identified three systematic reviews evaluating mechanical methods for induction of labour. The Cochrane review studied mechanical methods including laminaria tents, synthetic equivalents such as Dilapan, Foley catheters, and other types of balloon catheter for induction of labour. It included 45 RCTs that compared mechanical methods with PGE2, misoprostol, oxytocin, and placebo. Most trials had small sample sizes [29].
The authors did not find any advantage of mechanical methods compared with placebo or no treatment in the pre-specified outcomes of vaginal delivery not achieved within 24 hours or caesarean deliveries (1 trial, 48 women). There was no difference in caesarean deliveries between women receiving mechanical methods and women receiving placebo or no treatment (6 studies, 416 participants). There was no difference in any other outcome of interest, including chorioamnionitis (1 trial, 240 participants) and endometritis (2 trials, 288 participants) [29].
More subjects allocated to mechanical methods failed to deliver vaginally within 24 hours than those assigned to vaginal PGE2 (1 trial, 109 participants, 43/59 versus 21/50, RR 1.74, CI 1.21 to 2.49; NNH = 3), and more women allocated to mechanical methods required oxytocin augmentation (2 trials, 169 women, 30/89 versus 9/80, RR 2.90, 95% CI 1.40 to 6.00, NNH = 5), although this finding should be interpreted with caution, as there was significant heterogeneity between the 2 studies included in this comparison (P = 0.0008). Mechanical methods were less likely to result in uterine hyperstimulation with FHR changes in 6 trials with 484 women, (0/246 versus 14/238, RR 0.14, 95% CI 0.04 to 0.53, NNT = 20) and without FHR changes in 8 trials with 580 women (6/293 vs 28/287, RR 0.26, 95% CI 0.13 to 0.54, NNT = 14). There was no difference in caesarean deliveries between mechanical methods and vaginal PGE2 (12 trials, 786 women), but mechanical methods were associated with reduced need for instrumental vaginal deliveries (5 trials, 378 women, 36/192 versus 53/186, RR 0.65, 95% CI 0.46 to 0.93, NNT = 11) [29].
More women assigned to mechanical methods did not achieve vaginal delivery within 24 hours than those assigned to cervical PGE2 (1 trial, 100 participants, 34/50 versus 20/50; RR 1.70, CI 1.15 to 2.50; NNH = 3), and more women allocated to mechanical methods required oxytocin augmentation (1 trial, 185 women, 84/90 versus 63/95, RR 1.41, 95% CI 1.21 to 1.64, NNH = 3). However, there was no difference in caesarean sections in 12 trials that included 1614 women. Compared with cervical PGE2, mechanical methods were associated with less endometritis (4 trials, 693 participants, 9/352 versus 34/341 RR 0.26, 95% CI 0.13 to 0.52, NNT = 14). However, in 3 studies with 619 women that compared mechanical methods to cervical PGE2, there were more neonatal infections in babies born to mothers who had received mechanical methods compared to cervical PGE2 (24/316 versus 9/303, RR 2.45 95% CI 1.18, to 5.07, NNH = 24) [29].
Analysis of four studies with 198 women comparing mechanical methods with oxytocin found that mechanical methods resulted in fewer caesarean deliveries (18/103 versus 30/95; RR 0.55, 95% CI 0.33 to 0.91; NNT = 8). There was no difference in hyperstimulation without FHR changes, postpartum hemorrhage, or serious maternal morbidity or death in 1 trial with 60 women. No other outcomes could be evaluated for this comparison [29].
Four studies including 618 women compared mechanical methods to vaginal misoprostol. There were no statistical differences in the likelihood of achieving vaginal delivery within 24 hours (2 studies, 234 women) or in caesarean deliveries (4 studies, 618 women). There was reduced risk for uterine hyperstimulation with FHR changes seen in three trials including 434 women comparing mechanical methods to vaginal misoprostol (8/226 versus 19/208; RR 0.41, CI 0.20 to 0.87; NNT = 19). There were no differences noted in infectious morbidity or neonatal outcomes [29].
Unfavorable cervix subgroup
There was no difference in caesarean deliveries among 396 women enrolled in five randomized controlled trials comparing mechanical methods with placebo or no treatment, in 10 trials with 738 women comparing mechanical methods with vaginal PGE2 or in 12 trials with 1614 women comparing mechanical methods with intracervical PGE2. There was no difference in cesarean sections in three trials with 482 participants comparing mechanical methods with vaginal misoprostol. In the subgroup of women with unfavorable cervices, mechanical methods were less likely to result in caesarean delivery than oxytocin (3 trials, 178 women, 15/93 versus 27/85, RR 0.50, 95% CI, 0.29 to 0.87, NNT = 7) [29].
Other systematic reviews
Our search identified a second systematic review that compared mechanical methods of labour induction with PGE2, misoprostol, hyaluronidase or placebo [30]. This systematic review, which included 30 randomised, controlled trials with a total of 4468 participants, focused on the outcomes of maternal and neonatal infectious morbidity. The authors defined maternal infectious morbidity as maternal temperature greater than 38°C, endometritis or chorioamnionitis. They defined neonatal infectious morbidity as fever, suspected or proven sepsis, or need for antibiotics. Controls were the pooled group of women who had received other pharmacologic methods of labour induction. Compared with controls, women undergoing labour induction with mechanical methods were more likely to experience infectious morbidity (30 studies, 4468 participants, 252/2220 versus 188/2248; OR 1.38, 95% CI 1.12 to 1.68; NNH = 36). The authors reported no significant heterogeneity for this comparison. Similarly, infants born to mothers undergoing mechanical methods of induction were more likely to experience neonatal infectious morbidity than infants born to mothers undergoing induction with pharmacologic methods (8 trials, 1775 women, 40/893 versus 18/882; OR 2.03, 95% CI 1.19 to 3.51; NNH = 50). The authors reported that there was no significant heterogeneity for this comparison [30].
A third systematic review compared mechanical methods (Foley catheter balloon) with locally applied prostaglandins (vaginal PGE2, cervical PGE2 and vaginal misoprostol) [31]. This systematic review included 27 randomized controlled trials that included 3532 participants. When compared with all locally applied prostaglandins (LAPG) combined, there were no differences between mechanical methods and prostaglandins in caesarean deliveries (27 trial, 3532 participants), participants with cervices that were unfavorable or unchanged after 12 to 24 hours (6 trials, 613 participants), ripening to delivery interval (13 trials, 1270 participants), vaginal deliveries within 12 to 24 hours (13 trials, 1779 women), maternal fevers (19 trials, 2421 women), 5-minute Apgar scores less than 7 (14 trials, 1661 women), meconium staining (13 trials 1841 women), or admission of the neonate to a NICU (12 trials, 1796 women). Women who received LAPG were less likely to require oxytocin augmentation than those receiving mechanical methods (16 trials, 1644 participants, RR 0.73, 95% CI 0.62 to 0.86, P = 0.0002), but were more likely to experience excessive uterine activity, defined as tachysystole, hypertonus, or hyperstimulation syndrome (21 trials, 2661 participants, 244/1306 versus 147/1355, RR, 2.35; 95% CI, 1.41 to 3.90; P = .001, NNH for locally applied prostaglandins when compared with mechanical methods = 7). There was significant heterogeneity noted for the outcomes of excessive uterine activity, vaginal delivery within 12-24 hours, and for need for oxytocin augmentation [31].
The authors conducted subgroup analyses comparing mechanical methods with vaginal PGE2, cervical PGE2, and vaginal misoprostol. They found that mechanical methods were associated with a longer ripening to delivery interval than cervical PGE2 (5 trials, 552 subjects, weighted mean difference [WMD] 5.48 hours, 95% CI 2.79 to 8.16, P < 0.0001) and vaginal PGE2 (2 trials, 118 subjects, WMD,4.55 hours, 95% CI, 0.33 to 8.77; P = .03). Cervical PGE2 was associated with a higher risk for caesarean deliveries than mechanical methods in seven trials with 896 women (OR 1.27, 95% CI 1.01 to 1.59, P = 0.04). Vaginal misoprostol was associated with increased risk for excessive uterine activity compared with Foley balloon in 13 trials with 1847 participants, RR 3.41, 95% CI, 1.97 to 5.90; P = 0.0001) [31].
Randomised controlled trials published after the search date of systematic reviews
We included one additional trial that was published subsequent to the search date of the third systematic review [32]. This trial randomly assigned 240 women to receive vaginal misoprostol 25 μg or a Foley catheter for labor induction. This study found that the Foley catheter was associated with a longer induction to vaginal delivery interval than the vaginal misoprostol (20.2 hours versus 17.3 hours, P = 0.016). There were no significant differences in other outcomes of interest [32].
Summary: Mechanical methods are less likely to result in uterine hyperstimulation than PGE2 or vaginal misoprostol, but may be associated with increased maternal and neonatal infectious morbidity.
Membrane Sweeping
Our search identified one Cochrane systematic review of 22 trials which included 2797 subjects that compared membrane sweeping with oxytocin, PGE2, or no treatment [33]. There was no difference in rates of caesarean deliveries, serious neonatal morbidity, perinatal death, serious maternal or neonatal infections when comparing membrane sweeping with no treatment. A policy of routine membrane sweeping from 37 weeks onward reduced the likelihood of gestation continuing to both 41 (6 studies, 937 women, 77/473 versus 129/464; RR 0.59, 95% CI 0.46 to 0.74; NNT = 9) and 42 (6 studies, 722 women, 12/365 versus 43/357; RR 0.28, 95% CI 0.15 to 0.50; NNT = 12) weeks' gestation. Membrane sweeping was associated with reduced likelihood of not being in labour within 48 hours (five studies, 726 women, 234/367 versus 298/359; RR 0.77, 95% CI 0.70 to 0.84; NNT = 6). Membrane sweeping was also associated with reduced risk for not being delivered within one week (9 studies, 1375 women, 320/695 versus 440/680; RR 0.71, 95% CI 0.65 to 0.78; NNT = 6). Membrane sweeping was associated with more vaginal bleeding (3 trials, 391 women, 35/200 versus 18/191; RR 1.75, 95% CI 1.08 to 2.83; NNH = 15) and more maternal discomfort (2 studies, 320 women, 94/163 versus 32/157; RR 2.83, 95% CI 2.03 to 3.96; NNH = 3) compared with no treatment. Data comparing membrane sweeping with PGE2 and with oxytocin were insufficient to draw conclusions of relative efficacy [33].
Unfavorable cervix subgroup
There was no difference in caesarean sections among women allocated to membrane sweeping versus no treatment (3 trials, 200 women) nor in two trials with 252 women comparing membrane sweeping with vaginal prostaglandins nor in one trial with 69 women comparing membrane sweeping with oxytocin [33].
Randomised controlled trials published after the search date of systematic reviews or awaiting classification
Among the studies awaiting classification in the Cochrane review were five high quality RCTs [34–38]. In addition, we identified one additional high quality RCT that was published after the Cochrane review's search date [39]. Of these, five studies with 1700 participants compared membrane sweeping with no treatment or vaginal exam alone [35–39]. In a meta-analysis that added our independently-extracted data from the two studies [36, 37] that evaluating the effect of membrane sweeping on post-term gestations to the data reported in the Cochrane review, membrane sweeping significantly decreased the number of pregnancies progressing to 42 weeks' gestation (8 studies, 1874 participants, 102/902 versus 194/862, RR 0.53, 95% CI 0.43 to 0.65, NNT = 10). There was no significant heterogeneity for this comparison. Thus the addition of these two new trials did not alter the conclusion reached by the Cochrane review.
DeMiranda found that membrane sweeping significantly reduced the time from randomization to delivery by one day (3.50 versus 4.47 days, mean difference 0.97 days; 95% CI 0.60 to 1.35) [36]. In a more recent study involving 351 women, Yildirim and colleagues found that membrane sweeping significantly increased the likelihood of spontaneous labor by 41 weeks' gestation (162/179 versus 118/167, P = 0.0001) [38]. By contrast, Hamdan found that membrane sweeping did not increase the proportion of women planning trial of labor after prior cesarean section (TOLAC) who entered spontaneous labor [39]. The Hill study was designed to test whether membrane sweeping increases prelabour rupture of membranes, but found no overall difference in this outcome [37]. One study with 60 participants compared membrane sweeping with a single dose of intracervical PGE2 [34]. Use of cervical PGE2 resulted in a significantly shorter intervention to delivery interval than did membrane sweeping (26.23 hours versus 19.15 hours, P < 0.01) [34].
Summary: Membrane sweeping reduces the risk of post-term gestation.