Inclusion and exclusion criteria

We included quantitative studies that explored common indications for IOL, specifically: post-term pregnancy, premature rupture of membranes (PROM), twin pregnancy, antepartum haemorrhage, chorioamnionitis (including suspected), cholestasis of pregnancy, alloimmune disease or Rh disease, intrauterine growth restriction (IUGR), fetal distress, oligohydramnios, fetal gastroschisis, fetal macrosomia, fetal death, chronic/gestational hypertension, preeclampsia, diabetes, maternal age, elevated maternal body mass index (BMI), and other more uncommon maternal obstetric or medical indication (e.g. maternal cardiac disease, maternal melanoma, breast cancer, history of fast labour).

For each indication, we included systematic reviews/meta-analyses, randomised controlled trials (RCTs), prospective and retrospective cohort studies, and case control studies that compared maternal and neonatal outcomes for different modes or timing of birth i.e. IOL versus expectant management (EM); IOL versus immediate birth by caesarean section (CS); IOL at different time points (e.g. at 41 versus 42 weeks); and EM versus expedited birth.

To be included in the review, studies had to report on one or more of the following outcomes of interest: mode of birth, maternal morbidity, and fetal or neonatal morbidity and mortality. Following Mozurkewich et al. [10], maternal morbidity was defined as chorioamnionitis, endometritis, severe perineal trauma, blood transfusion, emergency CS or prolonged hospitalisation. Neonatal morbidity was defined as admission to a neonatal intensive care unit (NICU), 5-minute Apgar score <7, respiratory distress syndrome (RDS), shoulder dystocia, birth injury (as defined by the authors), meningitis, pneumonia, hypoxic ischaemic encephalopathy, meconium aspiration syndrome, or sepsis.

Studies were excluded if they were reported on in a systematic review or meta-analysis already included (as not to double count), where full text was not available or accessible, or if not published in English. Studies that compared different methods of IOL or evaluated IOL outcomes in the absence of a medical indication (e.g. maternal choice, routine IOL at 39 weeks) were also excluded.

Search strategy

To identify studies for inclusion we searched the databases PubMed (including Cochrane Library), Maternity and Infant Care (OVID), CINAHL, EMBASE and ClinicalTrials.gov from April 2008 to April 2018, which was then updated with an additional search until November 2019 to ensure the review was up to date at the time of publication. The databases were searched using the terms ‘labour induction’, ‘induction’ and ‘induction of labour’ in combination with the indications listed above. Following this we examined the reference lists of included articles for further studies. We also included studies that met our inclusion criteria previously included in the review by Mozurkewich et al. [10], which covered the period from January 1980 to April 2008 and used a comprehensive search strategy. This approach means that our review presents a comprehensive overview of the evidence from 1980 to date.

Data collection and extraction

All articles were screened for eligibility against the review criteria by reading the title and abstract by one reviewer (first author). All full text articles were reviewed by two authors to determine suitability for inclusion (See Fig 1). For each of the included studies, data were extracted by two reviewers using a purposely designed template that followed the PICO framework (method, population, intervention and comparator, outcomes) [30]. The quality of included studies was assessed by two reviewers independently, using the Cochrane the Risk of Bias in Randomised Trials Tool (RoB 2.0 tool) for RCTs [31], the Risk Of Bias In Non-randomised Studies of Interventions tool (ROBINS-I) for non-randomised studies [32] and the ROBIN tool [33] for systematic reviews.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flow of papers through review.

https://doi.org/10.1371/journal.pone.0228196.g001

Post-term pregnancy

There were 15 studies related to IOL for post-term pregnancy (> 40 weeks). These included a Cochrane review [6], a systematic review [34], an RCT [35], a prospective cohort study [36], nine retrospective cohort studies [37–45], and two secondary analyses of cohort studies [46, 47]. The Cochrane review assessed the effects of a policy of IOL at or beyond term compared with a policy of awaiting spontaneous labour (or until planned birth is deemed necessary) on maternal and neonatal outcomes [6]. This review includes 30 RCTs, with 12,479 women [48–77]. The majority of trials (about 75% of participants) adopted a policy of IOL at ≥ 41 weeks. IOL was associated with fewer perinatal deaths (2 vs 16) (risk ratio (RR) 0.33, 95% confidence interval (Cl) 0.14 to 0.78), lower NICU admissions (RR 0.88, 95% Cl 0.77 to 1.01), fewer babies with Apgar scores <7 at five minutes (RR 0.70, 95% Cl 0.50 to 0.98), and fewer CS (RR 0.92, 0.92, 95% Cl 0.85 to 0.99). The number needed to treat in order to prevent one perinatal death was 426. There was no significant difference between groups for perineal trauma (RR 1.09, 95% Cl 0.65 to 1.83), postpartum haemorrhage (RR 1.09, 95% Cl 0.92 to 1.30), length of maternal hospital stay (average mean difference -0.34 days, 95% Cl 1.00 to 0.33), or neonatal trauma (RR 1.18, 95% Cl 0.68 to 2.05). IOL was associated with an increase in operative vaginal births (RR 1.07, 95% Cl 0.99 to 1.16), in particular for IOL at < 41 weeks. Systematic reviews conducted prior to this Cochrane review included the same studies and, with the exception of one study [34], were excluded [12, 78–80]. The additional systematic review included reviewed the RCTs within the Cochrane review [6] and compared outcomes associated with IOL at 41 weeks versus 42 weeks [34]. This review identified four RCTs relevant to these timeframes and concluded that there was insufficient evidence to support IOL at 41 weeks instead of 42 weeks. A recent RCT not included in these systematic reviews compared IOL at 41 weeks with EM until 42 weeks found that IOL was associated with reduced adverse perinatal outcomes (1.7% vs 3.1%, absolute risk difference, -1.4%, 95% CI -2.9 to 0.0), however, this study was underpowered to demonstrate superiority of IOL at 41 weeks [35].

The remaining studies were cohort studies, mostly assessed as of moderate or severe risk of bias. In relation to outcomes associated with IOL at different gestational ages, three included retrospective cohort studies compared outcomes under different policy periods [38, 41, 45]. Bleicher et al. [41] compared outcomes for women who gave birth under a policy of IOL at 42 weeks (n = 968; from 2008–2009) with those who gave birth under a policy of IOL at 41 weeks (n = 962; from 2012–2013). This study found that, both the overall CS rate as well as the CS rate for women who underwent IOL, was lower during the 41 week policy period than during the 42 week policy period (15% vs 19.4%, p = 0.014 and 19% versus 27%, p = 0.007). IOL at 41 weeks was also associated with a significant reduction in 1st and 2nd degree perineal lacerations and neonatal readmission within 30 days of discharge. Similarly, Wolff et al. [45] compared outcomes for women who gave birth under a policy of IOL at 41+2 weeks (2012) (n = 8545) versus those who gave birth under a policy of IOL at 42 weeks (2010) (n = 9713). This study found a significant reduction in the rate of CS between 2010 and 2012 (p = 0.05), and a non-significant reduction in perinatal mortality of 60% (from 10 to 3). There were no significant differences in instrumental birth numbers or perinatal outcomes. Kassab et al. [38] compared CS rates associated with a policy of IOL at 41+3 days (n = 124; August 2006 and March 2007) versus a policy of IOL at 42 weeks (n = 227; April 2007 and July 2008), and found that the earlier IOL policy was associated with lower CS rates (p = 0.04) for nulliparous women.

Two cohort studies compared CS rates associated with IOL prior to 41 weeks versus IOL at 41 weeks [36, 47], also presenting opposing findings. A prospective cohort study by Haq et al. [36] compared CS rates for IOL at 40 weeks (n = 39) with IOL at 41 weeks (n = 39), and found that IOL at 41 weeks was associated with lower CS rates (16% vs 29%, p < 0.0001). Conversely, McCoy et al. [47] conducted a secondary analysis of a prospective cohort study to compare CS rates for IOL at term (between 37–40+6) (n = 700) with IOL at >41 weeks (n = 154) among women with an unfavourable cervix (Bishop score of <6 and cervical dilation <2 cm). This study found an increased risk of CS for women induced at >41 weeks versus those induced at term (46.8% vs 26.0%, p < .001).

The remaining seven papers included cohort studies that compared IOL with expectant management (EM) (spontaneous onset of labour or IOL if it becomes indicated) [37, 39, 40, 42–44, 46]. Mya et al. [46] conducted a secondary analysis of two multi-country surveys conducted by the World Health Organization (WHO) on maternal and newborn health to compare outcomes of IOL at 41 weeks (n = 4332) with EM (n = 26,720) (spontaneous onset of labour or birth at > 41 weeks). Compared to IOL, EM was significantly associated with decreased risk of CS in both databases (OR 0.70 and IOR 0.67). A retrospective cohort study by Mahomed et al. [37] compared CS rates associated with IOL at between 40 and 40+6 weeks (n = 2153) versus spontaneous birth between 41 and 41+6 weeks (n = 5658) for nulliparous women with uncomplicated singleton pregnancy, and found that CS rates were significantly higher in the IOL group (OR 1.52; 21% versus 14.9%).

Four other retrospective cohort studies compared the mode of birth associated with IOL at 41+1 weeks to EM until spontaneous onset of labour or IOL at 42–42+6 weeks [40, 42–44]. While, based on a relatively small sample size of 483 women (n = 211 in the IOL group; n = 277 in the EM group), Daskalakis et al. [40] found no significant differences in CS rate (36.5% vs 34.4%) or operative vaginal births (11.4% vs 9.2%) between the two groups, the other studies found that IOL was associated with increased risk of CS. A study by Thangarajah et al. [44], included 856 women (n = 400 in the IOL group; n = 456 in the EM group) and found that IOL was associated with increased rates of CS (33.8% vs. 21.1%, p < 0.001), and perineal lacerations (38.1% vs 26.4%, p = 0.002). Similarly, a study by Pavicic [42] including 1367 women (n = 722 in the IOL group; n = 645 in the EM group) found that IOL was associated with increased CS rates (25.4% vs 16.6%, p = 0.001) and a study by Teo [43] including 6501 women (n = 3588 in the IOL group; n = 2913 in the EM group) found that IOL was associated with higher rates of CS (29.4% vs 18.5%, p = 0.001) and instrumental birth (20.2% vs 17.7%, p = 0.012). The two latter studies were rated as having serious risk of bias as the IOL groups included a significantly higher proportion of women with comorbidities, which were not controlled for.

Lastly, Hermus et al. [39] conducted a retrospective matched cohort study (1:1 ratios for both age and parity) to compare outcomes for women who underwent IOL (= 377) at 42 weeks to EM beyond 42 weeks (n = 377). This study found that EM was associated with non-significantly lower rate of CS (12.5% vs 13.6%, RR 0.9, 95% CI 0.6 to 1.4), but higher incidence of shoulder dystocia (RR 4.3, 95% CI 1.3 to 15) and meconium-stained amniotic fluid (RR 1.8, 95% CI 1.4 to 2.3).

Premature rupture of membranes (PROM)

Five studies in relation to PROM were included, consisting of two Cochrane reviews [81, 82], one prospective cohort study [83], and two retrospective cohort studies [84, 85]. One Cochrane review and three studies addressed PROM at term (37–42 weeks) [81, 83–85], whilst the other Cochrane review addressed preterm PROM (<37 weeks) [82].

In relation to preterm PROM (< 37 weeks), the Cochrane review by Bond et al. [82] compared outcomes associated with women undergoing a planned early birth (by IOL or CS) with EM between 24 and 37 weeks. The review identified 12 RCTs, with 3617 women [7, 86–96]. In terms of neonatal outcomes, this review identified no clear differences in neonatal sepsis (RR 0.93), neonatal infection (RR 1.24, 95% CI 0.66 to 1.30), and overall perinatal mortality (RR 1.76, 95% CI 0.89 to 3.50), but found that early birth was associated with a higher rate of neonatal death (RR 2.55, 95% CI 1.17 to 5.56), respiratory distress syndrome (RR 1.26, 5% CI 1.05 to 1.53), need for ventilation (RR 1.27, 95% CI 1.02 to 1.58), and NICU admission (RR 1.16, 95% CI 1.08 to 1.24). In terms of maternal outcomes, early birth was associated with an increased rate of CS (RR 1.26, 95% CI 1.11 to 1.44) and endometritis (RR 1.61, 95% CI 1.00 to 2.59), and reduced rate of chorioamnionitis (RR 0.50, 95% CI 0.26 to 0.95). This review conducted a subgroup analysis by gestational age, and compared outcomes associated with planned birth <34 weeks and >34 weeks. The test for subgroup differences were not significant for neonatal infection, respiratory distress syndrome, CS, and chorioamnionitis. There was a decrease in endometritis in women randomised to early birth after 34 weeks [82]. The included studies were at low or unclear risk of bias, with the overall quality of evidence rated as moderate to high.

In relation to PROM at term (≥37 weeks), the Cochrane review by Middleton et al. [81] compared planned early birth (immediate IOL or within 24 hours) with EM (no planned intervention within 24 hours). The review identified 23 RCTs, involving 8615 women [97–117]. Early birth was associated with a reduced risk of maternal infectious morbidity (chorioamnionitis and/or endometritis) (RR 0.49, 95% CI 0.33 to 0.72), and neonates were less likely to have early-onset neonatal sepsis (RR 0.73, 95% CI 0.58 to 0.92). No clear differences were seen in CS rates (RR 0.84, 95% CI 0.69 to 1.04; 23); serious maternal morbidity or mortality (no events); definite early-onset neonatal sepsis (RR 0.57, 95% CI 0.24 to 1.33); or perinatal mortality (RR 0.47, 95% CI 0.13 to 1.66). The quality of included studies was low; only three of the included RCTs were of low risk of bias, while the remaining were of unclear or high risk of bias.

The remaining studies were cohort studies, assessed as of low to moderate risk of bias. A prospective cohort study by Omole-Ohonsi et al. [83] compared immediate IOL (N = 100) with delayed IOL after EM of 12 hours (N = 100). One-third in the delayed IOL group went into spontaneous labour before IOL. Immediate IOL was associated with significantly lower rates of CS (OR 0.18, 95% CI 0.07 to 0.47), operative vaginal birth (OR 0.26, 95% CI 0.07 to 0.88), and higher rates of vaginal birth (OR 6.10, 95% CI 2.76 to 13.75). There was no significant difference in neonatal outcomes.

A retrospective cohort study by Sadeh-Mestechkin et al. [84] compared immediate IOL (N = 213) versus delayed IOL after EM for 48 hours (N = 112). The delayed group had significantly higher rate of prolonged hospitalisation (p = 0.043) (as an indicator for maternal complications), and higher rates of CS (16.4% vs 7.1%, p = 0.024). There was no significant difference in the rate of clinical chorioamnionitis or postpartum endometritis, and there were no cases of early neonatal sepsis. A retrospective cohort study by Pintucci et al. [85] analysed outcomes associated with a policy of delayed IOL after EM for 48 hours (N = 1315). In total, 84% of the women went into spontaneous labour within 48 hours. There were low rates of clinical chorioamnionitis (2.3%) and neonatal infection rate (2.8%). The overall CS rate was 4.5%, which was lower for women who went into labour spontaneously than those who underwent IOL (OR 1.76, 95% CI 1.03 to 3.02).

Hypertension/preeclampsia

Fifteen studies in relation to preeclampsia, chronic or gestational hypertension were included. There were three Cochrane reviews [118–120], two RCTs [121, 122], a post hoc analysis of an RCT [123], three prospective cohort studies [124–126], four retrospective cohort studies [127–130], and two case control studies [131, 132]. A further systematic review was excluded as it did not identify any additional studies not already included [133]. The majority of studies focussed on severe preeclampsia as an indicator for IOL, with most of these in women giving birth <34 weeks gestation [119–121, 124–127, 130–132]. One study pertained to gestational hypertension [129], one to chronic hypertension [128], one to late preterm preeclampsia (34–36+6 weeks gestation) [122] and two studies included multiple hypertensive disorders as one group [118, 123].

In relation to hypertensive disorders broadly, Cluver et al. [118] conducted a Cochrane review comparing planned early birth versus EM from 34 weeks to term. The review identified five RCTs, with a total of 1819 women [134–138]. Both the HYPITAT-I (Hypertension and Preeclampsia Intervention Trial at Term) and HYPITAT-II trials were included [134, 135]. The HYPITAT-I trial compared IOL at 36–41 weeks (within 24 hours of randomisation) to EM until spontaneous onset of labour for pregnant women with mild to moderate gestational hypertension or mild preeclampsia [135]. The HYPITAT-II trial compared IOL at 34–36+6 weeks to EM until 37 weeks for pregnant women with gestational hypertension, mild preeclampsia, or deteriorating chronic hypertension [134]. Three further RCTs compared planned early birth via IOL or CS versus EM for pregnant women with mild preeclampsia or gestational hypertension [136, 138] or moderate essential chronic hypertension [137]. The review found a lower risk of composite maternal mortality and severe morbidity for women randomised to receive planned early birth. There was not enough information to draw conclusions about the effects on composite infant mortality and severe morbidity, with contrasting findings in HYPITAT-I (IOL from 36 weeks with RR 0.75, 95% CI 0.46 to 1.28 for composite infant outcome) [135] versus HYPITAT- II (IOL from 34 to 36+6 weeks, RR 1.4 95% CI 1.1 to 1.7 for any neonatal morbidity) [134]. Planned early birth was associated with higher levels of respiratory distress syndrome (RR 2.24, 95% CI 1.20 to 4.18), and NICU admissions (RR 1.65, 95% CI 1.13 to 2.40), with this finding driven by higher neonatal risks in the earlier planned birth group of HYPITAT II [134]. There was no clear difference between the groups for CS or length of hospital stay.

Tajik et al. [123] conducted a post hoc analysis of the HYPITAT-I trial to assess whether cervical ripeness should play a role in the decision for IOL for women with gestational hypertension or mild preeclampsia at term. This trial included a total of 756 women, with 377 in the IOL group, and 379 in the EM group (spontaneous labour). This study found that the superiority of IOL varied significantly according to cervical favourability. The length of the woman’s cervix was a predictor of outcome. For women who were managed expectantly, the longer the cervix, the higher the risk of maternal complications, whereas in women who were induced, cervical length was not associated with increased maternal complications. Similarly, IOL was more likely to reduce the CS rate in women with an unfavourable cervix.

Finally, two retrospective cohort studies sought to determine the optimal timing of birth for women with gestational hypertension [129] and chronic hypertension [128]. Both studies found that IOL between 38- and 39 weeks balances the lowest maternal and neonatal morbidity/mortality for both women with gestational hypertension [129] and those with chronic hypertension [128].

In relation to severe preeclampsia, the majority of included studies assessed outcomes remote from term, i.e. during the very preterm period of less than 34 weeks. Only one study focussed on late preterm preeclampsia (34–36+6 weeks gestation) [122]. An RCT by Chappell et al. (2019) compared planned birth (usually IOL) (n = 448) versus EM (n = 338) in women with late preterm pre-eclampsia from 34 to 37 weeks (The PHOENIX trial) [122]. This study found that planned birth reduced maternal morbidity and severe hypertension (65% vs 75%, RR = 0.86, 95% CI 0·79–0·94; p = 0·0005), but resulted in more neonatal admissions for prematurity (42% vs 34%, RR 1·26, 1·08–1·47; p = 0·0034).

In relation to preeclampsia remote from terms (<34 weeks), a Cochrane review by Churchill et al. [119] compared planned early birth versus EM for severe preeclampsia between 24–34 weeks’ gestation. The review included four RCTs, with a total of 425 women [139–142]. The study found that an expectant approach may be associated with decreased morbidity for the baby. Babies whose mothers were allocated to the early birth group had more intraventricular haemorrhage (RR 1.82, 95%CI 1.06 to 3.14), more hyaline membrane disease (RR 2.30, 95% CI 1.39 to 3.81), required more ventilation (RR 1.50, 95% CI 1.11 to 2.02), were more likely to have a lower gestation at birth, more likely to be admitted to neonatal intensive care (RR 1.35, 95% CI 1.16 to 1.58) and have a longer stay in the NICU, but less likely to be small-for-gestational age (RR 0.30, 95% CI 0.14 to 0.65). There was insufficient data for reliable conclusions on most outcomes for the mother, except that women allocated to the early birth group were more likely to have a CS (RR 1.09, 95% CI 1.01 to 1.18). There was also a second systematic review and meta-analysis of RCTs by Wang et al. [143] which largely included the same studies as the Cochrane review [119], with an additional two studies: the MEXPRE trial [121] and a trial by Duvekot et al. [144]. The latter was closed after 24 months because of low recruitment and the findings were reported by abstract only, and as such is excluded from our review. As the only additional study identified by this review is the MEXPRE trial [121], we have excluded the review by Wang et al. [143] and we report on the MEXPRE trial here.

The MEXPRE trial included 267 women and sought to determine whether EM in women with severe preeclampsia prior to 34 weeks results in improved neonatal outcome in countries with limited resources (i.e. low-middle income countries in Latin America) [121]. This study found no difference in the rate of perinatal mortality (9.4% vs 8.7%; RR 0.91, 95% CI 0.34 to 1.93), composite of neonatal morbidities (56.4% vs 55.6%; RR 1.01, 95% CI 0.81 to 1.26), or maternal morbidity (25.2% vs 20.3%; RR 1.24, 95% CI 0.79 to 1.94) with EM versus early birth. Small for gestational age (21.7% vs 9.4%; RR 2.27, 95% CI 1.21 to 4.14) and placental abruption were more common with EM (RR 5.07, 95% CI 1.13 to 22.7).

Four cohort studies [125–127, 130] and two case control studies [131, 132] assessed the association between mode of birth and maternal and neonatal outcomes for severe preeclampsia remote from term. A prospective cohort study by Ertekin [125] that compared EM (n = 33) versus early birth by IOL or CS (n = 37) for severe preterm preeclampsia on a range of maternal and neonatal outcomes, including the first year of neurological development of infant, did not find statistically significant differences between the two groups. However, there were seven fetal deaths in the immediate birth group versus two in the EM group (P = 0.058). This was most likely due to women with more significant risk factors (e.g. HELLP syndrome) being assigned to the immediate birth group.

The remaining studies addressed the safety of women undergoing an IOL versus a CS for severe preeclampsia remote from term. With the exception of one study [126], these concluded that IOL is a reasonable option that was not associated with poorer maternal and neonatal outcomes [127, 130–132]. A retrospective cohort study by Alanis at al. [127] assessed differences in neonatal outcomes with IOL (n = 282) versus planned CS (n = 209) in women with early onset severe preeclampsia and found that IOL was not associated with an increase in neonatal morbidity or mortality even after controlling for gestational age and other confounders. Similarly, Alexander et al. [130] conducted a retrospective cohort study comparing IOL (n = 145) with planned CS (n = 133) on neonatal outcomes in women whose pregnancies were complicated by preterm severe preeclampsia and birth of very low birth-weight infants. This study found that IOL in cases of severe preeclampsia was not harmful to very low birth-weight infants. While Apgar scores of ≤3 at 5 minutes were more likely in the IOL group (6% versus 2%, P = 0.04), other neonatal outcomes, including respiratory distress syndrome, grade 3 or 4 intraventricular haemorrhage, sepsis, seizures, and neonatal death, were similar in the two groups. Vaginal birth was accomplished by 50 (34%) women in the IOL group.

The two included case control studies had similar findings, and also indicated that IOL for severe preeclampsia should be considered as a reasonable option remote from term rather than a CS [131, 132]. Both studies conducted retrospective chart reviews to determine the rate of vaginal birth after IOL in severe preeclampsia remote from term to identify factors associated with its success and evaluate neonatal outcomes based on induction outcome. Based on a sample of 306 women, Nassar et al. [132] found that of the women that were induced, 48% gave birth vaginally. The Bishop score on admission was the best predictor of success, although the chance of successful IOL increased with advancing gestational age (ranging from 31.6% at </ = 28 weeks' gestation to 62.5% at >32 weeks' gestation). Based on a sample of 250 women, Blackwell et al. [131] found that attempted IOL did not increase neonatal morbidity, and that IOL success was significantly associated with gestational age (rarely successful at <28 weeks). The only study that presented different findings was a prospective cohort study by Mashiloane et al. [126] that compared outcomes associated with planned CS (n = 68), CS following attempted IOL (n = 14), and vaginal birth following successful IOL (n = 26) for severe preeclampsia from 26–32 weeks. This study found that perinatal mortality was significantly higher following IOL (p = 0.0004), and that planned CS contributed to a better perinatal outcome than vaginal birth.

We also identified two other studies in relation to preeclampsia in broader terms, not specified as remote from term. A prospective cohort study by Amorim et al. [124] evaluated the association between mode of birth (vaginal versus CS) and maternal outcomes for 500 women with severe preeclampsia. This study found that the risk of severe maternal morbidity was significantly greater in women in the CS group (54.0% versus 32.7%) irrespective of the presence of labour. Severe maternal morbidity was found to be associated with CS (OR 1.91). Amorim et al. [120] also conducted a Cochrane review to compare maternal and neonatal outcomes for women with severe preeclampsia who had a planned CS versus planned vaginal birth. However, this review did not identify any studies for inclusion and had no results to report.

Diabetes

Nine studies in relation to diabetes were included, consisting of two Cochrane reviews [9, 145], a secondary analysis of a trial that compared different treatments for mild gestational diabetes mellitus (GDM) [146] and six retrospective cohort studies [147–152]. With the exception of one study [145], all evaluated women with GDM, and excluded those with type I or II diabetes.

The Cochrane review by Bouvain et al. [145] included one RCT that compared outcomes for IOL ≥38 weeks (n = 100) versus EM (until 42 weeks) (n = 100) for pregnant women with diabetes (either type I or type II, or GDM) treated with insulin [153]. Of the 200 participants, 187 women had GDM and 13 had type 2 diabetes. This study found no significant difference between the two groups in terms of CS (RR 0.81, 95% CI 0.52 to 1.26). The risk of macrosomia was reduced in the IOL group (RR 0.56, 95% CI 0.32 to 0.98) and three cases of mild shoulder dystocia were reported in the EM group. No other types of perinatal morbidity were reported.

The Cochrane review by Biesty et al. [9] assessed the effect of planned birth for women with gestational diabetes and included one RCT, the GINEXMAL trial [154]. This trial included 425 women with GDM at term, randomised to IOL (between 38 and 39 weeks) (n = 214) or EM (n = 211) (until 41 weeks). This study found no difference between the two groups in terms of CS rates (RR 1.06, 95% CI 0.64 to 1.77; 12.6% in the IOL group versus 11.7% in the expectant group), or other maternal or neonatal outcomes. There were no maternal or fetal deaths. The study was underpowered and any reported differences between the two groups were very small and not clinically relevant.

There were four retrospective cohort studies that compared outcomes for women with GDM who underwent IOL versus EM, with CS rates as a primary outcome [147, 149–151]. A retrospective cohort study by Bettikher et al. [147] that compared outcomes for IOL (gestation not stated) in 43 women with EM until spontaneous labour in 188 women, found no significant difference between the two groups in terms of CS rate, or any of the other maternal or neonatal outcomes assessed (i.e. the frequency of uterine inertia, uncoordinated contractions and fetal distress). This study concluded that, in the absence of signs of fetal distress or macrosomia, planned early birth was not indicated. Similarly, Grabowska et al. [149] compared CS rates for 96 women who underwent IOL with 32 women who had a spontaneous labour and found that IOL did not increase the risk of CS (25% versus 25%, p = 0.66). A cohort study by Melamed [150] compared outcomes associated with IOL at 38 or 39 weeks and EM for 6417 women with GDM, and found that IOL at 38 or 39 weeks was associated with a lower CS rates but higher risk of NICU admission, when done at <39 weeks gestation.

Four studies assessed the impact of different timings of IOL for women with GDM in terms of CS rates [146, 148] or maternal and neonatal outcomes [151, 152]. A retrospective cohort study by Hochberg et al. (2019) compared outcomes for IOL at 37+0 and 38 + 6 weeks (n = 193) versus 39+0 and 40+6 weeks (n = 237) and found that the rates of composite maternal outcome and composite neonatal outcome did not differ between groups [152]. A retrospective cohort study by Vitner et al. (2019) found that IOL was associated with increased risk for adverse composite neonatal outcome or NICU admission when done prior to 39 weeks [151].

Specific to CS rates, Sutton et al. [146] conducted a secondary analysis of a trial investigating different treatments for mild GDM [155] to compare the rates of CS associated with IOL (n = 220) versus EM (n = 459) at different gestational ages. This study found that IOL was not associated with increased rates of CS at 37, 38, or 39 weeks, but was associated with a 3-fold increase in CS rates at 40 weeks and beyond. A retrospective cohort study by Feghali et al. [148] compared CS rates in women undergoing IOL at each week of gestation, with EM to a later gestational age. Similarly, IOL at 37 weeks, 38 weeks and 39 weeks was associated with similar rates of CS than EM, particularly for nulliparous women. The difference in CS rates between the two groups was only significant at 38 weeks for multiparous women. This study did not report on outcomes at 40+ weeks. The included cohort studies were all rated as having a moderate to serious risk of bias, as the reason for IOL was not clearly defined, therefore findings need to be interpreted with caution.

Twin pregnancy

We identified five studies in relation to twin pregnancy, consisting of a Cochrane review [156], and four retrospective cohort studies [157–160]. None of the included studies conducted an analysis by chorionicity. Chorionicity data was unavailable or incomplete in three studies [156, 158, 159], and two studies had the data but did not perform this sub-analysis [157, 160].

The Cochrane review [156] compared elective birth from 37 weeks to EM for women with an otherwise uncomplicated twin pregnancy. This review identified two RCTs for inclusion, involving a total of 271 women and 542 infants [156, 161]. Women in the elective birth group (n = 133) had a planned birth at 37 weeks by either CS or IOL in one of the included studies [162], and by IOL in the other study [161]. Women randomised to the EM group (n = 138) had their care according to local hospital guidelines, which involved either awaiting the spontaneous onset of labour, or having a planned birth after 38 weeks. The review found no statistically significant differences in risk of CS, perinatal death or serious infant morbidity, or maternal death or serious maternal morbidity, although both perinatal (RR 0.34, 95%CI 0.01 to 8.35) and maternal composite morbidity and mortality (RR 0.29, 95% CI 0.06 to 1.38) were lower (albeit not statistically) in the elective birth group.

A retrospective cohort study by Tavares et al. [160] compared outcomes for IOL versus spontaneous vaginal birth in twin pregnancy after 36 weeks of gestation. Of the 288 women with multiple pregnancies during the study period, 75 twin pregnancies >36 weeks of gestation were included, with 33 women undergoing IOL and 42 women who went into labour spontaneously. This study found no statistical differences between the two groups in terms of maternal and neonatal morbidity, or admission to the NICU, but did find a significant increase in CS in the IOL group (60.6 vs. 33.3%, p<0.05).

The remaining cohort studies examined birth outcomes in twin pregnancies. A study by de Castro et al. [158] measured twin pregnancy labour outcomes, with a subgroup comparison for women who underwent IOL (n = 287) with women who had a spontaneous vaginal birth for both twins, a CS or a vaginal birth for first twin and CS for 2nd twin (n = 596). This study found that IOL significantly reduced the chance of achieving vaginal birth (Foley catheter induction 74.7%; non-Foley induction 86.7%; no induction 88.9%, p < 0.001). Similarly, Jonsson [157] retrospectively compared outcomes for women who underwent IOL (n = 220) with those who had a spontaneous labour (n = 242), and found that IOL in twin pregnancies increased the risk of CS compared with spontaneous labour onset, especially if Foley catheters or prostaglandins were used. However, approximately 80% of babies born from women who had an IOL were born vaginally. These findings are not supported by Hamou et al. [159] which compared outcomes for women who had an IOL (n = 653); spontaneous birth (n = 2937) versus elective CS (n = 1015). While consistently this study found that the rate of vaginal birth in the IOL group was 81%, IOL was found to be associated with a lower rate of CS than spontaneous labour (77% reduction, OR 0.23) and lower rates of neonatal death (78% reduction, OR 0.22). However, this study was assessed as of serious risk of bias, and the high adverse outcome observed in the spontaneous twin group is likely because a large proportion of them (25%) were early preterm, prior to 34 weeks.

Intrahepatic cholestasis of pregnancy

In relation to intrahepatic cholestasis of pregnancy (ICP), we identified three studies, consisting of a Cochrane review [163] and two retrospective cohort studies [164, 165]. The Cochrane review [163] evaluated the effectiveness and safety of interventions in women with cholestasis of pregnancy. While this review identified 21 RCTs, only one included RCT, the PITCH trial, compared outcomes for early term birth to EM (and as such is the only RCT relevant here) [166]. The PITCH trial included 63 women, 30 of which were randomised to IOL between 37–37+6 weeks and 33 randomised to EM (spontaneous labour until 40 weeks or CS undertaken by normal obstetric guidelines, usually after 39 weeks). There were no stillbirths or neonatal deaths in either group and no significant differences in CS (RR 0.68), passage of meconium-stained liquor (RR 0.55, 95% CI 0.15 to 2.01) or admission to NICU (RR 0.55, 95% CI 0.05 to 5.76). This study was underpowered to detect other clinically important differences. Subgroup analysis results by bile acid level were not reported.

The retrospective cohort study by Kohari et al. [165] sought to determine the efficacy of a planned early birth policy for women with severe ICP (bile acids >40μmol/L) by comparing outcomes for women who gave birth under an active management policy (between 2009–2013) to those who were cared for prior to the introduction of this policy (between 2005–2008). Women with ICP who gave birth under the active management policy (n = 128) were managed as inpatients and had a planned birth between 36 and 37 weeks. Prior to the introduction of this policy, decisions around mode and timing of birth were made at the discretion of the health professional. The active management policy was found to be associated with a significant reduction in the incidence of stillbirth (0% versus 3.4%, p = 0.035). There was no difference in CS rates or NICU admissions. Women’s demographic characteristics were similar between the groups, with the exception of greater maternal age and GDM in the newer cohort. A retrospective cohort study by Puljic et al. [164] sought to determine the optimal timing of birth for pregnancies complicated by ICP (no stratification by bile acid level), by comparing outcomes by each additional week (from 34–40 weeks) of EM versus immediate birth. This study found that birth at 36 weeks gestation was associated with lower perinatal mortality.

Elevated maternal body mass index

In relation to elevated maternal BMI (≥30.0 kg/m2), four retrospective cohort studies were included [167–170], presenting mixed findings. Wolfe et al. [170] compared maternal and neonatal outcomes in obese (≥30.0 kg/m2), nulliparous women with an unfavourable cervix (modified Bishop score < 5) undergoing elective IOL between 39 and 41 weeks (n = 60) with EM after 39 weeks (n = 410). This study found that IOL was associated with higher rates of CS (40% vs 25.9%, p = .022), and NICU admissions (18.3% vs 6.3%, p = .001). Other maternal and neonatal outcomes were similar. These findings were not supported by the other included studies, described below, which found that IOL was associated with lower CS rates [167–169].

Kawakita et al. [167] compared the CS rate of elective IOL with EM in morbidly obese women (BMI ≥40 kg/m2) between 37 and 41+6 weeks who had singleton pregnancies, cephalic presentations and no previous CS or other comorbidity. In nulliparous women, elective IOL was not associated with increased risks of CS and was associated with decreased risks of macrosomia (2.2% vs 11.0%) at early term and decreased NICU admissions (5.1% vs 8.9%) at full term. In multiparous women, IOL was associated with a decreased risk of macrosomia at early term (4.2% vs 14.3%), CS at full term (5.4% vs 7.9%), and composite neonatal outcome (0% vs 0.6%) at full term.

Both Lee et al. [169] and Pickens et al. [168] compared outcomes between elective IOL and EM in obese women (BMI ≥30.0 kg/m2) with singleton pregnancies by analysing a Californian national dataset. Lee et al. [169] analysed 2007 data and compared outcomes for the two groups for each gestational age, from 37–40 weeks. This study found that IOL was associated with lower CS rates, and lower odds of macrosomia. There were no differences in the other reported outcomes. Similarly, Pickens et al. [168] analysed data from 2007 to 2011, comparing outcomes for these two groups at 39 and 40 and 41 weeks. This study found that IOL was associated with reduced CS rates (at 39 weeks gestation, frequencies were 35.9% vs 41.0%, p = <0.05), reduced composite of severe maternal morbidity (5.6% vs 7.6%, p = <0.05), and reduced NICU admissions (7.9% vs 10.1%, p = <0.05).

Maternal age

In relation to maternal age, we found two studies to include, one RCT [171] and one retrospective cohort study [172]. An RCT by Walker et al. [171] tested if IOL at 39 weeks reduces CS rates for nulliparous women who are ≥ 35 years by comparing outcomes for IOL (n = 305) with EM (n = 314). This study found no significant differences in the two groups in terms of CS rates (32% in the IOL group vs 33% in the EM group; RR 0.99, 95% CI 0.87 to 1.14), the percentage of women who had a vaginal birth with the use of forceps or vacuum (38% vs 33%, RR 1.30, 95% CI 0.96 to 1.77), the women’s experience of childbirth, or adverse maternal or neonatal outcomes. There were no maternal or infant deaths.

A retrospective cohort study by Knight et al. [172] compared perinatal mortality between IOL (at between 39 and 41 weeks) (n = 25,583) and EM (n = 51,744) for nulliparous women aged ≥ 35 years. Women with comorbidities or previous complicated births were excluded. IOL at 40 weeks was associated with a lower risk of in-hospital perinatal death (0.08% vs 0.26%; RR 0.33 95% CI 0.13 to 0.80; p = 0.015) and meconium aspiration syndrome (0.44% vs 0.86%; RR 0.52, 95% CI 0.35 to 0.78; p = 0.002), but an increased risk of instrumental vaginal birth (RR 1.06, 95% CI 1.01 to 1.11; p = 0.020) and CS (RR 1.05, 95% CI 1.01 to 1.09; p = 0.019). A number needed to treat analysis indicated that 562 women would require IOL at 40 weeks to prevent one perinatal death.

Maternal cardiac disease

One study, a prospective cohort study, addressed maternal cardiac disease [173]. This study examined the safety of IOL in women with cardiac disease by comparing outcomes for women who underwent IOL between 37 and 40 weeks (n = 47) versus EM (n = 74) (spontaneous onset of labour resulting in vaginal birth or emergency CS). There was no significant difference in complication rate between the two groups, however the groups were not well matched as women in the IOL group had more severe cardiac disease than those in the EM group.

Suspected macrosomia

In relation to suspected macrosomia as an indication for women undergoing an IOL, there were three studies, consisting of a Cochrane review [174], a systematic review and meta-analysis of RCTs and observational studies [175] and one retrospective cohort study [176]. We also identified a fourth study, a systematic review and meta-analysis, but this review included the same papers as the Cochrane review (the Cochrane was assessed as higher quality) and was therefore excluded [177].

The Cochrane review by Boulvain (2016) [174] sought to determine the effects of a policy of IOL of between 37 and 40 weeks for suspected fetal macrosomia on CS rates and maternal or perinatal morbidity. The review identified four RCTs [178–181] with a total of 1190 women, 590 in the IOL group and 600 in the EM group. In two of the included trials, women with diabetes were excluded [179, 180], one trial excluded women treated with insulin, but included participants who had GDM controlled by diet (10%) [178], whilst the participant inclusion criteria for the fourth trial was unclear [181]. Women were included when the fetal weight, estimated by ultrasound examination, was between 4000g and 4500g [180], and between 4000g and 4750g [179], when the fetus was estimated to weigh >97th percentile [181], or >95th centile [178]. This review found that compared to EM, IOL had no clear effect on the risk of CS (RR 0.91, 95% CI 0.76 to 1.09) or instrumental birth (RR 0.86, 95% CI 0.65 to 1.13) but did reduce shoulder dystocia (RR 0.60, 95% CI 0.37 to 0.98) and fetal fracture (any) (RR 0.20, 95% CI 0.05 to 0.79). There was no strong evidence of any difference between groups for low infant Apgar scores (<7 at one minute) (RR 1.51, 95% CI 0.25 to 9.02) or low arterial cord blood pH (RR 1.01, 95% CI 0.46 to 2.22). There were no clear differences between groups for brachial plexus injury, although this outcome was infrequent. Two studies reported third- and fourth-degree perineal tears, but only one had estimable data [178]; this study found the number of women with perineal tears was increased in the IOL group (RR 3.70, 95% CI 1.04 to 13.17). There was no perinatal mortality, and no differences in the groups in terms of the number of newborns with intraventricular haemorrhage (RR 1.06, 95% CI 0.19 to 5.96), or neonatal intensive care admissions (RR 0.66, 95% CI 0.35 to 1.24).

The systematic review and meta-analysis by Sanchez-Ramos et al. [175] included two of the RCTs included in the Cochrane review [179, 180], in addition to nine observational studies published between 1966 and 2002, with a total of 3751 women (2700 in the IOL group and 1051 in the EM group). The observational and RCT data were analysed separately; since we have already reported on the RCT findings, we only report the findings of the observational studies here. Analysis of the non-randomised studies indicates that the risk of CS may be increased when IOL is undertaken. Women who experienced spontaneous onset of labour had a lower incidence of CS in comparison to the IOL group (OR 0.39, 95% CI 0.30 to 0.50) and higher rates of spontaneous vaginal birth (OR 2.07, 95% CI 1.34 to 3.19). No differences were noted in rates of operative vaginal births, incidence of shoulder dystocia, or abnormal Apgar scores, in the analyses of the observational studies.

An observational study by Cheng et al. [176] based on known birthweight presented different findings. This retrospective cohort study compared the frequency of CS for women who had an IOL at 39 weeks with a neonatal birthweight of 4000 +/- 125g (birthweight 3875-4125g) with women who gave birth (following IOL or spontaneous onset of labour) at 40 weeks with birthweight 4075–4325g, at 41 weeks with a birthweight at 4275–4525g, or 42 weeks with a birthweight of 4475–4725g (assuming an intrauterine fetal weight gain of 200g per additional week of gestation). The frequency of CS in the IOL group was lower compared with women who gave birth at a later gestational age (35.2% versus 40.9%; OR 1.25, 95% CI 1.17 to 1.33). This study concluded that in the setting of macrosomia and known birthweight, IOL may reduce CS rates.

Fetal gastroschisis

In relation to known fetal gastroschisis, we identified two studies for inclusion, a Cochrane review [182] and a retrospective cohort study [183]. The Cochrane review [182] assessed the effects of elective preterm birth (<37 weeks) for fetal gastroschisis, and identified one RCT for inclusion [184]. This RCT assessed whether planned birth at 36 weeks reduces postnatal morbidity without exposing the infant to the added risks of prematurity by comparing outcomes for IOL at 36 weeks (n = 21) and spontaneous onset of labour (n = 21). The trial found no significant benefits or adverse effects associated with elective preterm birth, however, it was underpowered to detect clinically important differences. Two babies died in the planned birth group versus none in the spontaneous group. Seven women (33%) in the planned birth group and nine women (43%) in the spontaneous group had a CS (RR 0.78, 95% CI 0.36 to 1.70). There were no statistical differences in birthweight, ventilation requirements, necrotising enterocolitis and requirements for repeat surgery between the two groups. The average gestational age at birth was 35.8 weeks in the planned birth group and 36.7 weeks in the spontaneous onset of labour group.

An observational study by Al-Kaff et al. [183] analysed data from a national dataset in Canada that included 519 fetuses diagnosed with isolated gastroschisis between 2005 and 2013. This study compared outcomes for mode of birth (spontaneous labour, n = 190; IOL, n = 280; CS, n = 49) and timing of birth (≤35 weeks, n = 8; 36–37 weeks, n = 193; ≥38 weeks, n = 69). Neither mode nor timing of birth were associated with significant benefits or adverse effects. Planned IOL was not associated with decreased length of neonatal stay, total parenteral nutrition duration, or risk of the composite adverse outcome (RR 1.7, 95% CI 0.1 to 3.2) compared with birth following spontaneous onset of labour. Planned birth at 36–37 weeks was not associated with decreased length of neonatal stay, total parenteral nutrition duration or risk of composite outcome (RR 2.3, 95% CI 0.8 to 5.4) compared with planned birth at 38 weeks. Findings support awaiting the onset of spontaneous labour in pregnancies that are complicated by fetal gastroschisis.

Suspected fetal compromise

We identified four studies in relation to fetal compromise, consisting of two Cochrane reviews [185, 186] and two retrospective cohort studies [187, 188]. The included studies regarded compromise in the context of oligohydramnios [185, 187] as well as suspected IUGR [185, 186, 188]. The Cochrane review by Stock et al. [186] assessed the effects of immediate versus deferred birth of preterm babies with suspected fetal compromise on neonatal, maternal and long-term outcomes. This review included one trial of 548 women (588 babies) with pregnancies between 24 and 36 weeks and compared the outcomes for immediate birth (by IOL or CS) (n = 296) versus deferred birth (n = 291) (i.e. a set period of time until test results worsen, or until spontaneous onset of labour) [139, 189]. More women in the immediate birth group had a CS (RR 1.15, 95% CI 1.07 to 1.24), there were more babies who were ventilated for more than 24 hours (RR 1.54, 95% CI 1.20 to 1.97), and more had cerebral palsy at or after two years of age (RR 5.88, 95% CI 1.33 to 26.02). There was no real difference for other neonatal morbidity and mortality outcomes, perinatal mortality (RR 1.17, 95% CI 0.67 to 2.04), the composite outcome of death or disability at or after two years of age (RR 1.22, 95% CI 0.85 to 1.75), neurodevelopment impairment at or after two years (RR 1.72, 95% CI 0.86 to 3.41), death at or after two years of age (RR 1.04, 95% CI 0.66 to 1.63), or death or disability in childhood (6–13 years of age) (RR 0.82, 95% CI 0.48 to 1.40). The gestational age at birth was a median of four days earlier in women randomised to immediate birth. The review concluded that for preterm babies with suspected compromise and uncertainty about whether to hasten the birth or not, there appeared to be no benefit to immediate birth.

The Cochrane review by Bond et al. [185] assessed the effects of immediate birth (via IOL or CS) versus EM (until spontaneous onset of labour or planned early birth if it became necessary) for suspected fetal compromise at term (≥ 37 weeks) on neonatal, maternal and long-term outcomes. This review identified three RCTs for inclusion, with a total of 546 participants, of which 269 were randomised to immediate birth and 277 to EM. Two of the trials compared outcomes in 492 women whose pregnancies were complicated by IUGR [190–192], and one trial included 54 women with oligohydramnios [193]. This review found no difference in neonatal outcomes including perinatal mortality (no deaths in either group), major neonatal morbidity (RR 0.15, 95% CI 0.01 to 2.81), or neurodevelopmental disability/impairment at two years of age (RR 2.04, 95% CI 0.62 to 6.69). There was no difference in the risk of necrotising enterocolitis or meconium aspiration. There was also no difference in maternal mortality (RR 3.07, 5% CI 0.13 to 74.87), significant maternal morbidity (RR 0.92, 95% CI 0.38 to 2.22), CS rates (RR 1.02, 95% CI 0.65 to 1.59) or secondary maternal outcomes. Significantly more infants in the planned early birth group were admitted to an intermediate care nursery (RR 1.28, 95% CI 1.02 to 1.61). The gestational age at birth was an average of 10 days earlier in women randomised to immediate birth.

A retrospective cohort study by Rabinovich et al. [188] compared outcomes for IOL (n = 1428) versus EM (n = 804) between 34 and 38 weeks for IUGR. This study found that the IOL group had lower stillbirth and neonatal death rates (p < .001), higher 1 and 5 min Apgar scores and higher vaginal birth rates. IOL at 37 weeks protected from stillbirth but not from adverse composite neonatal outcomes. A retrospective cohort study by Brzezinski-Sinai et al. [187] compared outcomes for women with isolated oligohydramnios between 34 and 36.6 weeks who went into labour spontaneously (n = 33) versus those who underwent IOL (n = 111). Spontaneous labour was associated with statistically significant higher rates of CS (p < .001), as well as higher rates of maternal infection, chorioamnionitis, and transitory tachypnoea of the newborn. This study concluded that IOL may be beneficial to both the neonate and the mother; however some caution needs to be used interpreting these findings as the study was assessed as of moderate risk of bias.