A systematic scoping review of clinical indications for induction of labour

Background The proportion of women undergoing induction of labour (IOL) has risen in recent decades, with significant variation within countries and between hospitals. The aim of this study was to review research supporting indications for IOL and determine which indications are supported by evidence and where knowledge gaps exist. Methods A systematic scoping review of quantitative studies of common indications for IOL. For each indication, we included systematic reviews/meta-analyses, randomised controlled trials (RCTs), cohort studies and case control studies that compared maternal and neonatal outcomes for different modes or timing of birth. Studies were identified via the databases PubMed, Maternity and Infant Care, CINAHL, EMBASE, and ClinicalTrials.gov from between April 2008 and November 2019, and also from reference lists of included studies. We identified 2554 abstracts and reviewed 300 full text articles. The quality of included studies was assessed using the RoB 2.0, the ROBINS-I and the ROBIN tool. Results 68 studies were included which related to post-term pregnancy (15), hypertension/pre-eclampsia (15), diabetes (9), prelabour rupture of membranes (5), twin pregnancy (5), suspected fetal compromise (4), maternal elevated body mass index (BMI) (4), intrahepatic cholestasis of pregnancy (3), suspected macrosomia (3), fetal gastroschisis (2), maternal age (2), and maternal cardiac disease (1). Available evidence supports IOL for women with post-term pregnancy, although the evidence is weak regarding the timing (41 versus 42 weeks), and for women with hypertension/preeclampsia in terms of improved maternal outcomes. For women with preterm premature rupture of membranes (24–37 weeks), high-quality evidence supports expectant management rather than IOL/early birth. Evidence is weakly supportive for IOL in women with term rupture of membranes. For all other indications, there were conflicting findings and/or insufficient power to provide definitive evidence. Conclusions While for some indications, IOL is clearly recommended, a number of common indications for IOL do not have strong supporting evidence. Overall, few RCTs have evaluated the various indications for IOL. For conditions where clinical equipoise regarding timing of birth may still exist, such as suspected macrosomia and elevated BMI, researchers and funding agencies should prioritise studies of sufficient power that can provide quality evidence to guide care in these situations.


Introduction
Induction of labour (IOL) has been on the rise over recent decades [1,2], with significant variation within countries and between hospitals [1,3,4]. In Australia, the IOL rate rose from 27.3% in 2012 to 31.1% in 2016 [1].
IOL is generally undertaken with the aim of decreasing maternal and/or fetal morbidity or mortality i.e. when the risks of continuing the pregnancy to either mother or fetus are considered greater than the risks associated with planned birth [5]. For example, women are commonly induced for post-term pregnancy to reduce the risk of stillbirth [6]. Women with premature rupture of membranes are induced to decrease incidence of maternal sepsis and neonatal infection secondary to chorioamnionitis [7], women with preeclampsia to reduce the risk of stillbirth and severe maternal morbidity (renal failure, liver failure, coagulopathy, pulmonary oedema, eclamptic seizures) [8] and women with diabetes are induced to minimise macrosomia-associated birth complications and stillbirth risk [9].
While some indications for IOL are supported by high level evidence, others are not [10]. A systematic review of the evidence of indications by Mozurkewich et al. [10] conducted in 2008 found that the evidence at the time was insufficient to support IOL for women with common indications such as diabetes, twin gestation, suspected fetal macrosomia and oligohydramnios. The review called for further research to obtain a clearer picture of the risks and benefits associated with IOL [10].
In the 10 years since the initial review by Mozurkewich et al. [10], there continues to be debate about the acceptable use of IOL. There is no agreed external standard [11], and clinical guidelines vary considerably, both nationally and internationally [12][13][14][15][16][17]. The recent ARRIVE trial [18], which compared outcomes for low risk nulliparous women associated with IOL at 39 weeks (between 39+0 and 39+4) versus expectant management, seems to have further divided the maternity community in relation to IOL timing [19]. While this trial did not find any differences between the two groups for its primary outcome, that is, a composite of perinatal death and severe neonatal complications, it did find that IOL was associated with a reduction in caesarean section (CS) rate by 4%. This is at odds with some population studies that show that IOL is associated with a rise in CS rate [20]. Regardless of whether IOL is associated with a rise or reduction in CS rates, it is associated with increasing rates of early term birth [21], and its negative impact on child development [22]. Furthermore, IOL is often associated with a less positive birth experience for women compared to spontaneous onset of labour [23][24][25]. As such, the circumstances in which to offer a woman an IOL should be informed by the best available evidence.
The aim of this scoping review was to map the evidence in relation to indications for IOL and determine which are supported by evidence and where knowledge gaps exist. By building on the review by Mozurkewich et al. [10], this study presents a comprehensive overview of the available evidence to date.

Method
We undertook a systematic scoping review [26][27][28][29], informed by the method used by Mozurkewich et al. [10], and following the PRISMA-ScR reporting guidelines for systematic reviews as outlined in our protocol developed before the review commenced. The aim of a scoping review is to map the literature relevant to a broad research topic to gain insight into the nature of the evidence and identify research gaps [26,27,29].

Inclusion and exclusion criteria
We included quantitative studies that explored common indications for IOL, specifically: postterm 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. 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.

Results
Our search identified 2554 papers for screening, of which 272 articles were identified for full text review. We also reviewed the 34 studies identified by Mozurkewich et al. [10], of which 28 were included for full text review (we could not access full text for the remaining six). In total, 300 full text studies were reviewed, of which 68 were included (see Tables 1 and 2). Only six of the studies included in the review by Mozurkewich et al. [10] were included here, as the remaining were studies already included in more recent systematic reviews, or were systematic reviews that had been superseded by more recent reviews.
Included studies are listed in Tables 1 and 2. We did not identify any studies for inclusion in relation to fetal alloimmune disease or Rh disease, antepartum haemorrhage, (suspected) chorioamnionitis, or fetal death and only one study in relation to 'other' maternal medical indications, about maternal cardiac disease. The evidence for each indication for IOL is summarised below.

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][38][39][40][41][42][43][44][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 . 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][79][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       Summary statement. Women having IOL beyond 41-42 weeks is associated with fewer perinatal deaths and reduced CS rates, even though the number needed to treat to prevent perinatal mortality is high (approx. 450).  (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).
Summary statement. Early birth for PROM at term may help reduce maternal and neonatal infections without increasing CS rates.
Early birth for pre-term PROM increases the risk of infant death after birth, respiratory problems and NICU admissions, and CS rates, and is associated with a decreased incidence of chorioamnionitis.

Hypertension/preeclampsia
Fifteen studies in relation to preeclampsia, chronic or gestational hypertension were included. There were three Cochrane reviews [118][119][120] 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 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][140][141][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) [ ] 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.
Summary statement. RCT evidence suggests decreased maternal morbidity after IOL for preeclampsia from 34 weeks gestation, however at the cost of increased neonatal morbidity if undertaken at 34-37 weeks. There is little agreement on the timing of birth for women with chronic hypertension or gestational hypertension at term, but there is some evidence that indicates that planned birth between 38 and 39 weeks is associated with the lowest maternal and neonatal morbidity/mortality. EM for severe preeclampsia remote from term increases birthweight and reduces neonatal morbidity.

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][148][149][150][151][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][150][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.
Summary statement. There was little quality evidence to inform management between IOL at term or EM for women with diabetes during pregnancy, and the little evidence that was available was largely limited to GDM. Only one relevant study included women with pre-existing diabetes (Type I and Type 2), consisting of only 13 women.

Twin pregnancy
We identified five studies in relation to twin pregnancy, consisting of a Cochrane review [156], and four retrospective cohort studies [157][158][159][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. Summary statement. While some cohort studies found that IOL in twin pregnancies increases the risk of CS compared to spontaneous labour onset, other studies found the reverse. Evidence from two RCTs (included in a Cochrane review) found non-significant improvements in composite neonatal and maternal outcomes with planned birth for twins at 37 week.

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.
Summary statement. The evidence is mixed. One RCT found that early planned birth for ICP was not associated with improved outcomes, however this study was underpowered to detect clinically important differences. Evidence from retrospective cohort studies suggests that planned early birth was associated with a significant reduction in the incidence of stillbirths, and that planned birth at 36 weeks gestation was associated with lower perinatal mortality.
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).
Summary statement. The evidence is mixed and from retrospective cohort studies only. While some studies indicated that IOL for a high BMI was associated with reduced CS rates and improved maternal and neonatal outcomes, other studies demonstrated the reverse.

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.
Summary statement. Evidence from one RCT indicated that IOL does not improve outcomes or CS rates for women greater than 35 years, however this study was underpowered to identify the effect of IOL on perinatal death. Evidence from a retrospective cohort study suggested that IOL at 40 weeks reduces perinatal mortality.

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][179][180][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 thirdand 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.
Summary statement. Evidence from four RCTs included in a Cochrane review indicates that there appears to be little difference between IOL versus EM in terms of maternal and neonatal outcomes for women with suspected macrosomia.

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.
Summary statement. There was no evidence to support IOL for women who have pregnancies 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][191][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.
Summary statement. For preterm babies with suspected compromise and uncertainty about whether to plan birth early or not, there appears to be no benefit to immediate birth. However, included studies were largely underpowered and had different definitions of what is considered fetal compromise.

Discussion
The majority of indications for IOL are not supported by strong evidence. While there is high quality evidence in relation to IOL for post-term pregnancy, hypertension/preeclampsia and PROM, for all other indications there were conflicting findings and/or insufficient power to provide definitive evidence. A summary of the evidence and recommendations for future research are included in Table 3.
We did not identify any studies for inclusion in relation to fetal alloimmune disease, antepartum haemorrhage, chorioamnionitis (including suspected), or maternal mental health indications, and only one study in relation to maternal cardiac disease, which did not identify any adverse nor beneficial effects associated with IOL. This may be due to the low incidence of these issues and therefore the challenges in undertaking research. Clinical judgement rather than high level evidence will need to continue to drive practice.
The uncertainty in the evidence identified by this review raises questions about the implications for evidence in practice and the development of guidelines. While variations in clinical practice is often attributed to suboptimal guideline adherence [202][203][204], there is an increasing recognition that this may also be due to a shortage of clear clinical guidelines that provide consistent recommendations that are evidence based [203,205,206]. Specific to IOL a recent review of guidelines identified significant variation across guidelines in what are considered acceptable indications for IOL [17]. This variation can be understood in light of the evidence gaps in this space.
While further high quality evidence needs to be developed, there are challenges associated with RCTs, in terms of recruitment and funding. Until this evidence is available, there is a need to develop a better understanding of how to provide evidence-based care when the evidence is unclear. Clinical decision making should be informed by evidence from RCTs and cohort studies, but evolve to include the women's experience of care and preferences and include a process of shared-decision making. While the importance of shared decision making between women and practitioners is increasingly recognised [207,208], how shared decision making is best performed in the context of uncertainty or ambiguity remains less clear [209]. To support and inform evidence based care, more research is needed into shared decision making in a context of uncertainty. Furthermore, the experiences and preferences of women is largely absent from this literature, and RCTs that include women's experience of care as an outcome are required.
This review presents a comprehensive overview of the literature in relation to IOL. However, we could not access the full text of a small number of articles, and we did not include studies that were not in English that may have been relevant to our findings. Furthermore, in accordance with the scoping methodology [29,210], the quality assessment of the included studies was minimal and largely limited to an assessment of bias. Strengths of this study were that two authors reviewed each of the articles and recognised tools for data extraction and bias detection were used.

Conclusion
A large proportion of pregnant women have IOL at or near term, sometimes for indications that are well supported by evidence, and sometimes not. This study systematically mapped the available evidence for common indications for IOL. While for some indications, IOL is highly recommended, a number of common indications do not have strong supporting evidence. Overall, few RCTs have evaluated the various indications for IOL, and researchers and funding agencies should prioritise studies of sufficient power that can help to guide care in these situations. However, the entrenched nature of some IOL indications even when not well supported by evidence does present practical difficulties to RCT recruitment. Women should be provided Table 3. Evidence summary and recommendations for future research.

Post-term pregnancy
Evidence: 1. IOL versus EM, beyond 41-42 weeks is associated with fewer perinatal deaths and reduced CS rates. Therefore, although the number needed to treat is high for prevention of perinatal mortality, at approximately 450 IOL for every death prevented [6], and rate of assisted vaginal birth increased, post-term pregnancy is likely to remain a routine IOL indication. 2. The available evidence for optimal timing of IOL (41+ versus 42+) remains limited.
Future directions: 1. More studies with an adequate sample size would be ideal to improve the granularity of the data at different gestations. While further research is required, enthusiasm for post-term pregnancy trials may be low after the recently published ARRIVE RCT, where reduction in CS rates (and no evidence of maternal or infant harm) was found in women undergoing IOL at 39+ weeks versus EM [18]. Furthermore, given that offering IOL at 41+ weeks is ingrained into policy and practice in most high-income settings [13,16,194] it is likely any sufficiently powered RCT would need to be conducted in a low-to-middle-income country setting. These findings may then not necessarily be applicable to high-income settings.

PROM at term
Evidence: 1. Evidence from a Cochrane review that included 23 RCTs indicates that planned early birth may help reduce maternal and neonatal infections without increasing CS rates. The quality of the evidence was low due to many studies being at high risk of bias.
Future directions: 1. Ideally further research to assess the benefits versus harms of planned early birth would be performed, particularly in the early term (37-38 week) group. However, evidence that risk of chorioamnionitis increases from approximately 12 hours after term PROM [195], and associations of chorioamnionitis with the serious long-term morbidity of cerebral palsy, in addition to the short term morbidities assessed in the included studies [196], means there is unlikely to be clinical equipoise to perform such trials.

Pre-term PROM
Evidence: 1. Evidence from a Cochrane review that included 12 RCTs indicated that early birth increased the risk of infant death after birth, respiratory problems and NICU admissions, and CS rates without a clinically important difference in the incidence of neonatal sepsis. Early birth was associated with decreased incidence of chorioamnionitis. 2. For women with PROM <37 weeks with no contraindications to continuing the pregnancy, a policy of EM with careful monitoring was associated with better perinatal outcomes.
Future directions: 1. More research that explores the risk benefit ratio of early birth (late preterm) on long term developmental outcomes is required.

Hypertension
Evidence: 1. There was limited high quality evidence to inform decisions about optimal timing of birth. There was little or no agreement on the timing of birth for women with chronic hypertension, gestational hypertension or mild preeclampsia at term. Some evidence indicated that planned birth between 38 and 39 weeks was associated with the lowest maternal and neonatal morbidity/mortality for both women with gestational hypertension and those with chronic hypertension. 2. In preeclampsia or gestational hypertension, maternal morbidity was lower in RCTs comparing immediate (versus delayed) birth anytime from 34 weeks, however infant morbidity may be higher, particularly prior to 37 weeks. 3. The vast majority of studies regarded severe preeclampsia remote from term. There was evidence that indicated that EM for severe preeclampsia remote from term increases birthweight and reduces neonatal morbidity. 4. In relation to IOL versus CS, evidence indicated that while IOL is associated with high rates of CS, it is not associated with increased harm and should be considered a reasonable option.
Future directions: 1. Overall the strength of the evidence was weak, particularly regarding preterm preeclampsia and timing of birth at term for chronic hypertension and gestational hypertension, and more research is needed.
(Continued ) 1. There was little quality evidence to inform management between IOL at term or EM, and the little evidence that was available was limited to GDM. Only one relevant study included women with pre-existing diabetes (Type I and Type 2), consisting of only 13 women.
Future directions: • Further prospective cohort and RCT studies are required.

Twin pregnancies
Evidence: • The existing evidence does not definitively indicate that early planned birth for uncomplicated twin pregnancy improves outcomes. • While some cohort studies found that IOL in twin pregnancies increases the risk of CS compared to spontaneous labour onset, other studies found the reverse. • Evidence from two RCTs found non-significant improvements in composite neonatal and maternal outcomes with planned birth at 37 weeks.
Future directions: 1. More research is required. However, given population data suggests increased stillbirth risk in twin pregnancy beyond 37 weeks, there is unlikely to be uptake for trials of birth versus EM at later gestations [156]. Complicating interpretation of the twin pregnancy evidence is the non-reporting, missing data, or lack of power of the underlying studies to examine monochorionic (MCDA) versus dichorionic (DCDA) data. MCDA twins have substantially different risks to DCDA, including consequences of late intrauterine death of one twin on the other twin, making grouping of MCDA and DCDA in timing of birth studies problematic [197].

Intrahepatic cholestasis of pregnancy
Evidence: 1. Evidence from one RCT indicated that early planned birth was not associated with improved outcomes, however this study was underpowered to detect clinically important differences. 2. Evidence from retrospective cohort studies suggested that planned early birth was associated with a significant reduction in the incidence of stillbirths, and that planned birth at 36 weeks gestation was associated with lower perinatal mortality.
Future directions: 1. Further well-conducted cohort studies or RCTs of early term (37-38 weeks) versus full-term (39-40 weeks) birth are recommended. Timing of birth RCTs may not be ethical in the subgroup of women with high bile acids (>40μmol/L), where there is some cohort data to suggest an increased stillbirth risk directly related to ICP [198,199], and would certainly be hard to justify in those with bile acid levels >100 μmol/L where cohort evidence for increased stillbirth risk is strong [200]. However for the majority of cholestasis cases, where bile acid levels are 10-40 μmol/L, an elevated stillbirth risk above background is not proven [199,200], and early term birth with its potential adverse infant neurodevelopmental sequelae may not be justified [201].

Maternal elevated BMI
Evidence: 1. Evidence from retrospective cohort studies presented mixed findings. While some studies indicated that IOL was associated with reduced CS rates and improved maternal and neonatal outcomes, other studies demonstrated the reverse.
Future directions: 1. Further prospective cohort studies and RCTs are required.

Maternal age
Evidence: 1. Evidence from one RCT indicated that IOL does not improve outcomes or CS rates for women greater than 35 years, however this study was underpowered to identify the effect of IOL on perinatal death. 2. Evidence from a retrospective cohort study suggested that IOL at 40 weeks reduces perinatal mortality. with the best available evidence to help them make an informed choice about the risks and benefits of IOL. Clinicians should use the best available evidence to inform decision making and acknowledge when insufficient evidence is available.
Supporting information S1 Checklist. Preferred reporting items for systematic reviews and meta-analyses extension for scoping reviews (PRISMA-ScR) checklist.

Suspected macrosomia
Evidence: 1. Evidence from four RCTs included in a Cochrane review indicates that there appears to be little difference between IOL versus EM in terms of maternal and neonatal outcomes, but most included studies were underpowered. 2. Evidence from cohort studies present mixed findings, with some indicating that IOL is associated with a reduction in CS rates, and some indicating IOL is associated with increased CS rates.
Future directions: 1. Further adequately powered research is required.

Fetal gastroschisis
Evidence: 1. Neither the Cochrane review (with one RCT included) nor the retrospective cohort study found any significant benefits or adverse effects associated with IOL. 2. While the RCT was underpowered, these findings support EM in pregnancies that are complicated by fetal gastroschisis.
Future directions: 1. As more women globally are exposed to antenatal ultrasound, more babies may be identified in utero with fetal gastroschisis. Therefore, further research is necessary to determine the impact of IOL especially in low to middle income settings.

Fetal compromise
Evidence: 1. Evidence from RCTs included in Cochrane reviews indicates that for preterm babies with suspected compromise and uncertainty about whether to plan give birth early or not, there appears to be no benefit to immediate birth. However, these studies were largely underpowered.