The effect of dexamethasone on labor induction: a systematic review

Background To evaluate the effect of dexamethasone administration on the interval between initiation of labor induction and active phase of labor. Methods The databases including PubMed, Cochrane Library, Embase, Scopus and Web of Science were searched for studies published up to June 27, 2021. Two types of articles were included: a) full-text articles published in English or any other languages, and b) Randomized Controlled Trials (RCTs). Participants were primi- or multigravida women with term or post-term pregnancy. The intervention group received parenteral or extra-amniotic dexamethasone whereas the control group received normal saline or no treatment before initiation of labor induction. All data were analyzed using Review Manager 5.3. Results Seventeen studies involving 1879 patients were included in the meta-analysis. Administration of dexamethasone reduced the interval between the initiation of labor induction and the beginning of active phase by about 70 min [MD: - 1.17 (− 1.37, − 1.00); P < 0.00001]. Duration of the first stage of labor in the dexamethasone group was about 88 min shorter than that in the control. There were no maternal and fetal adverse effects. Conclusions Dexamethasone could significantly reduce the length of induction-active phase interval, and length of the first stage of labor, with no difference in maternal or fetal adverse effects. Supplementary Information The online version contains supplementary material available at 10.1186/s12884-021-04010-1.


Background
Induction of labor refers to the process of stimulating contractions before the spontaneous onset of labor, either with or without ruptured membrane. Labor induction can be considered as a therapeutic intervention when the benefits of precipitous delivery to either the mother or the fetus are greater than the risks of pregnancy continuation [1]. Induction of labor is the most common obstetric procedure and the fastest-growing medical technique in the United States [2]. Most recently, the prevalence of induction has been estimated to be 23.3% in the United States, 22.4% in Ethiopia, and 9% in Sudan [3][4][5]. Rupture of the membranes without labor, gestational hypertension, oligohydramnios, nonreassuring fetal status, post-term pregnancy, and various maternal medical conditions such as chronic hypertension and diabetes are the common cause of start termination of pregnancy [6]. Several medical and non-medical methods are used for the induction of labor [7,8]. While oxytocin is the most conventional method, there are other effective methods including the use of prostaglandins such as misoprostol and dinoprostone as well as the application of some mechanical methods like stripping of membranes, artificial rupture of membranes, extraamniotic saline infusion, trans cervical balloons, and hygroscopic cervical dilators [9]. The success of induction and progression of labor depends on several factors including multiparity, low body mass index (BMI), ruptured membranes, tall height, low estimated fetal weight, and absence of comorbidities associated with placental insufficiency (e.g., preeclampsia) [10].
According to a study by Kawakita et al., of nulliparous women who underwent induction of labor in their study, 77.4% had a vaginal delivery [11]. The time required for the induction of labor, especially when there is an unfavorable cervix, increases the risk of cesarean delivery [10]. Some studies have reported the role of administering glucocorticoids such as dexamethasone in cervical ripening [12]. Although the role of glucocorticoids in the initiation of labor is still unknown, there is evidence supporting the role of glucocorticosteroid receptors in the fetal membranes at the beginning of the labor process [13,14].
Corticotropin-releasing hormone (CRH) in the female reproductive system is the main regulator of the hypothalamic-pituitary-adrenal axis [12,15]. During pregnancy, the placenta and fetal membranes secrete large amounts of CRH. Also, CRH levels in plasma increase during labor, with the maximum value at vaginal delivery [16]. There is considerable evidence for the effect of dexamethasone on the duration of labor stages. The aim of the current systematic review was to examine the effect of dexamethasone on the length of the interval between the initiation of labor induction and the beginning of the active phase of labor. Moreover, the length of different stages of labor and maternal and neonatal outcomes were investigated.

Methods
Systematic Reviews and Meta-Analyses of Randomized Controlled Trials (RCTs) were used to conduct this systematic review [17].

Inclusion and exclusion criteria
Two types of articles were included in this review: a) full-text articles published in English or any other languages, and b) Randomized Controlled Trials (RCTs). Abstracts, comments, letters to editor, and observational studies were excluded.

Participants
The criteria for including the participants were as follows: an unfavorable cervix with a Bishop score of ≤4; singleton pregnancy with a duration of at least 37 weeks according to a reliable date for the last menstrual period and a first trimester ultrasound evaluation; cephalic presentation and intact membrane; and normal amniotic fluid. Women with any of the following conditions were excluded from the study: uterine malfunction, macrosomia, placenta previa or placenta abruption risk, history of surgery on uterus, uterine contractions, fetal distress, and fetal occiput posterior position.

Types of interventions
The included studies involved intervention groups receiving dexamethasone before initiation of labor induction by any route of administration (intramuscular, intravenous or extra-amniotic) compared with control groups (placebo or no intervention).

Types of outcome measures
The primary outcomes were as follows: The interval between initiation of induction of labor and active phase of labor, length of different stages of labor, the interval between initiation of induction of labor and delivery and Bishop Score after induction. Two types of secondary outcomes were as follows: Maternal and neonatal outcomes. The maternal outcomes were Caesarean section rate, and adverse events. Neonatal outcomes included Apgar score at the 1st and 5th minutes after birth, fetal distress, meconium-stained liquid, and admission to neonatal intensive care unit (NICU).

Study selection
Following the initial search in the targeted databases, two of the authors (ZM) and (SH F), independently screened titles and abstracts of the search results. Full text screening was conducted by the same two authors. Covidence was used for all screening, data extraction, and quality assessment. Any probable disagreement was resolved by discussion or asking assistance from the third author (E.K).

Data extraction
The authors used Covidence for data extraction and entered the data into Review Manager Software (RevMan 5.3). Details of the studies including the design of the study, inclusion and exclusion criteria, baseline characteristics, interventions, and outcomes were extracted by two of the authors (ZM and SH F), independently.

Assessment risk of bias in included studies
The risk of bias for each study was independently assessed by two reviewing authors (ZM and SH F) who used seven criteria suggested by Cochrane for the quality assessment of randomized controlled trials. These criteria included selection bias, performance bias, detection bias, attrition bias, selective reporting, and other risks of bias. If the authors had any discrepancy, they deliberated an issue to resolve it.

Statistical analysis
Different statistical procedures were taken for continuous and dichotomous data. Mean and standard deviation with 95% CIs were used for continuous data such as the interval between initiation of induction of labor and active phase of labor, the length of different stages of labor, Apgar score at the 1st and 5th minutes, and Bishop Score after induction. For dichotomous data, the results were presented as summary risk ratio or odds ratio (OR) with 95% confidence intervals. Outcome measurement in all trials was similar. To demonstrate the effect size and CI, Forest plots were used. Moreover, heterogeneity between the included studies was assessed by I 2 . By default, we used fixed effects for all pooled studies. If I 2 > 50%, the random effect model was used for the primary results of heterogeneity. Furthermore, sensitivity analyses were conducted to discover the potential source of heterogeneity if it was statistically significant across the studies. The authors performed sensitivity analyses by sequentially omitting one single study each time to test the robustness of uncertainty in the meta-analysis. Finally, all the data were analyzed using Review Manager (RevMan 5.3) statistical software from the Cochrane group. The significance level was set at 0.05 for random effects and fixed effects.

Subgroup analysis and investigation of heterogeneity
The following subgroup analyses were carried out: parenteral versus extra-amniotic administration of dexamethasone. Some outcomes were also used including the length of the interval between initiation of induction and delivery, mode of delivery, fetal distress, and admission to NICU. Figure 1 Shows the flowchart of the selection process of studies. In our search of databases, 2672 articles were obtained as follows: Cochrane Library (n = 80), Scopus (n = 493), Web of Science (n = 178), PubMed (n = 1119), and EMBASE (n = 802). After removing duplicates (n = 949) using Covidence, 1723 papers were screened of which, 26 eligible articles were selected for full-text screening, but we could not have access to the full-text of nine papers because these papers were published before 2000. Therefore, 17 articles were finally included in the study

Characteristics and quality assessment of studies
Out of the 17 studies included in this review, four investigated the effect of dexamethasone as an extra-amniotic infusion with a Foley catheter [18][19][20][21], and the rest assessed the effect of parenteral injection of dexamethasone on the duration of labor induction. We examined the outcomes of these two groups separately. Among these articles, eight studies were published in Egypt [12,[22][23][24][25][26][27][28], seven in Iran [7,19,20,[29][30][31][32], one in Iraq [18], and one in Israel [21]. All of them had an RCT design with fourteen using placebo and three having no intervention in their control groups. All of the studies assessed the effect of dexamethasone on nulliparous women except four studies that examined multiparous and nulliparous women [19][20][21]29]. The range of women's age was between 18 and 35 years, and the gestational age was 37-42 weeks. All papers except one [23] used oxytocin for labor induction about 6 to 12 h after dexamethasone injection. Two papers did not mention their induction protocol [19,30]. The number of participants in studies differed from 31 to 86 in each group. The characteristics of the studies are demonstrated in Table 1.
Quality assessment of papers was conducted by two reviewers (ZM, SHF) according to Cochran Risk of Bias tool. The result of the assessment is presented in Fig. 2. The lowest biases were related selection bias, reporting bias, and attrition bias, respectively. However, most of the detection bias and allocation concealment signifies in the unclear risk. In terms of other types of bias, 50 % of papers were in the low-risk zone and the others were in the high-risk zone.
Overall meta-analysis The interval between induction of labor and active phase of labor Ten articles including 1126 participants were analyzed in order to assess the effect of dexamethasone on the interval between the initiation of induction of labor and the active phase of labor. The result of the pooled analysis showed that the mean difference of this interval was shorter in the experimental group compared with the control group [MD: -0.96, CI 95% (− 1.06, − 0.85); P < 0.00001]. Because of high heterogeneity (I 2 = 74%; P < 0.0001), sensitivity analysis and random-effect analysis were done. By omitting one study [24], the heterogeneity reached 31%. However, this interval was about 70 min shorter in the intervention group compared with the control group [MD: -1.17, CI 95% (− 1.37, − 1.00); In the subgroup of extra-amniotic administration of dexamethasone with Foley catheter, two papers reported this outcome. There were 163 participants. The analysis showed that the length of the interval between induction  [I 2 = 97%; P < 0.00001]. However, after removing 2 studies during sensitivity analysis [23,26], the length of the first stage of labor in the dexamethasone group was nearly 88 min shorter than that in the control [MD: -1.47, CI 95% (− 1.78, − 1.16); P < 0.00001] [I 2 = 30%; P = 0.24].

Length of the second stage of labor
We assessed the length of the 2nd stage of labor in 10 articles involving 874 participants. The pooled analysis of these studies showed that the mean difference of the  second stage in experimental group is lower than that in the control group [MD: -11.98, CI 95% (− 12.81, − 11.15); P < 0.00001] [I 2 = 98%; P < 0.00001]. We conducted random-effect and sensitivity analysis to reduce heterogeneity. By omitting the effect of three articles [25,26,31], heterogeneity reached 0%, and the mean difference of the second stage in the experimental group was still lower than that in the control group [MD: -4.21, CI 95% (− 5.43, − 2.99); P < 0.00001]. That is, the second stage of labor was almost 4 min shorter in the experimental group (Fig. 7).

Length of the third stage of labor
The pooled analysis of nine papers with 1071 participants showed that although the length of the third stage in the experimental group was shorter than that in the control group based on the fixed-effect model [MD: -0.67, CI 95% (− 0.96, − 0.39); P < 0.00001] [I 2 = 98%; P <

Mode of delivery
This outcome was reported in two groups of induction of labor by parenteral dexamethasone and extra-amniotic injection of dexamethasone with a Foley catheter. Thus, we analyzed this outcome as a subgroup analysis.   Fig. 9).

Induction to delivery time interval
This outcome was reported in one study using parenteral dexamethasone and three using Foley catheter. In both subgroups, there was a significant difference between the intervention and control groups in terms of the length of induction to delivery time interval. In the

Neonatal outcomes
Fetal distress, Apgar score, meconium-stained liquid, and admission to the Neonatal Intensive Care Unit (NICU) were assessed in the included articles. None of the mentioned outcomes showed a statistically significant difference between the intervention and control groups in the subgroups of parenteral or extra-amniotic injection dexamethasone.

Neonatal Apgar at the 5th minute
Nine papers in the parenteral subgroup reported the 5th minute Apgar score and involved 991 participants. There were no significant differences between groups [MD: 0.09, CI 95% (0.00, 0.18); P = 0.05].

Maternal outcomes
Only three articles evaluated the mother's blood pressure (BP) as a maternal complication of parenteral injection of dexamethasone [22,26,32]. There were no significant differences between the intervention and control groups in terms of systolic BP

Discussion
This systematic review aimed to evaluate the effect of dexamethasone on labor induction. Our meta-analysis of 17 Randomized Controlled Trials (13 papers in the parenteral subgroup and four in the extra-amniotic injection subgroup) showed that the use of dexamethasone before the initiation of labor induction by any route of administration decreases the interval between induction and active phase, and the length of the first, second, and third stages of labor, and improves Bishop score after induction. We also found that dexamethasone injection does not affect maternal outcomes (caesarean section rate and blood pressure) or neonatal outcomes (Apgar score at 1st and 5th minute after birth, fetal distress, Meconium strained liquid, NICU admission). Physiological processes regulating childbirth represent a series of biochemical changes in the uterus and cervix that result from endocrine and paracrine signals from the mother and fetus [1]. The role of Corticosteroids in the onset of labor is well unknown. Investigations on animals have shown that secretion of cortisol from the maturing fetal hypothalamus-pituitary-adrenal axis is important in initiating labor. According to these studies, after injection of a glucocorticosteroid, preterm labor was observed in lamb fetuses [33]. Glucocorticoids, especially cortisol, increase amniotic cyclooxygenase, increase prostaglandin synthesis, inhibit the activity of the chorionic prostaglandin dehydrogenase, and inhibit prostaglandin metabolism [34,35]. Glucocorticoids are also strong stimulants in the production of placental CRH. In the second half of pregnancy, CRH levels in the mother's bloodstream gradually increase and peak in the last six to 8 weeks before delivery. Women with high plasma CRH levels give birth earlier while those with lower CRH levels give birth later, indicating the impact of CRH production as an important factor in the onset of labor.
CRH does not have a direct inotropic effect on the human myometrium, but it does cause uterine vasodilation by affecting the placenta locally. It can also stimulate the secretion of dehydroepiandrosterone sulfate (DHEA-S) in fetal adrenal cortex cells. CRH enhances the effects of estrogen on these tissues of the uterus and cervix, increases prostaglandins in the amniotic sac, chorion, and decidua, and enhances the effect of oxytocin [35]. Several studies have reported that the injection of corticosteroids provokes successful induction of labor in lambs and humans. Since glucocorticoid receptors are present in the amniotic sac, glucocorticosteroids could carry out a possible role in parturition through paracrine or autocrine mechanisms [36]. Kalantaridou et al. (2007) reported that the corticotrophin-releasing hormone (CRH) is the main corrector of the hypothalamic-pituitary-adrenal axis. Circulating placental CRH is responsible for the physiologic hypercortisolism of the second half of pregnancy and plays a role in the commencement of labor [15].
In humans, the production of CRH by the placenta and the increase of this hormone in maternal plasma are associated with the timing of parturition [37]. Recently, it has been shown that CRH stimulates the placenta for the production of estrogens and inhibits the production of progesterone [38]. Increasing the ratio of estrogen to progesterone in the mother's serum may progress the expression of contraction-associated proteins in the myometrium, thus facilitating the beginning of parturition [1,39]. Furthermore, glucocorticoids induce the production of CRH by the placenta and the production of prostaglandins (PGF2 and PGE2) by fetal membranes [40].

Limitations of this study
Several limitations existed in this meta-analysis: 1) publication bias was not been measured; 2) Although most studies included in this review had examined the effect of dexamethasone on nulliparous women, some had not examined this effect on a parity basis.; 3) Most studies were conducted in Iran and Egypt, and the number of articles from other countries was small; 3) Maternal and neonatal outcomes had not been reported in all included studies completely. These limitations could have contributed to heterogeneity. On the other hand, given the possible effect of dexamethasone on neonatal outcomes [41,42], none of the studies examined other outcomes such as neonatal hypoglycemia and the consequences of neurological and behavioral development in neonates born to these mothers.

Conclusion
Dexamethasone could significantly reduce the length of the time interval between induction and active phase and length of the first stage of labor with no difference in maternal or fetal adverse effects. Considering the high heterogeneity and quality of the studies included in this review, high-quality double-blind clinical trials are needed to be included in future reviews in order to draw more solid conclusion in this regard.