The ARRIVE trial (A Randomized Trial of Elective Induction Versus Expectant Management) tested the hypothesis that elective induction of labor at 39 weeks of gestation compared with expectant management among low-risk nulliparous women reduces the risk of a composite of perinatal mortality and severe neonatal morbidity. The primary perinatal outcome in the ARRIVE trial was not significantly different between the induction of labor and expectant management groups (relative risk [RR] 0.80; 95% CI 0.64–1.00), whereas the major secondary outcome of cesarean delivery was significantly less common in the induction of labor group (RR 0.84; 95% CI 0.76–0.93).1
In the wake of the ARRIVE trial, it is theorized that outcomes with induction of labor at 39 weeks of gestation and expectant management may be different in certain subgroups of low-risk women. Characteristics such as maternal age, maternal weight, or Bishop score could be used to better identify which low-risk women at 39 weeks of gestation are at greatest risk of adverse outcomes, and therefore might preferentially benefit from labor induction or from expectant management. Our objective in this secondary analysis was to evaluate the association between patient characteristics and maternal and perinatal adverse outcomes among women who underwent intended elective labor induction at 39 weeks of gestation or expectant management. We also wanted to determine whether a multivariable model could be developed to predict outcome based on characteristics.
We performed a secondary analysis of the ARRIVE trial of the Eunice Kennedy Shriver National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Women eligible for enrollment in the initial trial were low-risk nulliparous women with a live, nonanomalous singleton fetus in a vertex presentation, and who had no contraindication to vaginal delivery, no planned cesarean delivery, and no indication for induction before 40 5/7 weeks of gestation. Eligible women who gave consent were randomized between 38 0/7 weeks and 38 6/7 weeks of gestation to either induction of labor or expectant management. Women in the induction group were assigned to undergo induction of labor at 39 0/7–39 4/7 weeks of gestation; women in the expectant management group were asked to forego elective delivery before 40 5/7 weeks of gestation and to have delivery initiated no later than 42 2/7 weeks. Full descriptions of the eligibility criteria and study procedures of the initial trial are published elsewhere.1 The study protocol was reviewed and approved by all participating institutions’ review boards. Written informed consent was obtained from all participants before randomization.
To estimate the associations between patient characteristics and adverse outcomes for women for whom these associations would be relevant in an actual clinical setting, the current analysis excluded women in the trial whose labor, induction or cesarean delivery began before 39 0/7 weeks of gestation, even if the actual delivery was at 39 0/7 weeks of gestation or later. Because an inclusion criterion for the trial was nonanomalous fetuses, and this secondary analysis evaluated a perinatal outcome, the 12 enrolled pregnancies with anomalies detected after randomization were also excluded from this analysis. Women whose management and delivery were not compliant with the intended intervention were also excluded. Therefore women assigned to induction were excluded if their delivery began after 39 4/7 weeks of gestation unless the delay was due to a newly developed medical indication (eg, awaiting resolution of active herpes) and women assigned to expectant management were excluded if an induction without a medical indication was initiated before 40 5/7 weeks of gestation. Women in either group were excluded if they had elective cesarean delivery without labor.
Outcomes evaluated included a composite of perinatal mortality and severe perinatal morbidity, third- or fourth-degree perineal lacerations, and postpartum hemorrhage. The perinatal composite outcome for this analysis was the same as the primary outcome of the parent trial, defined as one or more of the following: perinatal death, the need for respiratory support within 72 hours after birth, 5-minute Apgar score 3 or less, hypoxic–ischemic encephalopathy, seizure, infection (confirmed sepsis or pneumonia), meconium aspiration syndrome, birth trauma (bone fracture, neurologic injury, or retinal hemorrhage), intracranial or subgaleal hemorrhage, or hypotension requiring vasopressor support. Postpartum hemorrhage was defined as any of the following: use of two or more uterotonics (methergine, prostaglandin F2-alpha, prostaglandin E1) other than oxytocin, blood transfusion, or other surgical interventions for bleeding such as uterine compression sutures (B-Lynch stitch), uterine artery ligation, vascular embolization, hypogastric artery ligation, uterine balloon or packing, curettage, or nonelective hysterectomy.
In addition to treatment group, 11 baseline maternal characteristics that could be identified before induction or expectant management undertaken at 39 weeks of gestation were evaluated: maternal age, self-reported race–ethnicity, body mass index (BMI, calculated as weight in kilograms divided by height in meters squared) at randomization, marital status, employment status, insurance type, smoking, drinking, previous pregnancy loss (at less than 20 weeks of gestation), assisted conception history, and modified Bishop score at randomization.
We compared differences in study outcomes by treatment group and patient characteristics using the χ2 test for categorical variables and the Wilcoxon rank sum test for continuous variables. Log-binomial multivariable regression models were used to estimate the RRs and 95% CIs. Only those patient characteristics that showed a significant association with outcome in univariate analysis (P<.05) were included in the multivariable models. Backwards selection was used to retain significant characteristics. To account for patient clustering within a hospital, random effects for hospital were included in the final models. Interactions were then assessed between each patient characteristic and treatment group, with interactions retained if P<.05. Area under the curve (AUC) was estimated for each final model with and without random effects for hospital. Area under the curve values from the receiver operating characteristic curves used to characterize the classification ability of the models were as follows: 0.5 was considered no better than chance, greater than 0.5 to less than 0.7 poor, 0.7 to less than 0.8 acceptable, 0.8 to less than 0.9 excellent, 0.9 or higher outstanding.2 All statistical analyses were performed using SAS 9.4; all tests were two-sided, with P<.05 considered statistically significant. No imputation for missing data was performed.
Of the 6,106 women enrolled in the ARRIVE trial, 6,096 (99.8%) had delivery data. Of these, 5,007 (82.1%) achieved 39 weeks of gestation, had no fetal anomalies and adhered to the study protocol (Fig. 1). The maternal characteristics for those included in the analysis and those excluded because they were not delivered per protocol are presented in Appendix 2, available online at http://links.lww.com/AOG/C13.
Overall, the perinatal composite occurred in 252 (5.0%) newborns, 166 (3.3%) women had third- or fourth-degree perineal lacerations, and 237 (4.7%) deliveries were complicated by postpartum hemorrhage. The perinatal composite outcome occurred in 103 (4.1%) women in the induction of labor group compared with 149 (6.0%) women in the expectant management group (P=.003). Conversely, frequencies of third- or fourth-degree perineal laceration and postpartum hemorrhage were similar between women in the induction of labor and expectant management groups (laceration: 3.5% vs 3.2%, P=.53; postpartum hemorrhage: 4.7% vs 4.8%, P=.94). Outcomes were similar between those included in the analysis and those excluded because they were not delivered per protocol (perinatal composite 5.0% vs 5.3%, P=.82; third- or fourth-degree perineal laceration 3.3% vs 2.8%, P=.62; postpartum hemorrhage 4.7% vs 3.8%, P=.42).
The final unadjusted and multivariable models showing the characteristics significantly associated with each outcome are presented in Table 1. Intended elective induction of labor at 39 weeks of gestation was associated with a reduced risk of the perinatal composite outcome (4.1% vs 6.0%; adjusted relative risk [aRR] 0.71; 95% CI 0.55–0.90) and BMI (per unit increase) was associated with an increased risk of the perinatal composite outcome (aRR 1.04; 95% CI 1.02–1.05). A decreased risk of third- or fourth-degree perineal laceration was observed with increasing BMI (per unit increase) (aRR 0.96; 95% CI 0.93–0.98), and in Black women (compared with White women) (1.2% vs 3.9%; aRR 0.34; 95% CI 0.19–0.60). A higher risk of third- or fourth-degree perineal laceration was observed in Asian women in the unadjusted model, but not in the adjusted model. Hispanic women (compared with White women) were at an increased risk for postpartum hemorrhage (6.3% vs 4.0%; aRR 1.64; 95% CI 1.18–2.29). Patient characteristics associated with adverse outcomes were similar between treatment groups (P for interaction >.05; treatment group by BMI interaction for the perinatal outcome P=.32; treatment group by BMI interaction for third- or fourth-degree perineal lacerations P=.35; treatment group by race and ethnicity interaction for third- or fourth-degree perineal lacerations ranged from P=.47 to P=.88; treatment group by race and ethnicity interaction for postpartum hemorrhage ranged from P=.09 to P=.53).
As demonstrated by the AUC values of the receiver operating characteristic curves, the final multivariable models (without the random effects of hospital) were poor at predicting adverse outcomes (AUC for the perinatal composite 0.58, 95% CI 0.55–0.62; AUC for third- or fourth-degree laceration 0.64, 95% CI 0.60–0.68; AUC for postpartum hemorrhage 0.56, 95% CI 0.52–0.59). The AUCs remained weak when the random effects of hospital were added to the models (perinatal composite 0.70, 95% CI 0.67–0.73; third- or fourth-degree laceration 0.67, 95% CI 0.63–0.71; postpartum hemorrhage 0.67, 95% CI 0.63–0.71).
We found that few baseline patient variables were significantly associated with adverse outcomes in low-risk nulliparous women who achieved 39 weeks of gestation. In this secondary analysis of the ARRIVE trial, women were included only if the delivery process had not started by 39 weeks of gestation and subsequently were managed consistently with their randomly assigned approach to delivery (ie, intended 39-week labor induction or expectant management). Such a cohort composition is necessary if the results are to provide insight into the fundamental questions at hand: for women at 39 weeks of gestation who have the choice to undergo immediate labor induction or expectant management, are there particular individual characteristics that could provide insight into whether they may be at particular risk of adverse outcomes and is pursuit of one strategy better than the other? The patient characteristics selected were those that could be identified before induction or expectant management, and thus potentially inform patient counseling. In multivariable analysis, few factors remained associated with the adverse outcomes. Notably, for all of the outcomes evaluated, no characteristics or groups of characteristics could be used to identify a subgroup of women who were more likely to benefit from expectant management compared with labor induction.
Induction at 39 weeks of gestation in this analysis was associated with a reduction in the perinatal composite outcome compared with expectant management. This finding has been reported in several earlier observational studies.3–7 and reflects a key difference between the parent trial and this secondary analysis. In the trial, women were randomized as early as 38 0/7 weeks of gestation and women who delivered before 39 weeks of gestation in both groups would be likely to have similar outcomes. In the current study, only women who reached 39 weeks of gestation were analyzed, given that it would be analytically not possible to determine factors associated with adverse outcomes after 39 weeks of gestation for women who were no longer pregnant at that gestational age. Correspondingly, this analysis reveals that among women who are at 39 weeks of gestation—the only group of women for whom the decision to proceed with induction or expectant management is truly actionable—intended elective induction of labor is associated with a lowering of the adverse perinatal composite outcome.
The results of the multivariable analysis demonstrating an increased risk of the perinatal composite outcome with increasing BMI are consistent with other studies.8,9 Other studies also have described a decreased risk of third- or fourth-degree perineal laceration with increasing BMI and in Black women.10,11 The association with an increased risk of postpartum hemorrhage in Hispanic women has been previously noted as well.10 Nevertheless, our analysis demonstrated that characteristics associated with adverse outcomes were few, had relatively weak associations, and were poorly predictive; moreover, none of these associations varied significantly by treatment group.
The strengths of our analysis arise from the large and robust study from which it is derived (which included 41 community and academic geographically diverse centers), which had strict inclusion criteria, prospectively collected data, and reliable and detailed, clinically relevant outcomes. Limitations of our study are those inherent to any observational study, as well as the generalizability of the findings.
In summary, intended labor induction at 39 weeks of gestation was associated with a reduction in the perinatal composite outcome. Our analysis demonstrated that few maternal characteristics were associated with adverse outcomes after 39 weeks of gestation, and none could reliably predict which women were most likely to have these adverse outcomes. Furthermore, although induction of labor at 39 weeks of gestation was associated with a reduction in the perinatal composite outcome, no characteristics identified a subgroup of women who would preferentially benefit from undergoing expectant management. Such information may be useful to women and their obstetric providers as they consider their choices regarding timing of delivery.
Authors' Data Sharing Statement
- Will individual participant data be available (including data dictionaries)? Yes.
- What data in particular will be shared? Individual de-identified data on all of the randomized participants, including baseline characteristics, details of delivery, pregnancy and neonatal outcomes and adverse events. Data may be truncated or combined to ensure confidentiality and privacy.
- What other documents will be available? Other documents include the original data forms, the study protocol, a comprehensive data dictionary including variable names, formats, definitions and whether the variable has been recoded and how. Explanatory notes are also included.
- When will data be available (start and end dates)? Data will be available by 3/2021 through NICHD Data and Specimens Hub (DASH), assuming IRB approval.
- By what access criteria will data be shared (including with whom, for what types of analyses, and by what mechanism)? Terms and conditions for access are given athttps://dash.nichd.nih.gov/.
1. Grobman WA, Rice MM, Reddy UM, Tita AT, Silver RM, Mallett G, et al. Labor induction versus expectant management in low-risk nulliparous women. New Engl J Med 2018;379:513–23.
2. Hosmer DW, Lemeshow S, Sturdivant RX. Applied logistic regression, 3rd ed. Hoboken, NJ: John Wiley and Sons; 2013.
3. Cheng YW, Kaimal AJ, Snowden JM, Nicholson JM, Caughey AB. Induction of labor compared to expectant management in low-risk women and associated perinatal outcomes. Am J Obstet Gynecol 2012;207:502.e1–8.
4. Darney BG, Snowden JM, Cheng YW, Jacob L, Nicholson JM, Kaimal A, et al. Elective induction of labor at term compared with expectant management: maternal and neonatal outcomes. Obstet Gynecol 2013;122:761–9.
5. Gibson KS, Waters TP, Bailit JL. Maternal and neonatal outcomes in electively induced low-risk term pregnancies. Am J Obstet Gynecol 2014;211:249.e1–16.
6. Osmundson S, Ou-Yang RJ, Grobman WA. Elective induction compared with expectant management in nulliparous women with an unfavorable cervix. Obstet Gynecol 2011;117:583–7.
7. Stock SJ, Ferguson E, Duffy A, Ford I, Chalmers J, Norman JE. Outcomes of elective induction of labour compared with expectant management: population based study. BMJ 2012;344:e2838.
8. Aune D, Saugstad OD, Henriksen T, Tonstad S. Maternal body mass index and the risk of fetal death, stillbirth, and infant death: a systematic review and meta-analysis. JAMA 2014;311:1536–46.
9. Scott-Pillai R, Spence D, Cardwell C, Hunter A, Holmes V. The impact of body mass index on maternal and neonatal outcomes: a retrospective study in a UK obstetric population, 2004–2011. Int J Obstet Gynaecol 2013;120:932–9.
10. Grobman WA, Bailit JL, Rice MM, Wapner RJ, Reddy UM, Varner MW, et al. Racial and ethnic disparities in maternal morbidity and obstetric care. Obstet Gynecol 2015;125:1460–7.
11. Blomberg M. Maternal body mass index and risk of obstetric anal sphincter injury. Biomed Research International 2014;2014:395803.