Last year I wrote that “all obstetric clinicians know that induction of labor causes more cesarean deliveries. If you spend time on the labor floor, you can see that women who are induced end up with more cesareans than those who experience spontaneous labor. However, we have been confused and wrong regarding this issue for decades. This is because at any gestational age, the options for a clinician and patient are not induction of labor vs spontaneous labor but induction of labor vs expectant management. Expectant management encompasses having the patient progress into the future where a wide variety of things could occur.” In studies that are designed to compare induction of labor with expectant management, many of them find either no difference in the risk of cesarean delivery or a decreased risk of cesarean delivery and improvement in other perinatal outcomes.

This issue of expectant management is methodologically tricky. With expectant management of pregnancy, women may go into spontaneous labor, but they may also develop a complication of pregnancy such as preeclampsia, fetal growth restriction, or oligohydramnios. Furthermore, they may progress to late term (41 weeks) or postterm (42 weeks) gestations at which time most clinicians will induce, regardless of antenatal testing.

These various outcomes can be easily studied in a prospective, randomized, controlled trial that compares women induced at a particular gestational age with those expectantly managed. However, in retrospective cohort studies, the comparisons must be made carefully. Whereas identifying women with induction of labor is generally not that challenging, clearly identifying those without a medical indication can be more methodologically challenging. The comparison group of expectant management can also be challenging and several controversies have arisen in the literature.

First, when comparing a 38 week induction of labor, does the expectant management group include women who experienced spontaneous labor at 38 weeks’ gestation? In the initial study using this methodology, spontaneous labor at 38 weeks was not included because there would be women induced on the same day as those going into spontaneous labor, not a possible comparison in reality. However, it was pointed out that if women were induced at 38 weeks and 1 day, an alternative for those expectantly managed would be spontaneous labor at 38 weeks and 2 days, etc.

Two studies that perform the analyses both ways have been conducted. In examining the impact on perinatal mortality, Stock et al found that there was little impact of whether or not they used the spontaneously laboring patients at the same week. However, when the investigators examined cesarean delivery, they found that when these patients were excluded from the expectant management group, there was a small reduction in the risk of cesarean delivery with induction of labor. Alternatively, when the spontaneously laboring women at the same week of gestation were included in the expectant management group, there was a small increase in the risk of cesarean with induction of labor.

In the study by Darney et al, the effect size of decrement in both perinatal mortality and cesarean delivery was large enough that, regardless of the methodology, the induction of labor group experienced lower rates of these complications.

The other methodological issue is what complications may arise in the expectant management group. Chronic conditions such as pregestational diabetes and chronic hypertension need to be excluded from both groups in such a study because these conditions are commonly indications for induction of labor and exist prior to the term pregnancy.

Even gestational diabetes, which usually predates a term pregnancy, should be excluded. However, women with preeclampsia should be excluded from the nonmedically indicated induction group but kept in the expectant management group. This is because one of the complications that can arise in women who are expectantly managed is that they may develop preeclampsia. Similarly, one should not exclude women beyond 41 or 42 weeks’ gestation from the expectant management group because this is a complication of expectant management.

In the current paper by Bailit et al in this month’s Journal, a secondary analysis from the APEX study conducted by the Eunice Kennedy Shriver National Institute of Child Health and Human Development–funded Maternal-Fetal Medicine Units Network, the authors compare nulliparous women induced without a medical indication with those expectantly managed. The authors found that some maternal morbidity such as perinatal infection and third- or fourth-degree perineal lacerations were reduced in the women undergoing induction of labor, most consistently at 39 weeks’ gestation. From a neonatal perspective, many of the outcomes were not adequately powered, even in this large study, but they did identify a reduction in neonatal intensive care unit admissions with induction at 39 weeks’ gestation and a reduction in respiratory complications with induction at 40 weeks’ gestation.

They also found an increase in the risk of cesarean delivery with nonmedically indicated induction of labor at 38 and 40 weeks’ gestation although no difference at 39 weeks’ gestation. They also demonstrated a longer admission to delivery time with induction of labor, although no difference in the overall length of stay.

This is the first multicenter study of this question to use primarily chart-abstracted data. The most concerning finding was of the increased risk of cesarean delivery at 38 and 40 weeks’ gestation. However, it appears that the authors kept the women who underwent spontaneous labor in their expectant management group. As noted in previous text, this remains controversial; thus, it would be interesting to know whether there was an increased risk of cesarean delivery had these women been excluded from the expectant management group.

The other methodological issue that arose in this study was that the women beyond 42 weeks’ gestation were excluded. As noted above, reaching such a gestational age is a potential complication of expectant management, and thus, such patients should be included in an expectant management group. It is well known that women delivered at or beyond 42 weeks’ gestation have a greater risk of cesarean delivery and perinatal morbidity; thus, their exclusion could have biased the study toward expectant management. Such a bias may have affected the direction of the cesarean delivery rates, however, because the perinatal morbidity generally favored induction of labor, and this bias would make those findings only more robust.

As with many well-done studies, this one adds to a confusing and controversial literature about management of the term pregnancy. There has been a great push over the past 5 years to avoid nonmedically indicated deliveries prior to 39 weeks’ gestation, but there is less clear guidance at 39 and 40 weeks’ gestation. In pregnancies with an indication for an induction of labor, it has been deemed that risks of induction of labor are outweighed by the benefits of avoiding expectant management.

For such indications as preeclampsia or fetal growth restriction, there are surprisingly few prospective, randomized studies of this question. Rather, it became the standard of care to deliver because the perceived risk to either the mother or baby was believed to be too high to risk further expectant management. In pregnancies without a medical indication, this occurs at some point as well. Currently, there is good evidence that at 41 weeks’ gestation, there is a reduction in the risk of cesarean delivery from induction of labor and a reduction in the risk of perinatal mortality. Therefore, at 41 weeks’ gestation, clearly the research supports delivery at that gestational age as opposed to expectant management.

Between 39 0/7 weeks and 40 6/7 weeks, we simply do not have enough evidence to support routine delivery. Furthermore, although there may be some modest benefit, it is unclear whether the marginal cost of the induction of labor is worth that benefit. In the current study, the time to delivery was 3 hours longer, on average. The labor and delivery unit is an expensive place to spend 3 hours, so we must be careful not to commit to the usage of our scarce medical resources without a demonstration of adequate benefit. Specifically, although it has been demonstrated that induction of labor was cost effective at 41 weeks’ gestation, it was not found to be cost effective at 39 weeks’ gestation.

It is my understanding that the Maternal-Fetal Medicine Units Network has started a prospective trial of induction at 39 or 40 weeks’ gestation vs expectant management. Such a study is likely to provide us with higher-quality information than what we currently have. However, until such a study is completed, we still have to face this question clinically every day. In that realm, the current study adds to the increasing literature regarding this question to suggest that there are likely tradeoffs when considering a nonmedically indicated induction at 39 or 40 weeks’ gestation.

For patients requesting this option, it is important to bring up the potential impacts including the possibility of an increased risk of cesarean delivery and the comorbidities that may accompany this outcome in current and future pregnancies. However, it is also reasonable to discuss the potential improvement in outcomes as well suggested by this and other studies to empower women to participate in shared medical decision making.