The rate of labor induction has more than doubled in the United States over the past decade.1 It is suggested that the growth in the induction rate is likely due to an increase in elective inductions, where labor is initiated for convenience purposes and without a medical or obstetric indication.2 Observational studies show that the elective induction of labor is associated with an increase in the risk of cesarean delivery in nulliparous women,3–7 especially in women with an unfavorable cervix.6,8,9 However, it remains unclear how labor progresses when it is induced. Does induced labor progress differently from spontaneous labor, which in turn, increases the risk of cesarean delivery? Do the physicians intervene too soon? To answer these questions, we examined labor progression in electively induced labor with or without preinduction cervical ripening and risk of cesarean delivery among low-risk nulliparous women with a term pregnancy.
This study is a retrospective analysis of obstetric data from the Christiana Hospital in Newark, Delaware, a tertiary-care teaching hospital that serves residents of the state of Delaware and in neighboring regions of Pennsylvania, Maryland, and New Jersey. Christiana Hospital delivers approximately 7,000 newborns each year. In 1998, this institution established an electronic obstetric records system. This database consists of prospectively collected labor and delivery information and manually abstracted information from the prenatal record. When a woman is admitted to labor and delivery, the nursing staff creates an entry for her in the database. Information such as obstetric interventions performed as well as the details of each vaginal examination, including the time of the examination and cervical dilation, are noted in real time. Full-time trained abstractors extract and input the information from the prenatal and neonatal record within a month after the delivery. External validation of this database is performed every 3–4 months, and the information has been shown to be 95% accurate.
Between January 2002 and March 2004, 9,447 women delivered a singleton infant at this institution, of which 4,391 were nulliparous and 3,405 were nulliparous and delivered between 37 + 0 and 40 + 6 weeks. Further excluded from this analysis were 1,205 women with 1 or more of the following maternal or fetal conditions during pregnancy or at the time of admission: diabetes, hypertension, maternal medical history of cardiovascular, infectious, pulmonary, renal, mental, or thyroid disorders, multiple gestation, intrauterine growth restriction, uterine bleeding, oligohydramnios, breech presentation, elective cesarean delivery without a trial of labor, or clinically indicated induction. Thus, our final sample consisted of 2,200 low-risk nulliparous women with a singleton term pregnancy. The study protocol was reviewed and approved by the institutional review board at Christiana Care Health System and has been exempted by the National Institutes of Health.
Our exposure of interest was the elective induction of labor, which was defined as an induction without a medical or obstetric indication, or induction for “postdates” before 41 weeks of gestation. “Elective” and “postdates” are two of the options under “reason for induction” in the obstetric database that physicians were free to select. In the current study, 56% of women who had an elective induction were identified because their physicians had chosen “elective” as the indication. The rest of subjects (44%) were “postdates” before 41 weeks. Our induction population was further subdivided into 2 subgroups: those who had preinduction cervical ripening followed by oxytocin induction (n = 143) and those who had oxytocin induction without preinduction cervical ripening (n = 286). Although information on the patient's Bishop score at the time of admission was not recorded in the obstetric database, preinduction cervical ripening was generally applied if the Bishop score was less than 6. Our comparison group was composed of 1,771 nulliparous women who had spontaneous onset of labor. Labor status was determined at the time of admission by an attending or resident physician.
The only method of preinduction cervical ripening used at this institution during the study period was by an intracervical Foley catheter. No other mechanical or pharmacologic agents were used during the study period. Women in the Foley group had a 16-F Foley catheter placed supracervically, the balloon filled with 30 ml of sterile water, and the catheter placed on traction and gently taped to the maternal thigh. After extrusion of the Foley catheter, oxytocin induction began at 1.0 mIU/min and increased by 1 to 2 mIU/min every 30 minutes until adequate labor was obtained as dictated in a standard protocol. For women with a favorable cervix, oxytocin was consistently used per protocol and the only agent for induction.10
The primary outcomes of interest were labor progression and the risk of cesarean delivery. Median duration of labor by each centimeter of cervical dilation was computed based on data from serial vaginal exams and used as a measurement of labor progression. In this analysis, the starting point of labor was at vaginal examination at admission for those in spontaneous labor or start of oxytocin infusion for those with an induced labor. To adjust for potential confounders, we also included information on mother's age, race, patient payment status (resident compared with private), timing of epidural analgesia placement, infant birth weight (representing fetal size), and gestational age at delivery. To examine how long physicians had waited for cervical change to occur before proceeding to a cesarean delivery and whether it varied by labor onset status, we computed the median duration of time that cervical dilation was arrested before cesarean delivery for dystocia. For example, if a woman had cervical dilation of 5 cm at 10:00 (first time seen at 5 cm) and the last vaginal examination was done at 13:00 and the dilation was still 5 cm, the duration that the physician had waited before deciding on a cesarean delivery was 3 hours.
We first compared various baseline characteristics of the subjects, stratified by labor status at admission. For continuous variables with a normal distribution, mean and standard deviation values were calculated and Student t test or analysis of variance were used. For continuous variables with a skewed distribution, median and the values at the 10th and 90th percentiles were calculated and the Wilcoxon rank sum test was performed. For categorical data, percentages were calculated and the χ2 test was used. Next, unadjusted and adjusted risk ratios and 95% confidence intervals were computed from a log-binomial model for the overall risk of cesarean delivery, as well as the risk for a first-stage cesarean delivery, and first-stage cesarean delivery for dystocia. The log-binomial model was fitted using PROC GENMOD of SAS (SAS Institute, Inc., Cary, NC), where the convergence was assisted by forcing the estimated probabilities to be 1 or less during iteration. The overall risk of primary cesarean delivery in this cohort was 18%. Risk ratios were chosen because odds ratios overestimate the risk ratio when the outcome in a cohort study is common.11 Potential confounders were included in the full model if they were risk factors for cesarean delivery and associated with labor onset status, based on a P of .20. These covariates included maternal age, gestational age at delivery, and infant birth weight. However, only those covariates that changed the β coefficient of the exposure variable by greater than 10% (maternal age and birth weight) were retained in the final model for all analyses.
Lastly, we used survival analysis to quantify the duration of the first stage of labor progression and, specifically, the median time elapsed for women to proceed from 1 cm of cervical dilation to the next for each group. Because continuous monitoring of cervical dilation was not performed and women were admitted into labor at different degrees of cervical dilation, it is impossible to know exactly when an individual first reached a given level of dilation. However by considering these times as censored, the median duration of labor can be estimated.12
For each interval of cervical dilation (eg, from 3 cm to 4 cm), a lower and upper possible time range was computed. Thus, each individual contributed an interval-censored value at a given level of dilation. Interval-censoring can be defined as “when the time of event occurrence is known to be somewhere between times a and b, but we don't know exactly when.”13 For example, if we knew that an individual labor took at least 2 hours to progress from 4 to 5 cm but no longer than 5 hours, the interval (2–5) would be used to model the underlying distribution of the duration of labor from 4 to 5 cm. It is well established that the duration of labor has a skewed distribution leaning toward the left (ie, some labors produce a long right tail of the distribution). This distribution generally fits a log normal distribution. Thus, a natural assumption for the data comprising the time interval is that they are log normally distributed, which was consistent with our data (Shapiro-Wilk test for normality P < .001 before log transformation and P = .24 after the transformation).
We fitted a model with a log normal distribution and interval-censored time-to-event data to assess the median duration of time elapsed in hours for each centimeter of cervical dilation during labor using the LIFEREG procedure in SAS. Adjustment was made for baseline characteristics that were associated with labor status and the duration of labor, based on a P of .20. These covariates included maternal age, race, patient payment status, the timing and use of epidural analgesia, oxytocin use, gestational age at delivery, and fetal size. Median traverse times in each group were then estimated numerically by finding the time for which the average fitted probability of the event equaled 0.5. Because more than one comparison was made, we used the method of Holm to correct P values for pair-wise comparisons.14
Table 1 presents the baseline sociodemographic and intrapartum characteristics of the overall study population, stratified by labor onset status. Women with an elective induction were slightly older and more often white, compared with those admitted in spontaneous labor. In regard to obstetric interventions, 57% of women with a spontaneous onset of labor received oxytocin for labor augmentation, and epidural analgesia was commonly administered at this institution. Women with an elective induction were more likely to deliver a heavier infant and at a later gestation compared with women admitted with a spontaneous onset of labor. Overall, there were 8 infants that weighed 4,500 g or more, and all were born to women admitted with a spontaneous onset of labor.
Women whose labor was induced after cervical ripening were significantly more likely to deliver by cesarean compared with those in spontaneous labor (41.3% compared with 13.9%, P < .001; adjusted risk ratio 2.41; 95% confidence interval 1.95–2.98) (Table 2). In contrast, labors induced without the need of cervical ripening were not associated with an increased risk of a cesarean delivery (adjusted risk ratio 1.04; 95% confidence interval 0.79–1.37). The majority of cesarean deliveries were performed due to dystocia during the first stage of labor, and the pattern of increased risk of cesarean delivery remained the same. The 2- to 3-fold increase in the cesarean delivery rate was consistent across the gestational weeks (Fig. 1).
After adjusting for potential confounders, women induced without cervical ripening experienced a significantly shorter median duration of labor from 4 to 10 cm, compared with women admitted in spontaneous labor (266 compared with 358 minutes, P < .01) (Table 3). However, no significant difference in the second stage of labor was observed between these 2 groups. In contrast, women who were induced after cervical ripening had a slower duration of labor from 4 to 10 cm compared with those who were admitted with a spontaneous onset of labor (439 compared with 358 minutes, P = .02). Upon further examination, this prolonged labor seemed to be mostly attributable to the slower labor progression from 3 to 6 cm. No significant differences were seen between the 2 groups in either the late active phase or the second stage of labor. This may in part be attributable to progressive dropout of women with a cesarean delivery for dystocia, which is more prominent in induced labor than in spontaneous labor (Fig. 2). The selective dropout of protracted labor may, therefore, have masked the difference between the 2 groups in late active phase and second stage.
Last, we calculated the median duration of time that physicians had waited before the decision was made to perform a cesarean delivery for dystocia (n = 111) during the first stage of labor (Table 4). Both women with an elective induction with and without cervical ripening experienced longer wait times (5.5 and 3.2 hours, respectively) compared with women admitted in spontaneous labor (2.6 hours) when the cesarean was performed before 6 cm. However, when the cesarean was performed on or after 6 cm but before 10 cm, women with an elective induction without cervical ripening had a shorter median wait time compared with women admitted in spontaneous labor (1.5 compared with 2.7 hours, respectively, P = .02). No noticeable difference for women with an elective induction after cervical ripening and those admitted in spontaneous labor was noted.
Our study revealed that women who had an elective induction with cervical ripening, indicating an unfavorable cervix, had a significantly longer latent and early active phase and a 2- to 3-fold increased risk of cesarean delivery compared with those with a spontaneous onset of labor. We also found that physicians in this hospital had waited a substantially long time before performing a cesarean delivery in women with an unfavorable cervix. Despite cervical ripening, oxytocin, and a long wait for cervical change, a very high percentage of women with an unfavorable cervix will develop dystocia even at 40 weeks, and the cesarean delivery rate may reach 40% or higher. On the other hand, elective induction without cervical ripening, indicating a favorable cervix, has a faster active phase, probably due to continuous oxytocin.
Our findings on the increased risk of cesarean delivery among women who received preinduction cervical ripening are consistent with other studies which showed a 2- to 3-fold increased risk of cesarean delivery in women with an unfavorable cervix.6,8,9 Moreover, our labor progression findings are consistent with those from 2 previous studies.15,16 In a retrospective analysis of obstetric data from the Royal Victoria Hospital in Canada, Smith and colleagues15 reported a shorter mean duration of labor in induced nulliparas without cervical ripening (n = 175) compared with those in spontaneous labor (n = 1,487; 8.6 compared with 10.9 hours). In a smaller prospective study, Vierhout and colleagues16 report a faster mean duration of the first stage of labor in induced nulliparas with a favorable cervix (n = 58), compared with those in spontaneous labor (n = 40; 5.8 compared with 8.0 hours). Results of statistical significance tests between the 2 groups were not documented in either study.
However, the differences between previous studies and the current one are important to consider. First, women with an unfavorable cervix at induction seem to have different baseline characteristics as well as pattern of labor progression compared with those with a favorable cervix. Thus, adjustment for potential confounders is very important, which was inadequately done in some previous studies. Second, our study shows that labor progression and risk of cesarean delivery are very different in induced labor with and without a favorable cervix. Thus, it is inappropriate to combine these 2 groups of women together, nor is it informative to simply exclude women who received cervical ripening agents.17
Our study used advanced statistical methods to simultaneously examine labor progression centimeter by centimeter while adjusting for potential confounding factors. Only oxytocin and Foley catheter were used for induction in this institution, which gave us a unique opportunity to study induction issues without confounding by various induction methods. Previous studies have compared the efficacy of prostaglandin agents with the Foley catheter and found that they were similarly effective for the purposes of cervical ripening.10,18,19 No differences in cervical ripening time or in total time from the start of cervical ripening to delivery were noted between the two methods.
Nonetheless, the findings of this study must be interpreted with recognition of its limitations. First, the measurement of cervical dilation was subjective. Continuous monitoring of cervical dilation was not performed, and the measurements were based on vaginal examinations probably performed by several physicians. This potential misclassification would likely be nondifferential in nature and would bias our results toward the null. Second, women who experienced labor protraction and arrest in early stage of labor were progressively dropped out of the patient pool because of cesarean delivery. Although all groups are affected to some degree, the induced group that had an unfavorable cervix has more and earlier cesarean deliveries than the other 2 groups (Fig. 2). Thus, the dropout (or censoring) had a larger effect on that group than on the other groups despite the fact that we adjusted for several factors to reduce the potential for selection bias. This may explain why labor progression in late active phase and second stage did not differ between these groups.
Finally, we acknowledge that our observational study cannot directly address a clinically important question: Does elective induction in women with an unripe cervix at 40 weeks of gestation lead to a higher cesarean rate than expectant management does? A definitive answer has to come from a randomized trial. However, our retrospective data may provide some clue. The cesarean rates at 41 weeks or later were 16% for spontaneous labor, 20% for induction without ripening, and 41% for induction with ripening at this institution. They were virtually the same as those at 40 weeks (Fig. 1). Assume all women who delivered at 41 (n = 305) or 42 weeks (n = 22) had an unripe cervix at 40 weeks. If all of them had been induced at 40 weeks, 147 (45%) would have had cesarean delivery. Because they were delivered at 41 weeks or later, only 68 were delivered by cesarean (a compound rate of 21%). These numbers suggest that elective induction with an unripe cervix at 40 weeks is likely to result in a significantly higher cesarean rate than expectant management would.
In conclusion, our data indicate that the pattern of labor progression differs substantially for women in electively induced labor compared with those in spontaneous labor. Elective induction with an unfavorable cervix carries a high risk of labor arrest and subsequent cesarean delivery even in late gestation. A cesarean delivery rate above 40% among term, nulliparous women and the potential for repeat cesarean deliveries in future pregnancies seems to be too high a price to pay for a marginal cause.
1. Zhang J, Yancey MK, Henderson CE. U.S. national trends in labor induction, 1989–1998. J Reprod Med 2002;47:120–4.
2. Rayburn WF, Zhang J. Rising rates of labor induction: present concerns and future strategies. Obstet Gynecol 2002;100:164–7.
3. Maslow AS, Sweeny AL. Elective induction of labor as a risk factor for cesarean delivery among low-risk women at term. Obstet Gynecol 2000;95:917–22.
4. Cammu H, Martens G, Ruyssinck G, Amy JJ. Outcome after elective labor induction in nulliparous women: a matched cohort study. Am J Obstet Gynecol 2002;186:240–4.
5. Seyb ST, Berka RJ, Socol ML, Dooley SL. Risk of cesarean delivery with elective induction of labor at term in nulliparous women. Obstet Gynecol 1999;94:600–7.
6. Yeast JD, Jones A, Poskin M. Induction of labor and the relationship to cesarean delivery: a review of 7001 consecutive inductions. Am J Obstet Gynecol 1999;180:628–33.
7. Dublin S, Lydon-Rochelle M, Kaplan RC, Watts DH, Critchlow CW. Maternal and neonatal outcomes after induction of labor without an identified indication. Am J Obstet Gynecol 2000;183:986–94.
8. Prysak M, Castronova FC. Elective induction versus spontaneous labor: a case-control analysis of safety and efficacy. Obstet Gynecol 1998;92:47–52.
9. Macer JA, Macer CL, Chan LS. Elective induction versus spontaneous labor: a retrospective study of complications and outcome. Am J Obstet Gynecol 1992;166:1690–7.
10. Sciscione AC, Nguyen L, Manley J, Pollock M, Maas B, Colmorgen G. A randomized comparison of transcervical Foley catheter to intravaginal misoprostol for preinduction cervical ripening. Obstet Gynecol 2001;97:603–7.
11. Robbins AS, Chao SY, Fonseca VP. What's the relative risk? A method to directly estimate risk ratios in cohort studies of common outcomes. Ann Epidemiol 2002;12:452–4.
12. Klein JP, Moeschberger ML. Survival analysis: techniques for censored and truncated data. Berlin (Germany): Springer; 1997.
13. Allison PD. Survival analysis using the SAS system: a practical guide. Cary (NC): SAS Institute Inc.; 1995.
14. Holm S. A simple sequentially rejective multiple test procedure. Scand J Stat 1979;6:65–70.
15. Smith LP, Nagourney BA, McLean FH, Usher RH. Hazards and benefits of elective induction of labor. Am J Obstet Gynecol 1984;148:579–85.
16. Vierhout ME, Out JJ, Wallenburg HC. Elective induction of labor: a prospective clinical study, I: Obstetric and neonatal effects. J Perinat Med 1985;13:155–62.
17. Wigton TR, Wolk BM. Elective and routine induction of labor: a retrospective analysis of 274 cases. J Reprod Med 1994;39:21–6.
18. Barrilleaux PS, Boffill JA, Terrone DA, Magann EF, May WL, Morrison JC. Cervical ripening and induction of labor with misoprostol, dinoprostone gel, and a Foley catheter: a randomized trial of three techniques. Am J Obstet Gynecol 2002;186:1124–9.
19. Chung JH, Huang WH, Rumney PJ, Garite TJ, Nageotte MP. A prospective randomized controlled trial that compared misoprostol, Foley catheter, and combination misoprostol-Foley catheter for labor induction. Am J Obstet Gynecol 2003;189:1031–5.