Materials and Methods
The protocol was instituted on September 23, 1997 in the Maternal-Fetal Medicine service of the University of Alabama at Birmingham Hospital, which has approximately 3000 deliveries annually. Deliveries in this predominantly publicly funded service are done by resident physicians under direct 24-hour supervision of one of 11 faculty or fellows of the Maternal-Fetal Medicine Division. Before institution of the protocol, all members of the Maternal-Fetal Medicine Division agreed to manage eligible women according to it, and the resident staff received in-service education. Women were eligible if gestational age was at least 36 weeks and they were undergoing indicated induction (eg, ruptured membranes, hypertensive disease) with a vertex, singleton gestation and a cervix dilated no more than 2 cm. Women with dead fetuses, chorioamnionitis before initiation of oxytocin, and previous cesareans were not eligible. In general, women with intact membranes and cervical dilatation less than 2 cm received extraamniotic saline infusion (for up to 12 hours) for cervical ripening concomitant with initiation of oxytocin, the technique of which is described in a previous investigation from our institution.14 Other methods of cervical ripening (eg, prostaglandins [PG]) were not used. All women received dilute intravenous oxytocin that was initiated at 2 mU/minute and increased every 15 minutes over 2 hours to 30 mU/minute or until regular painful uterine contractions ensued or labor progressed. The protocol required amniotomy within 24 hours of starting induction. After membrane rupture, our general policy was to place an intrauterine pressure catheter and a fetal scalp electrode and titrate the oxytocin to achieve more than 200 Montevideo units.
If the fetal heart rate (FHR) pattern was reassuring, cesarean delivery before the active phase of labor (defined as cervical dilatation of at least 4 cm and 90% effacement or 5 cm of cervical dilatation, regardless of effacement) was not permitted for nonprogressive labor before oxytocin had been administered for at least 12 hours with ruptured membranes. Women who met those criteria were eligible for cesarean for failed labor induction. Once in the active phase, labors were managed according to our previously detailed active phase labor management protocol.15
University of Alabama at Birmingham institutional review board approval was obtained to assess outcomes after using this protocol. To ensure complete ascertainment of protocol-eligible patients, oxytocin orders were available only in sequentially numbered envelopes on the labor and delivery suite. On a daily basis a research nurse determined for which women oxytocin had been ordered and identified all eligible women by chart review. The charts of eligible women and their infants were abstracted for selected demographic and clinical variables. Specific maternal complications assessed included clinically diagnosed chorioamnionitis, postpartum hemorrhage, postpartum endometritis, need for blood transfusion, and abdominal incision seroma or infection. Intrapartum chorioamnionitis was diagnosed when the woman's temperature was at least 100F and she had at least one of the following symptoms or signs: uterine tenderness, maternal or fetal tachycardia (greater than 100 beats per minute or greater than 160 beats per minute, respectively), or purulent amniotic fluid (AF) or cervical discharge and the absence of any other identifiable source of infection (eg, pyelonephritis).16 The diagnosis of intrapartum chorioamnionitis precluded diagnosis of postpartum endometritis, which was defined as a temperature of at least 100.4F postpartum and at least one of the following symptoms or signs: uterine tenderness, maternal tachycardia (greater than 100 beats per minute), or purulent cervical discharge and absence of any other identifiable source of infection. Throughout the period of the study, our policy was to administer intrapartum antibiotics to women at risk of early-onset neonatal group B streptococcal infection, ie, previous infants with invasive group B streptococcal infections, group B streptococcal bacteriuria during the current pregnancy, delivery before 37 weeks, membrane rupture for at least 18 hours, and intrapartum temperature of at least 38C.17
Neonatal outcomes and complications assessed included 1 and 5 minute Apgar scores; umbilical cord arterial blood gas measurements; bacteremia and bacterial meningitis; shoulder dystocia; brachial plexus injury; need for antibiotics; supplemental oxygen requirement outside the delivery room; mechanical ventilation; and diagnosis of necrotizing enterocolitis, intraventricular hemorrhage, seizures, birth asphyxia,18 or death.
Nulliparas and paras were analyzed separately and compared for certain outcomes. Within each parity grouping, we compared complication rates between women who remained in the latent phase with those who progressed to the active phase or delivered at three epochs (after 6, 9, or 12 hours of oxytocin with ruptured membranes). Outcomes also were assessed and reported separately for the few women without informative cervical examinations (within ± 1.5 hours of the nominal 6-, 9-, or 12-hour examination). We calculated 95% confidence intervals (CI) for selected key outcomes. Proportional data were compared by χ2 or Fisher exact test, as appropriate. Normally distributed data were compared using the Student t test, and non-normal continuous data were compared with the Wilcoxon rank-sum test. Data were analyzed using the SAS system version 8.0 for personal computers (SAS Institute, Cary, NC). P < .05 was considered statistically significant. We established a target sample of at least 500 women. A cohort of that size yields high precision (narrow CIs) around most observed event rates. For example, with a sample of 500, the upper limit of the exact binomial 95% CIs for an event with an observed rate of 1% is less than 3% percent.
From September 23, 1997 through October 27, 1999, 509 women were managed by our protocol. Three hundred sixty women (71%) were nulliparous and 149 (29%) were parous. Parous women were significantly older, heavier, of slightly less advanced gestational age, initially somewhat more dilated, and less likely to receive epidural analgesia (Table 1). Twenty-five percent of nulliparas and 9% of paras were delivered by cesarean (P = .001).
In nulliparas, the median (range) interval from oxytocin initiation to rupture of membranes for the 247 (69%) women who began induction with intact membranes was 6.3 (0–26.6) hours, and the median induction-to-delivery interval was 14.2 (2.3–42.4) hours. For the 113 nulliparas (31%) who began induction with ruptured membranes, the median induction-to-delivery interval was 9.2 (1.5–32.8) hours.
After 6 hours of oxytocin with ruptured membranes, at least 14% (n = 51) of nulliparas were still in the latent phase (16 [4.4%] lacked an informative cervical examination for this time stage and might still have been in the latent phase). Thirty-nine percent (n = 20) of the 51 women still in the latent phase at 6 hours were delivered vaginally (95% CI 26%, 53%). Seven percent (n = 25) of nulliparas were still in the latent phase after 9 hours, and their vaginal delivery rate was 28% (n = 7, 95% CI 10%, 46%). Only 4% (n = 15) remained in the latent phase after 12 hours, and their vaginal delivery rate was 13% (n = 2, 95% CI 2%, 39%; Table 2). Nulliparas in the latent phase at each stage (after 6, 9, and 12 hours of oxytocin with ruptured membranes) ultimately had higher rates of infectious morbidity (ie, chorioamnionitis and endometritis) than women who had progressed to the active phase or had delivered at those stages, although these differences were statistically significant only for endometritis (Table 2).
In parous women, the median (range) interval from oxytocin initiation to rupture of membranes for 107 (72%) women who began induction with intact membranes was 5.2 (0.7–27.9) hours, and their median induction-to-delivery interval was 8.3 (1.9–33.4) hours. For the 42 women (28%) who began induction with ruptured membranes the median induction-to-delivery interval was 5.5 (2.2–16.1) hours. Those intervals were all significantly shorter than the respective intervals for nulliparas.
Compared with nulliparas, parous women were less likely to still be in the latent phase after 6 hours of oxytocin and ruptured membranes (14% versus 3%, respectively, P = .001). Of the five parous women still in the latent phase at 6 hours, two were still in the latent phase at 9 hours, and none was in the latent phase at 12 hours. All five delivered vaginally. Parous women had significantly lower rates of chorioamnionitis and endometritis than nulliparas, 6% versus 16%, and 6% versus 13%, P = .002 and P = .03, respectively. Too few parous women were in the latent phase at and beyond 6 hours of oxytocin and ruptured membranes to conduct meaningful analyses of the relationship between remaining in the latent phase and the risk of maternal infection.
In 26 nulliparas and five parous women (6% of the total cohort) who had not yet delivered, we were unable to document cervical examinations in at least one of the stages (19 at 6 hours, 9 at 9 hours, and 6 at 12 hours). In general, at each respective stage, the vaginal delivery and maternal infection rates of those women were intermediate between the rates of women who had progressed into the active phase or had been delivered and the rates of women who remained in the latent phase.
Twenty-nine (28%) of 104 cesareans were done in the latent phase of labor, 26 in nulliparas and 3 in parous women. Thirteen nulliparas (4%) but no parous women, had failed labor induction (ie, cesarean in the latent phase for lack of cervical dilation). The 13 nulliparas received a median (range) of 15 (8–24) hours of oxytocin after membrane rupture. Only three of 13 did not receive oxytocin for the protocol-mandated 12 hours after membrane rupture. However, they all received at least 8 hours, with a total duration of oxytocin (before and after membrane rupture) of 9–26 hours.
No woman suffered severe morbidity and, except cesarean delivery and infectious morbidity among nulliparas, no maternal outcome correlated with duration of the latent phase, including postpartum hemorrhage (overall rate 5%), blood transfusion (1%), wound seroma infection (n = 4), or length of stay (median 3 days for vaginal delivery and 5 days for cesarean).
We excluded 11 infants with serious congenital anomalies from the analysis of neonatal morbidity. Fetal and neonatal outcomes for the remaining 498 infants were generally good, with no stillbirths, neonatal deaths, or cases of bacterial meningitis, intraventricular hemorrhage, seizures, or birth asphyxia. Mean birthweight ± standard deviation (SD) was 3194 ± 525 g. Median 1- and 5-minute Apgar scores were 8 and 9, respectively.
Twelve infants (2.4%, 95% CI 1.1%, 3.7%) had 13 serious neonatal morbidities or some potential marker thereof, including 5-minute Apgar score of 3 or less (n = 1), umbilical artery (UA) pH under 7.0 (n = 3), bacteremia (n = 2), brachial plexus injury (n = 2), mechanical ventilation (n = 3), and necrotizing enterocolitis (n = 2; Table 3). Three of those 12 infants were born to parous women and nine to nulliparas, a rate of 2.1% among parous women and 2.5% among nulliparas (P = 1.0). The rate of serious neonatal morbidity (or potential markers of such morbidity) was unrelated to latent phase duration. Ten of the 12 infants were born to mothers who had progressed to active phase or were delivered after 6 hours of oxytocin and ruptured membranes. Only one of the 12 infants (with group B streptococcal bacteremia) was born to a woman (a nullipara) who was in the latent phase after 6 hours. She had a cesarean for failed induction after 22.5 hours of oxytocin with ruptured membranes and inadvertently did not receive group B streptococcal prophylaxis. The other infant was born to one of 31 women who lacked at least one informative cervical examination. That infant had a 5-minute Apgar score of 3 with UA pH of 7.04, received 3 days of intravenous antibiotics but had negative blood cultures, did well, and was discharged 4 days after birth (Table 3). Thus, the rate of severe neonatal morbidities was not significantly different in infants of nulliparas who remained in the latent phase after 6, 9, and 12 hours of oxytocin and ruptured membranes compared with those who had progressed into the active phase or delivered, 2.0% versus 2.4%, 4.0% versus 2.5%, and 6.7% versus 2.4%, respectively.
Rates of less severe neonatal morbidities, eg, antibiotic receipt (overall rate 11%), shoulder dystocia (3%), supplemental oxygen requirement (1%), and length of neonatal hospitalization (median 2 days for vaginal delivery and 4 days for cesarean) did not correlate with latent phase duration.
We have reported the results of an initial attempt to standardize the definition of failed labor induction based on two indices, duration of ruptured membranes after initiation of oxytocin and lack of progression into active phase of labor. We chose those because they are reproducible and pertinent to labor induction. Ruptured membranes imply commitment to delivery, and, in spontaneous labor, entry into active phase demarcates the transition from the often desultory progression in the latent phase to more rapid dilation (and traditionally less passively managed) of the active phase.
In terms of vaginal delivery, we found benefit to continued induction for women still in the latent phase at 6 and, to a lesser degree, 9 hours after oxytocin with ruptured membranes. The largest incremental benefit was from continued induction for the 51 nulliparas still in the latent phase after 6 hours of oxytocin with ruptured membranes; 20 (39%) of those women had vaginal deliveries. All five parous women in the latent phase at that stage also delivered vaginally. By that point, the total median duration of oxytocin (before and after membrane rupture) administered to the 30 nulliparas and paras still in the latent phase after 6 hours who began induction with intact membranes was 12.5 hours. After that duration of oxytocin, it is likely that in some practice settings, those women would have had cesarean deliveries. Our data suggest that continued induction is better because the only serious neonatal morbidity among those infants was a case of group B streptococcal bacteremia that occurred after membrane rupture for 22.5 hours and failure to administer antibiotic prophylaxis. Overall, the rate of serious neonatal morbidity among the 51 infants of nulliparas who were still in the latent phase at that point (2.0%) did not differ significantly from the rate in infants of nulliparas who progressed into active phase or delivered (2.4%).
Our study had several limitations. As usually defined, active phase is a phenomenon of spontaneous labor, signaled by an abrupt change in the slope of the curve that results when cervical dilatation is plotted against time.13 Typically, it occurs when the cervix reaches 3–4 cm of dilatation.13 Thus, our definition of active phase labor (and conversely, our definition of latent phase labor) was arbitrary, in that it was unrelated to the observed rate of cervical change, and none of the women to whom it was applied were in spontaneous labor. Nevertheless, although arbitrary, it forced explicit acknowledgement that latent phase labor typically progresses more slowly than active phase, thus, thresholds for intervention should be modified accordingly.
Another obvious limitation of our study is that it was observational and not randomized. Comparison of labor induction protocols (with different definitions of failed labor induction) against one another in a proper randomized clinical trial would be the optimal, least biased manner of evaluation. However, most randomized clinical trials compare standard with experimental therapies or alternate but accepted therapies. There is no standard definition of failed labor induction.10–12 Our study design allowed for noncomparative assessment of a definition of failed labor induction and for comparison of the observed incremental benefits (in terms of vaginal delivery) and the potentially attributable incremental risks (in terms of maternal and neonatal morbidity) of continued labor induction despite nonprogression to active phase for up to 12 hours of oxytocin and ruptured membranes. At one level our study should be viewed as the first step toward a randomized trial of labor induction protocols.
Because our study was not randomized, the only incontrovertible statement we can make about the benefits and risks of continued labor induction despite nonprogression to active phase is that continued labor induction allowed some women to have vaginal deliveries. The higher rate of infectious morbidity observed in the women who remained in the latent phase at these stages (relative to women who had progressed into the active phase or had delivered) might have been related to continued labor induction. However, the possibility that those women might have had infections regardless of how their labors were managed must be acknowledged. Four of 12 cases (33%) of chorioamnionitis in women who were in the latent phase at 6 hours had already been diagnosed by the 6-hour examination. Clearly, these women did not experience infection as a result of continued labor induction. Thus, the morbidity associated with continued labor induction almost certainly overestimates the morbidity attributable to continued labor induction, especially when the alternative to continued induction is cesarean delivery.
A further limitation of our study was that even with more than 500 eligible women, few (56, 11%) were still in the latent phase after 6 hours of oxytocin and ruptured membranes, and fewer still (27, 5%) after 9 hours. That was a circumstance that we had not anticipated. Although infants born to those women were no more likely to have serious morbidity than infants of the women who had progressed or delivered after these intervals (2.0% versus 2.4%, and 4.0% versus 2.5%, respectively, among infants of nulliparas), our statistical power to detect a difference in such morbidity was limited. For example, we calculated that to have 80% power to detect a doubling of the observed 2.4% rate of severe morbidity in the infants of nulliparas who progressed or delivered at 6 hours would require a sample of approximately 5000 women.
Likewise, our sample size limited the opportunity for informative analysis by preinduction membrane status or initial cervical dilatation. For instance, 5.3% of infants of the 95 women who began induction with a closed cervix and who had progressed beyond the latent phase (or had been delivered) after 6 hours of ruptured membranes and oxytocin had severe neonatal morbidity or some marker thereof. Although the infants of the 26 women who began induction with closed cervices but remained in the latent phase after 6 hours did not have a higher mordidity rate (in fact their rate was lower, but not significantly so, 3.9%, P > .999), to have 80% power to detect a doubling of the observed rate of 5.3% would require a sample of approximately 1250 women. The safety of this protocol clearly requires further evaluation.
Most women entered the protocol with intact membranes and had cervical ripening by extraamniotic saline infusion, which might limit the applicability of our results to women in whom other commonly used cervical ripening-induction agents, such as PGE1 or PGE2, are used. We focused on vaginal delivery as an outcome, so aspects of management unrelated to the induction protocol per se, eg, oxytocin dosing and management of the active phase and second stage of labor, might have influenced our results. The high rate of epidural analgesia use (92% by nulliparas and 74% by parous women) precluded informative assessment of this potential confounder. Finally, we chose a threshold of 12 hours because it was the greatest minimum duration for which a consensus could be reached among managing physicians. We realized that a valid and unbiased assessment of our protocol required near-complete adherence to whatever threshold was established. In this regard we succeeded, in that only three deliveries (0.6% of the total) were not managed according to protocol.
Our investigation represents an important initial attempt to standardize the definition of failed labor induction based on reproducible and pertinent clinical parameters. Given the increasing frequency with which labor induction is used and the lack of any recognized standards for failed induction, additional investigation along those lines is urgently needed.
1. Mathews TJ. Trends in stimulation and induction of labor 1989–1995. Stat Bull 1997; Oct-Dec.
2. Ventura SJ, Martin JA, Curtin SC, Mathews TJ. Births: Final data for 1997. National Center for Health Statistics. Nat Vital Stat Rep 1999;47:1–96.
3. 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.
4. 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.
5. Jarvelin MR, Hartikainen-Sorri AL, Rantakallio P. Labour induction policy in hospitals of different levels of specialisation. Br J Obstet Gynaecol 1993;100:310–5.
6. Smith LP, Nagourney BA, McLean FH, Usher RH. Hazards and benefits of elective induction of labor. Am J Obstet Gynecol 1984;148:579–85.
7. Williams MC, Krammer J, O'Brien WF. The value of the cervical score in predicting successful outcome of labor induction. Obstet Gynecol 1997;90:784–9.
8. Watson WJ, Stevens D, Welter S, Day D. Factors predicting successful labor induction. Obstet Gynecol 1996;88:990–2.
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. Cunningham FG, MacDonald PC, Gant NF, Leveno KJ, Gilstrap LC III, Hankins GDV, et al. Williams obstetrics. 20th ed. Stamford, Connecticut: Appleton & Lange, 1997.
11. Gabbe SG, Niebyl JR, Simpson JL, eds. Obstetrics. Normal & problem pregnancies. New York: Churchill Livingstone, 1996.
12. American College of Obstetricians and Gynecologists. Induction of labor. ACOG practice bulletin no. 10. Washington, DC: American College of Obstetricians and Gynecologists, 1999.
13. American College of Obstetricians and Gynecologists. Dystocia and the augmentation of labor. ACOG technical bulletin no. 218. Washington, DC: American College of Obstetricians and Gynecologists, 1995.
14. Guinn DA, Goepfert AR, Christine M, Owen J, Hauth JC. Extraamniotic saline infusion, laminaria or PGE2
get for induction of labor in women with an unfavorable cervix. Obstet Gynecol 2000;96:106–12.
15. Rouse DJ, Owen J, Hauth JC. Active-phase labor arrest: Oxytocin augmentation for at least 4 hours. Obstet Gynecol 1999;93:323–8.
16. Gibbs RS, Duff P. Progress in pathogenesis and management of clinical intraamniotic infection. Am J Obstet Gynecol 1991;164:1317–26.
17. American College of Obstetricians and Gynecologists. Use and abuse of the Apgar score. ACOG committee opinion no. 174. Washington, DC: American College of Obstetricians and Gynecologists, 1996.
18. American College of Obstetricians and Gynecologists. Prevention of early-onset group B streptococcal disease in newborns. ACOG committee opinion no. 173. Washington, DC: American College of Obstetricians and Gynecologists, 1996.