The cesarean delivery rate in the United States continues to rise, reaching 31.1% in 2006, a new all-time high.1 Arrest in the active phase of labor is a significant contributor to this increase,2 and has been shown to raise the risk of cesarean delivery fourfold to sixfold.3,4 In 1989 the American College of Obstetricians and Gynecologists (ACOG) recommended the diagnosis of active phase arrest only be made with no cervical change for a minimum of 2 hours in the setting of an adequate uterine contraction pattern, defined as contractions with 200 or more Montevideo units in a 10-minute period.5 However, in clinical practice, cesarean deliveries performed for lack of progress often fail to meet these criteria.6 More recent ACOG recommendations on the management of labor dystocia recommend oxytocin administration be considered in the setting of arrested labor for effective uterine activity while avoiding fetal compromise.7 However, there is little information on the effect of active phase arrest on maternal and neonatal outcomes.
Although ACOG recommends at least a “2-hour minimum” before a diagnosis of active phase arrest is considered, recent studies have demonstrated that if women diagnosed with active phase arrest are given at least 4 hours, a majority will go on to achieve vaginal delivery. In these two studies, Rouse et al8,9 report that this protocol seems safe and effective, with low rates of maternal and neonatal complications in all women with active phase arrest, with similar rates in those with cesarean and vaginal deliveries. We sought to confirm these findings and evaluate the perinatal outcomes in our cohort of women with active phase arrest.
To explore the obstetric management of women with active phase arrest as well as to estimate the risks associated with active phase arrest, we created two comparison groups. First, we included only women with active phase arrest and compared the rates of adverse perinatal outcomes among women who had vaginal deliveries to those who had cesarean deliveries. Next, to examine the risk associated with active phase arrest, we compared the outcomes of women diagnosed with active phase arrest who achieved vaginal delivery to women without active phase arrest who also delivered vaginally.
MATERIALS AND METHODS
We undertook a large retrospective cohort study to examine the rates of adverse perinatal outcomes in women with active phase arrest. The cohort included all women with a live, term, cephalic, singleton birth diagnosed with active phase arrest at the University of California, San Francisco, who delivered between 1991 and 2001 (n=1,014). We excluded women with multiple gestations, delivery before 37 weeks of gestation, and anomalous or nonviable fetuses. The Committee on Human Research at University of California, San Francisco approved the study. The primary independent variable of interest was mode of delivery (vaginal delivery compared with cesarean delivery). Maternal outcomes included the frequency of chorioamnionitis, endomyometritis, postpartum hemorrhage (more than 500 mL of blood loss for vaginal delivery and more than 1,000 mL of blood loss for cesarean delivery) severe postpartum hemorrhage (more than 1,000 mL of blood loss for vaginal delivery and more than 1,500 mL of blood loss for cesarean delivery), and blood transfusion. Neonatal outcomes included 5-minute Apgar score less than 7, acidemia (umbilical cord arterial pH less than 7.0 or umbilical artery base excess –12 or less), neonatal sepsis (as diagnosed by the managing pediatrician), and the frequency of admission to the neonatal intensive care unit (NICU).
To evaluate the risks associated with vaginal delivery in the setting of active phase arrest, we constructed an alternative cohort that included all women with term vaginal deliveries that occurred at the University of California, San Francisco from 1991 to 2001 (n=12,901). All women with a term, singleton, live, cephalic, nonanomalous fetus who delivered vaginally during the study period were included. As above, we excluded women with multiple gestations, preterm delivery before 37 weeks, cesarean delivery, and anomalous or nonviable fetuses. All deliveries during the study period were performed by the attending physicians, certified nurse midwives, or resident physicians with attending supervision.
For this cohort, the primary independent variable was the diagnosis of active phase arrest. We evaluated several outcomes. Maternal outcomes included frequency of operative vaginal delivery (including forceps and vacuum-assisted vaginal delivery), chorioamnionitis, severe (third or fourth degree) perineal lacerations, endomyometritis, postpartum hemorrhage, and blood transfusion. Neonatal outcomes examined included the frequency of 5-minute Apgar score of less than 7, acidemia, neonatal sepsis, admission to the NICU, shoulder dystocia, clavicular fracture, Erb's palsy, and cephalohematoma (as diagnosed by the pediatrician caring for the neonate).
During the study period, the diagnosis of active phase arrest was defined as absence of cervical change during the active phase of labor (4 cm or more cervical dilation) for at least 2 hours in the presence of adequate uterine contractions (200 or more Montevideo units per 10-minute period, as measured by an intrauterine pressure catheter). The diagnosis of active phase arrest was made by the managing physician at the time of delivery according to these criteria. Management decisions were under weekly morbidity and mortality peer review for quality assurance according to institutional standards of care. To evaluate management strategies at our institution, we have reviewed the charts of 191 women with a diagnosis of active phase arrest and report that 48% were expectantly managed beyond 2 hours of active phase arrest, 26% beyond 4 hours of active phase arrest, and 26% beyond 6 hours of active phase arrest.
We extracted all data from a large electronic database containing information regarding prenatal records, labor management, and perinatal outcomes that is prospectively collected, coded, and maintained. All clinical data were recorded at the time of admission and delivery by the managing physicians and midwives. Trained data abstractors also perform daily chart review to ensure accurate and complete information reporting.
All data were analyzed using Stata 9.0 (StataCorp, College Station, TX). Univariable analyses using t tests and χ2 statistics were performed to compare maternal demographic variables as well as vaginal delivery rates across continuous and dichotomous predictors, respectively. A statistical significance level of P<.05 was used. The frequencies of adverse maternal outcomes and neonatal outcomes were compared by mode of delivery using the Fisher exact and χ2 tests for dichotomous variables. Next, multivariable logistic regression models were constructed to control for potential confounders, including maternal age, parity, maternal race/ethnicity, maternal prepregnancy body mass index (BMI), prior cesarean delivery, and delivery year. Each model evaluated the risk of adverse outcome associated with cesarean delivery.
Demographic data for women with and without active phase arrest were evaluated using basic univariable comparisons, χ2 tests for dichotomous variables and Student t tests for continuous variables. The rates of adverse maternal and neonatal outcomes were compared using χ2 tests as well as by constructing relative risks and associated 95% confidence intervals (CIs) associated with the diagnosis of active phase arrest for each outcome of interest.
To control for the effects of several confounders, we constructed multivariable logistic regression models to estimate the effect of the diagnosis of active phase arrest on each of the outcomes of interest. In all cases, covariates in the model included maternal age, parity, race/ethnicity, BMI, Medicaid insurance status, history of prior cesarean delivery, induction of labor, epidural use, and delivery year. A single model was constructed for each outcome of interest, each evaluating the effect (adjusted odds ratio, and 95% CI) associated with a diagnosis of active phase arrest. We were not able to construct models in the instances where there were no cases of an outcome in the active phase arrest group. Model goodness of fit was examined using the Hosmer-Lemeshow test.
We identified 1,014 women with active phase arrest: 33% (335) went on to deliver vaginally, and 28% (95) of these women underwent an operative vaginal delivery. Parity, maternal age, and ethnicity did not differ between women who delivered vaginally and those who had cesarean delivery. Compared with women who had cesarean delivery, women who delivered vaginally had a lower BMI (mean BMI 23.4 compared with 25.3 kg/m,2 P<.001), and delivered slightly smaller infants (mean birth weight±standard deviation 3,533 g (±658) compared with 3,700 g (±493), P<.001; Table 1).
Among women with active phase arrest, the univariable analysis revealed an increased rate of all adverse maternal outcomes examined in women who had a cesarean delivery (Table 2). These included chorioamnionitis, endomyometritis, postpartum hemorrhage, severe postpartum hemorrhage, and maternal blood transfusion. However, there were no differences in the rates of adverse neonatal outcomes between women who delivered vaginally and those who had a cesarean delivery.
When the frequencies of adverse outcomes were compared using a multivariable logistic regression model to control for potential confounders, we found that cesarean delivery was associated with an increased risk of chorioamnionitis (adjusted odds ratio [aOR] 3.37, 95% confidence interval [CI] 2.21–5.15), endomyometritis (aOR 48.41, 95% CI 6.61–354), postpartum hemorrhage (aOR 5.18, 95% CI 3.42–7.85), and severe postpartum hemorrhage (aOR 14.97, 95% CI 1.77–126). Adverse neonatal outcomes, however, were not statistically significantly associated with cesarean delivery (Table 3).
This cohort consisted of 12,901 women with a term vaginal delivery, 355 (2.6%) women with active phase arrest, and 12,566 without labor dystocia. Demographic and obstetric characteristics were compared between these two groups (Table 4). There were more nulliparous women (74%) who delivered vaginally with a diagnosis of active phase arrest compared with women delivered vaginally without active phase arrest (53%, P<.001). Compared with women without active phase arrest, more women with active phase arrest were aged more than 35 years at time of delivery (21% compared with 17%, P=.05), and delivered at 41 or more weeks of gestation (28% compared with 20%, P<.001). Fewer women in the active phase arrest group had induction of labor (10% compared with 18%, P<.001), but there was a greater usage of epidural analgesia among women with active phase arrest (84% compared with 55%, P<.001).
The rates of several maternal and neonatal adverse outcomes associated with a diagnosis of active phase arrest were compared between the study groups (Table 5). Among women with active phase arrest, there was an increased rate of operative vaginal delivery (28% compared with 17%, P<.001) as well as increased rates of several adverse maternal outcomes, including, chorioamnionitis (18% compared with 8%, p<.001), third- or fourth-degree perineal lacerations (16% compared with 9%, P=.001), and postpartum hemorrhage (26% compared with 17%, P<.001), compared with other women having a vaginal delivery without active phase arrest. The rates of endomyometritis and blood transfusion were not statistically different between the two groups. Examination of neonatal outcomes revealed increased rates of shoulder dystocia (4% compared with 2%, P<.01) and 5 minute Apgar less than 7 (5% compared with 2%, P<.001) among women with active phase arrest compared with those without (Table 5). The rates of other adverse neonatal outcomes and active phase arrest were similar, including sepsis, NICU admission, clavicular fracture, Erb's palsy, and acidemia.
To evaluate the effect of active phase arrest on adverse maternal and neonatal outcomes while controlling for potential confounders, we constructed several multivariable logistic regression models (Table 6). The models were tested against the “true” model using the Hosmer-Lemeshow test for goodness-of-fit, which showed no difference (P>.05). When controlling for potential confounders, women with active phase arrest had significantly increased odds of chorioamnionitis (aOR 2.70, 95% CI 1.22–2.36) and shoulder dystocia (aOR 2.37, 95% CI 1.33–4.25). Unlike the univariable models, the odds of operative vaginal delivery did not differ significantly between the groups (aOR 1.00, 95% CI 0.73–1.36), nor did the odds of postpartum hemorrhage (aOR 1.35, 95% CI 0.99–1.83) or severe perineal lacerations (aOR 0.92, 95% CI 0.63–1.36). Finally, after controlling for confounders, infants born to mothers with active phase arrest did not have statistically significantly increased odds of 5-minute Apgar scores less than 7 (aOR 1.75, 95% CI 0.85–3.61).
To systematically evaluate the rates of adverse perinatal outcomes among women with active phase arrest, we made two comparisons. First, we looked only at women with active phase arrest and compared the outcomes by mode of delivery: vaginal delivery to cesarean delivery. In women with active phase arrest, cesarean delivery was associated with an increased risk of chorioamnionitis, endomyometritis, and postpartum hemorrhage. However, cesarean delivery was not associated with adverse neonatal outcomes in women with active phase arrest. These findings suggest that efforts to achieve vaginal delivery in the setting of active phase arrest may reduce the maternal risks associated with cesarean delivery without additional risk to the neonate.
To get a sense of the number needed to treat to avoid postpartum hemorrhage and blood transfusion, using the raw numbers found in our analysis, we estimate that only three women would need to achieve vaginal delivery to prevent one postpartum hemorrhage, and 33 women would need to achieve vaginal delivery to prevent one maternal blood transfusion. The number of women needed to be managed expectantly to achieve vaginal delivery and prevent these complications would vary depending on the a priori chance of achieving vaginal delivery. From our work and prior studies, this proportion seems to range from 33% to 60%. Considering these values, the number needed to manage expectantly to prevent postpartum hemorrhage would range from five to nine and to prevent a maternal blood transfusion would range from approximately 50 to 100. Thus, determining which women are likely to achieve vaginal delivery in this setting is an important next step in this area of research.
In addition, our study illustrates several important differences in the perinatal outcomes of women with and without active phase arrest who experienced vaginal deliveries. Although many of the outcomes we examined did not differ between the two groups, we did observe a positive association between the diagnosis of active phase arrest and an increased risk of chorioamnionitis and shoulder dystocia. Our data also suggest an association between postpartum hemorrhage and active phase arrest. However, when we examined the more serious sequelae associated with these three outcomes, including neonatal sepsis, Erb's palsy, and maternal blood transfusion, we did not find an increased risk in the setting of active phase arrest.
Although we did not have information regarding the timing of the diagnosis of chorioamnionitis relative to that of active phase arrest, it could be that chorioamnionitis may be an effect of active phase arrest. We think, however, it is more likely that chorioamnionitis leads to dysfunctional uterine contractions and is a precursor to active phase arrest. In turn, it may be that the increased chorioamnionitis may lead to greater rates of postpartum hemorrhage; thus, intervening with a cesarean delivery seems to be unlikely to be protective against either of these outcomes. However, we also found a twofold increase in shoulder dystocia in women who experienced active phase arrest as compared with other women with vaginal deliveries. This seems consistent with the biology and anatomy of labor, that a tighter cephalopelvic relationship may both prolong the length of labor and increase the risk for shoulder dystocia. In contrast, it is also possible that there is diagnostic bias such that clinicians are more likely to diagnose a shoulder dystocia in a woman with active phase arrest. Interestingly, we note that no Erb's palsies occurred in the women with active phase arrest, so it is unclear whether laboring in the setting of a diagnosis of active phase arrest increases the actual neonatal morbidity from shoulder dystocia.
As noted above, we did not observe an increased risk of severe sequelae associated with chorioamnionitis and postpartum hemorrhage, namely, neonatal sepsis and maternal blood transfusion. However, this observed absence of association may be attributed to the rare nature of these outcomes such that the possibility of type II error exists. For example, for neonatal sepsis, we only had 15% power to find a 50% difference between the two groups and for transfusion, 20% power to find a 50% difference.
There are several other important limitations to this study. First, as an observational study, our results may be prone to confounding bias. We attempted to control for this through the use of multivariable logistic regression analyses. In addition, our effect estimates may be confounded by indication. For example, it may be that some women more likely to experience adverse outcomes had cesarean delivery and thus were excluded from our larger study cohort. However, previous studies examining only women with active phase arrest indicated low and comparable rates of adverse maternal and neonatal outcomes in women delivering vaginally and by cesarean delivery.8,9
Despite these limitations, we provide a useful comparison of outcomes in women who achieved vaginal delivery in the setting of active phase arrest that can be used to inform obstetric management and counsel women. We believe that with the lower rate of complications seen in women with active phase arrest who deliver vaginally as compared with those who have cesarean deliveries, that it is reasonable to use oxytocin augmentation and tincture of time to attempt to achieve vaginal delivery in women diagnosed with active phase arrest. In this era of falling vaginal birth after cesarean rates, the old adage of “once a cesarean, always a cesarean” has become a reality again in the majority of clinical practice settings. Thus, attempting to prevent the first cesarean by expectantly managing a prolonged labor or even with a formal diagnosis of active phase arrest seems reasonable. At the very least, we would encourage a multicenter, prospective trial of the management of active phase arrest described by Rouse et al8,9 to attempt to reduce the cesarean delivery rate in women's first labors.