|Home | About | Journals | Submit | Contact Us | Français|
To study neonatal outcomes in early preterm births by delivery route
Delivery precursors were analyzed in 4,352 singleton deliveries, 24 0/7–31 6/7 weeks’ gestation. In a subset (N=2,906) eligible for a trial of labor, neonatal mortality in attempted vaginal delivery (VD) was compared to planned cesarean delivery (CD) stratified by presentation.
Delivery precursors were classified as maternal or fetal conditions (45.7%), PPROM (37.7%) and preterm labor (16.6%). For vertex presentation, 79% attempted VD and 84% were successful. There was no difference in neonatal mortality. For breech presentation, at 24 0/7 – 27 6/7 weeks’ gestation, 31.7% attempted VD and 27.6% were successful; neonatal mortality was increased (25.2% versus 13.2%, p=0.003). At 28 0/7 – 31 6/7 weeks’ gestation, 30.5% attempted VD and 17.2% were successful; neonatal mortality was increased (6.0% vs. 1.5%, P= 0.016).
Attempted VD for vertex presentation has a high success rate with no difference in neonatal mortality unlike breech presentation.
Fifty four percent of all infant deaths in the United States occur among the 2 percent of infants born at less than 32 weeks’ gestation (2007).1 The optimal route of delivery for the early preterm fetus remains controversial. Some observational studies have shown a lower neonatal mortality for planned cesarean delivery as compared with vaginal delivery for vertex 2–4 and breech early preterm pregnancies 2, 4–9 whereas other studies do not show a difference by route of delivery for vertex10–12 or breech presentation.13–15
The vertical uterine incision often required for cesarean delivery at this gestational age increases the risks of hemorrhage, bladder injury and other complications. There is also an increased risk of uterine rupture, placenta previa and placenta accreta in subsequent pregnancies.16–17
Six trials have attempted to randomize the route of delivery for women in preterm labor at high risk for delivery. Recruitment difficulties limited combined enrollment in all of these trials to only 122 women. A metaanalysis of these trials found no statistically significant differences in neonatal outcomes by route of delivery, except for lower cord pH values among infants delivered by cesarean delivery.18
Because randomized trials to answer this question have not proven feasible, a study using recent cohort data to determine the effect of fetal presentation, gestational age, and the intended route of delivery on outcome would be valuable. Therefore, the purpose of this study was to use a contemporary cohort that reflects current obstetric and neonatal clinical practice to identify the precursors of early preterm delivery ≤ 32 weeks of gestation and to assess the effect of intended route of delivery on neonatal mortality for viable singleton early preterm births, stratified by presentation.
The Consortium on Safe Labor (CSL) was a study conducted by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health and has been described in detail elsewhere.19 Briefly, CSL was a retrospective cohort study involving 228,668 deliveries between 2002 and 2008 from 12 clinical centers and 19 hospitals representing nine American Congress of Obstetricians and Gynecologists districts. All deliveries at 23 weeks’ gestation or greater were included in the CSL cohort. Women could have more than one pregnancy in the cohort; so to avoid intra-person correlation only the first pregnancy enrolled was included for a total of 208,695 women.
Demographic data, medical history, prenatal, labor and delivery information as well as postpartum and neonatal outcomes were extracted from electronic medical records from each institution. Data from the neonatal intensive care unit were collected and linked to the newborn record. Maternal and newborn discharge ICD-9 codes were also collected for each delivery. Data were transferred in electronic format from each site and were mapped to common categories for each pre-defined variable at the data coordinating center. Data inquiries, cleaning and logic checking were performed. Validation studies for 4 key outcome diagnoses (cesarean for nonreassuring fetal heart rate tracing, asphyxia, NICU admission for respiratory conditions, and shoulder dystocia) confirmed high level of accuracy. There was greater than 95% concordance with the medical chart for 16/20 variables examined with the lowest concordance of 91.1% for clinical diagnosis of shoulder dystocia.19 Institutional Review Board approval was obtained by all participating institutions.
All singleton deliveries occuring between 24 0/7–31 6/7 weeks of gestation comprised the cohort for this analysis (N=5,055). Figure 1 summarizes the study population for analysis. Due to a high percentage of missing data for some neonatal and pregnancy variables, two sites were eliminated from further analyses (N=703), resulting in 4,352 pregnancies. We first categorized the possible precursors for preterm delivery. The admission indications, indications for delivery and pregnancy complications in the electronic medical record fields were used to classify the precursors to delivery into 3 overall categories: preterm labor (PTL), preterm premature rupture of membranes (PPROM), and indicated delivery. These 3 categories were mutually exclusive, prioritizing preterm labor then PPROM and then indicated for classification. The conditions leading to an indicated preterm birth included preeclampsia, placental abruption, non-reassuring fetal status, fetal growth restriction, severe maternal medical disease, antepartum stillbirth and major anomalies; these conditions were not mutually exclusive so multiple indications could be coded for a single pregnancy. For example, a pregnancy complicated by a major anomaly, fetal growth restriction, and non-reassuring fetal status would be counted in each of these 3 categories of indicated early PTB. Major anomalies were defined by one or more ICD-9 codes (Table 1).
Next we investigated neonatal outcomes in pregnancies eligible for attempted vaginal delivery (VD) versus planned cesarean delivery (CD). Pregnancies with conditions requiring immediate cesarean delivery such as fetal distress, placenta previa and placental abruption or associated with such poor neonatal outcome that route of delivery was unlikely to affect outcome including antepartum stillbirth and fetal anomalies were eliminated from further analyses (N=1,466), resulting in 2,906 singleton pregnancies. These 2,906 singleton non-anomalous pregnancies were then stratified by gestational age blocks: 24 0/7 – 27 6/7 weeks of gestation (N=1,102) and 28 0/7 – 31 6/7 weeks of gestation (N=1,804). These gestational age categories were based on the fact that the highest rates of neonatal mortality and morbidity occur between 24 0/7 – 27 6/7 weeks of gestation. Within each gestational age block, pregnancies were then stratified by vertex or breech presentation. (Figure 1)
The following maternal characteristics were analyzed: race (white/Asian, black, Hispanic/other); maternal age (continuous variable); marital status (married, not married/unknown); insurance (private, public/self pay/other/unknown); parity (nulliparas, multiparas); preexisting diabetes; preeclampsia; PTL; PPROM; smoking prior to/during pregnancy; alcohol use prior to/during pregnancy; illicit drug use prior to/during pregnancy; pre-pregnancy body mass index (continuous variable); gestational age week (continuous variable); and birthweight (continuous variable). Antenatal corticosteroid use (yes/no) was analyzed in a subset of pregnancies where this information was reported (N=1,094).
Attempted vaginal delivery was then compared to planned cesarean delivery for the following neonatal outcome variables: death (intrapartum death + neonatal death), asphyxia, respiratory distress syndrome (RDS), pneumonia, intraventricular hemorrhage (IVH), necrotizing enterocolitis (NEC), sepsis, and need for ventilation. The local NICU definitions as recorded in the neonatal medical chart were used. Diagnoses were also supplemented with ICD9 codes.
Univariable analysis for all baseline and outcome variables was performed using chi-square test comparing women undergoing attempted VD and planned CD. Multivariable analysis was then performed calculating adjusted relative risks and 95% confidence intervals to assess the strength of the relationship between attempted route of delivery stratified by presentation and the occurrence of neonatal mortality or morbidity. Missing values for maternal age (0.2% of cohort) and pre-pregnancy body mass index (27.7% of cohort) were replaced with mean values. For birth weight, 4% of births were missing values and were replaced by the mean value within each gestational age week. In order to estimate risk ratios instead of odds ratios and also avoid the convergence pitfall associated with log-binomial models, we used Poisson regression with a robust variance estimator as described by Zou20 and dealt with clustering by site using fixed effects covariates in the model. Statistical analyses were performed using SAS (SAS Institute Inc., Cary, NC) version 9.2.
In order to address the precursors of early preterm birth, the entire cohort of 4,352 singleton pregnancies at 24 0/7–31 6/7 weeks’ gestation with complete information on baseline and outcome variables was analyzed. The precursors leading to delivery by gestational age groupings (24–27 weeks’ gestation and 28–31 weeks’ gestation) are presented in Table 2. The distribution of precursors was as follows: maternal or fetal indications (45.7%), PPROM (37.7%) and preterm labor (16.6%). Preeclampsia and major congenital anomalies were the leading contributors to indicated early preterm births. The distribution of the precursors leading to delivery were relatively similar across the two gestational age groupings.
Univariable analyses comparing attempted VD to planned CD for gestational age subgroups stratified by presentation are presented in Tables 3 and and4.4. At 24 0/7 – 27 6/7 weeks of gestation with a vertex presentation (N= 714), 22.8% underwent planned CD and 77.2% attempted VD. Of those with attempted VD, 84.8% had a successful vaginal delivery. The planned CD pregnancies were more likely to be multiparous, have diabetes or have preeclampsia and less likely to have PPROM and preterm labor when compared to the attempted VD group (P < 0.05). (Table 3) Overall neonatal mortality in the group was 14.8% (106/714). When attempted VD was compared to planned CD, there were no differences in neonatal mortality or other neonatal outcomes. (Table 4)
At 24 0/7 – 27 6/7 weeks’ gestation with a breech presentation, (N= 388), 68.3 % underwent planned CD and 31.7 % attempted VD. Of those with attempted VD, only 27.6% had a successful vaginal delivery. By univariable analyses, the planned CD pregnancies were more likely to have preeclampsia and less likely to be complicated by preterm labor when compared to the attempted VD group (P < 0.05). (Table 3) Overall neonatal mortality in this group was 17.0% (66/388). When compared to planned CD, attempted VD was associated with a higher rate of neonatal mortality (25.2% versus 13.2%, P=0.003), but a lower rate of neonatal sepsis (P=0.01) and decreased need for neonatal ventilation (P=0.023). (Table 4)
At 28 0/7 – 31 6/7 weeks’ gestation with a vertex presentation, (N=1,424), 20.1% underwent planned CD and 79.9% attempted VD. Of those with attempted VD, 84.3% had a successful vaginal delivery. By univariable analyses, the planned CD pregnancies were more likely to have private insurance, be multiparous, have diabetes or have preeclampsia and less likely to have PPROM and preterm labor when compared to the attempted VD group (P < 0.05). (Table 3) Overall neonatal mortality in this group was 2.4% (34/1,424). When attempted VD was compared to planned CD, there were no differences in neonatal mortality. Compared to planned CD, attempted VD was associated with lower rates of perinatal asphyxia (P = 0.004), fewer cases of RDS (P=0.003) and a reduced need for assisted ventilation (P=0.004) but higher rates of IVH (P=0.0017). (Table 4)
At 28 0/7 – 31 6/7 weeks’ gestation with a breech presentation (N= 380), 69.5% underwent planned CD and 30.5% attempted VD. Of those with attempted VD, only 17.2% had a successful vaginal delivery. By univariable analyses, the planned CD pregnancies were more likely to have preeclampsia and less likely to be complicated by preterm labor when compared to the attempted VD group (P < 0.05). (Table 3) Overall neonatal mortality in this group was 2.9% (11/380). Attempted VD compared to planned CD was associated with increased neonatal mortality (6.0% vs. 1.5%, P= 0.016). (Table 4)
Multivariable analyses were then performed examining the previously defined neonatal outcomes as well as a composite outcome of death or asphyxia. (Table 5) For vertex presentation at 24 0/7 – 27 6/7 weeks’ gestation, there were no differences in individual neonatal outcomes or in the composite outcome between attempted vaginal delivery and planned CD. However, for breech presentation in this gestational age range, there was a three-fold increase in mortality with attempted vaginal delivery (RR = 3.0, 95% CI 1.8, 5.1) and a 2.4 fold increase in the composite outcome of death or asphyxia (RR =2.4, 95% CI 1.5–4.0). The differences in the risks of neonatal sepsis and need for ventilation seen on univariable analysis were no longer significant. Vertex presentation with attempted vaginal delivery at 28 0/7 – 31 6/7 weeks’ gestation was not associated with an increased risk of neonatal mortality or with the composite outcome of death or asphyxia, but was associated with an increased risk of IVH (RR=1.8, 95% CI 1.1, 2.9) when compared to planned CD for vertex presentation in this gestational age range. For breech presentation at this gestational age range, attempted vaginal delivery was associated with an increased neonatal mortality (RR =5.1, 95% CI 1.3, 19.9) compared to planned CD. (Table 5) Sensitivity analysis was performed by removing the 872 subjects who had missing values for maternal age, pre-pregnancy BMI or birth weight. There was no change in the main findings of our study with the sensitivity analysis based on the remaining 2,034 subjects with complete information. For vertex presentation, there was no increase in neonatal mortality with attempted vaginal delivery compared to planned CD. For breech presentation, there remained an increased neonatal mortality rate with attempted vaginal delivery compared to planned CD.(data not shown) Multivariable analyses was then also performed in the subset of 5 sites reporting information on antenatal corticosteroids (ACS) administration (N=1,094). 75.4% (825/1,094) reported administering ACS prior to delivery. When ACS use was added to the multivariable model there was essentially no change in the neonatal mortality results for the 24–27 weeks’ gestation group. There was no increased risk of neonatal mortality for vertex presentation (RR=1.34, 95% CI 0.49, 3.64) and there remained an increased risk of neonatal mortality for breech presentation (RR=4.0, 95% CI 1.3, 12.2) even after accounting for ACS administration. There were too few deaths in 28–31 weeks’ gestation (n= 13 deaths) to evaluate the effects of route of delivery stratified by presentation in this subset with information on ACS use. However, there was no longer an increased risk of IVH with attempted VD at 28–31 weeks’ gestation in vertex presentation after controlling for ACS administration (RR=1.57, 95% CI 0.72–3.40).
Despite the uncertainty of benefit, the use of cesarean delivery for infants at less than 32 weeks of gestation has increased over the last twenty years as survival has improved. 21 In this large recent cohort with detailed medical record data, PPROM and indicated deliveries accounted for the majority of early preterm births; preeclampsia was the most common obstetric reason for indicated preterm delivery. This distribution differs from the usually quoted rates of about 30–35% of preterm births being indicated, 40–45% due to spontaneous preterm labor, and 25–30% due to PPROM.22 This difference likely results from our focus on preterm births < 32 weeks’ gestation rather than on all preterm birth (< 37 weeks’ gestation).
In contrast to some previously published reports, 2–4 we found that attempting vaginal delivery with vertex presentation at 24 0/7 – 27 6/7 weeks of gestation did not significantly affect neonatal mortality. Furthermore, in such situations, more than 80% of attempted vaginal deliveries were successful. The findings, however, were different at 24 0/7 – 27 6/7 weeks of gestation if the fetal presentation was breech. In this subgroup, the majority of deliveries were by planned CD and the failure rate of attempted vaginal delivery was high. Planned cesarean delivery was associated with significantly lower neonatal mortality.
The findings in the deliveries at 28 0/7 – 31 6/7 weeks’ gestation also differed by presentation. In this subgroup, if the fetal presentation was vertex, the majority of attempted vaginal deliveries succeeded. There was also no difference in the neonatal mortality rate compared to planned CD. It is important to note that the risk of IVH in the subgroup at 28–31 weeks of gestation in vertex presentation was no longer increased in the attempted vaginal delivery group after accounting for antenatal corticosteroid administration. For breech-presenting fetuses, attempting vaginal delivery was less successful and planned CD was associated with lower neonatal mortality.
Previous studies examining the effect of route of delivery on neonatal mortality for early preterm births with a vertex presentation have defined route of delivery by actual, not attempted, delivery route. Two studies based on U.S. birth certificate data found that actual CD was associated with decreased neonatal mortality for vertex presentation 500–749 grams2 as well as up to 1,300 grams.3 The detailed information available in our study, but not available in birth certificate data, enabled us to account for attempted route of delivery as well as indications for delivery on neonatal mortality. This information is crucial for counseling families about the benefits and risks of attempting vaginal delivery in this situation. Our results are consistent with the findings of 3 previous studies that analyzed singleton vertex-presenting neonates, less than 1500 grams10,12 as well as those born at 24–34 weeks of gestation,11 which showed no survival advantage for cesarean delivery.
For breech pregnancies less than 32 weeks of gestation, planned CD was associated with lower neonatal mortality in our study. This finding has been noted in other studies based on birth certificate data 2, 4–9 but not in other, smaller studies which use detailed medical record data from individual institutions. 13–15
There are numerous strengths to this study. Examining the attempted route of delivery instead of actual route of delivery and controlling for indications leading to early preterm delivery is a major strength over previous studies because this information has direct clinical applications. We were able to control for multiple maternal, demographic, and obstetric covariates such as gestational age that influence neonatal outcome when examining planned delivery route. Some studies are based solely on birth weight and not gestational age which is a limitation since gestational age is more directly related to neonatal outcome. Other strengths include using a relatively large recent cohort of early preterm deliveries with medical record level data available from multiple hospitals throughout the U.S rather than relying upon birth certificate data. Previous studies have been single-center or multicenter trials with relatively small sample sizes and therefore underpowered to demonstrate a difference or accumulated larger numbers by evaluating outcomes over a decade, in which time neonatal intensive care has improved.
The limitations of this study include lack of information regarding antenatal corticosteroid administration in the entire cohort. However, recent data demonstrates that 85% of infants less than 28 weeks of gestation are exposed to antenatal steroids.23 Information regarding the grade of IVH and NEC was not collected. In addition, because the decision on mode of delivery was made by the physician and influenced by local practice, bias may have been introduced in that those with perceived worse outcome potential may have been more likely to be delivered vaginally. Likewise, neonatal outcomes were defined clinically as recorded in the patient’s medical records, so criteria for defining these conditions may have varied among hospitals. Misclassification may have occurred in that a planned CD may have been classified as an attempted vaginal delivery: eg. breech presentation that has progressed through labor quickly and delivers vaginally but would have been a planned CD if the woman was earlier in the course of labor. This misclassification may result in bias in either direction. Lastly, there are limited data on maternal outcomes and long-term neurological and physical disability outcomes of these infants which is important when evaluating the overall risks and benefits of either route of delivery.
Selecting a route of delivery at less than 32 weeks’ gestation is a difficult clinical decision given the high rate of neonatal mortality and morbidity as well as the maternal risks associated with cesarean delivery and requires detailed counseling. For vertex-presenting fetuses less than 32 weeks’ gestation, there was no improvement in neonatal mortality with planned CD. For breech-presenting fetuses less than 32 weeks’ gestation, attempting vaginal delivery was associated with a low success rate and planned CD was associated with lower neonatal mortality rates. Because it is unlikely that a randomized control trial will ever be conducted in the U.S.24, research with prospective data collection is needed to further delineate the short and long term benefits and risks of attempted route of delivery in the early preterm gestation.
The authors would like to thank Mr. Maurice Davis for assistance with the figures and tables in this manuscript.
The data included in this paper were obtained from the Consortium on Safe Labor, which was supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, National Institutes of Health, through Contract No. HHSN267200603425C. Institutions involved in the Consortium include, in alphabetical order: Baystate Medical Center, Springfield, MA; Cedars-Sinai Medical Center Burnes Allen Research Center, Los Angeles, CA; Christiana Care Health System, Newark, DE; Georgetown University Hospital, MedStar Health, Washington, DC; Indiana University Clarian Health, Indianapolis, IN; Intermountain Healthcare and the University of Utah, Salt Lake City, Utah; Maimonides Medical Center, Brooklyn, NY; MetroHealth Medical Center, Cleveland, OH.; Summa Health System, Akron City Hospital, Akron, OH; The EMMES Corporation, Rockville MD (Data Coordinating Center); University of Illinois at Chicago, Chicago, IL; University of Miami, Miami, FL; and University of Texas Health Science Center at Houston, Houston, Texas.
Reprints not available
Disclosure: None of the authors have a conflict of interest
The opinions and views expressed in this paper are those of the authors and do not necessarily reflect those of the National Institutes of Health or the NICHD.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.