Objective
The objective of the study was to compare neonatal mortality and morbidity in very preterm twins with the first twin in cephalic presentation in hospitals with a policy of planned vaginal delivery (PVD) and those with a policy of planned cesarean delivery (PCD).
Study Design
Women with preterm cephalic first twins delivered after preterm labor and/or premature preterm rupture of membranes from 26 0/7 to 31 6/7 weeks of gestation were identified from the databases of 6 perinatal centers and classified as PVD or PCD according to the center’s management policy from 1999 to 2010. Severe neonatal morbidity was defined as any of the following: intraventricular hemorrhage grades 3-4, periventricular leukomalacia, necrotizing enterocolitis, bronchopulmonary dysplasia, and hospital death. The independent effect of the planned mode of delivery, defined by the center’s management policy, was tested and quantified with a 2-level multivariable logistic regression.
Results
The PVD group included 248 women, and the PCD group 63. Maternal characteristics did not differ between the 2 groups. The rate of vaginal delivery was 85.9% (213 of 248) vs 20.6% (13 of 63) ( P < .001), and the rate of cesarean delivery for the second twin was 1.6% (4 of 248) vs 4.8% (3 of 63) ( P = .13) for PVD and PCD. PVD had no independent effect on either newborn hospital mortality or severe neonatal composite morbidity.
Conclusion
A policy of planned vaginal delivery of very preterm twins with the first twin in cephalic presentation does not increase either severe neonatal morbidity or mortality.
The incidence of twin pregnancy has increased worldwide over the past 20 years by approximately 80%, largely because of the growing use of assisted reproductive technologies and the increase in average age at first childbirth. Twin deliveries today account for 2-3% of live births in France and other developed countries, but they account for approximately one third of very preterm births, that is, of babies born before 32 weeks of gestation.
Twin delivery remains a challenging event in daily obstetric practice, and this challenge is still more difficult in cases of very preterm birth. Several large retrospective population-based studies have suggested that cesarean delivery may decrease morbidity for term and preterm second twins. Although these studies based on large databases have excellent statistical power, they are limited by selection bias. For example, because there is no linkage to the original medical file, they lack information about maternal-fetal condition on admission and the reason for cesarean or vaginal delivery, and most are unable to compare planned vaginal (PVD) to planned cesarean (PCD) deliveries.
Instead they compare 3 groups: one with both twins delivered by cesarean delivery (C-C), one with both twins delivered vaginally (V-V), and another with the second twin delivered by cesarean delivery after vaginal delivery of the first twin (V-C). Because this kind of analysis does not take the intended treatment into account, it may introduce a selection bias by allocating to the vaginal delivery group (V-V) the cases for whom a cesarean is not performed because of a poor prognosis. Conversely, allocating to the cesarean delivery group (C-C) cases in which fetal condition is uncertain (fetal growth restriction, for instance) may introduce bias by worsening neonatal prognosis is this group.
Recently a large multicenter randomized controlled trial showed that PCD does not decrease neonatal morbidity for either first or second twins compared with a PVD for twins delivered after 32 weeks’ gestation (Twin Birth Study). These data conflict with the results of the large retrospective population-based studies described in the previous text and underline the weaknesses of the latter.
Before the Twin Birth Study, both the American and French Colleges of Obstetricians and Gynecologists considered that available data were insufficient to allow recommendations to be issued about the best route of delivery of first twins in cephalic presentation before 32 weeks of gestation. Obstetricians are nonetheless troubled about the risk of neonatal morbidity and mortality related to complications of vaginal delivery. This is a major concern for second twins, in particular, because of the possibly higher risk of hypoxia following obstetric delivery maneuvers, cord prolapse, or premature placental separation, especially for very low birthweight infants.
A recent Cochrane systematic review focusing on the best mode of delivery for preterm infants concluded that recruiting difficulties are likely to make a randomized study on this topic impossible. The best alternative study design uses observational studies comparing hospitals with different policies. Consequently, we investigated neonatal outcomes for very preterm twins when the first twin was cephalic by comparing centers with policies of either PVD or PCD.
Materials and Methods
This retrospective study included all consecutive women with twin pregnancies with a first twin in cephalic presentation admitted for preterm labor who gave birth between 26 0/7 and 31 6/7 weeks’ gestation, except those meeting the exclusion criteria. These women were identified from the databases of 6 perinatal centers and classified as PVD or PCD according to the center’s management policy from 1999 to 2010.
PVD was standard policy at center 1 (Port Royal Maternity Hospital), center 2 (Rouen University Hospital), center 3 (Créteil University Hospital) and center 4 (Montreuil General Hospital), whereas PCD was standard policy at center 5 (Angers University Hospital) and center 6 (Poissy University Hospital). Each center retrieved the paper files for each patient, and 3 independent investigators (A.O., E.N., and A.-C.B.) carefully examined the clinical notes from all files. The Ethics Committees of all institutions approved this protocol (2013 78).
Women were excluded from the study if the first twin’s presentation was noncephalic at the time the patient presented for delivery and in all cases likely to introduce confounding biases and increase neonatal morbidity in 1 of the 2 groups, such as monoamniotic pregnancy, induced preterm delivery, including for preeclampsia (defined according to criteria from the International Society for the Study of Hypertension in Pregnancy ), abruptio placentae, placenta previa, twin-to-twin transfusion syndrome (defined according to the Quintero staging system ), fetal growth restriction (FGR) (defined as fetal abdominal circumference below the 10th percentile and/or estimated fetal weight below the 10th percentile ), delayed birth of the second twin (defined as a delivery interval between the 2 twins longer than 2 hours), fetal death, and any prenatally diagnosed malformation of either twin.
Except for the policy of mode of delivery, standard procedures for the management of twin pregnancies as well as of preterm labor and premature preterm rupture of membranes (PPROM) were similar in all centers and did not change during the study period. In these 6 tertiary referral centers, all obstetrics records are reviewed every morning by a senior obstetrician during the daily staff meeting to verify compliance with these protocols and to improve their enforcement.
Gestational age was established by the last menstrual date and the first-trimester ultrasound scan, if available, as it generally is in France. The scan date was preferred if the menstrual date was uncertain or if there was discrepancy of more than 5 days between the 2 estimates. Ultrasound examination including an estimation of fetal weight and Doppler umbilical artery velocimetry was routinely performed every 2 weeks for monochorionic and every 4 weeks for dichorionic twin pregnancies.
Tocolytics were administered according to each hospital’s protocol up to 33 weeks unless there were laboratory or clinical symptoms of chorioamnionitis. Patients with PPROM received 2 g/d of amoxicillin for 7 days at admission. Prenatal corticosteroids were administered to women admitted for preterm labor or PPROM to enhance fetal lung maturity. Fetal heart rate (FHR) monitoring was performed at least once a week before labor and continuously during labor for both twins. Epidural analgesia was routinely offered.
All vaginal twin deliveries in both groups were performed by a senior attending obstetrician. Second twins in breech presentation or transverse position were usually delivered by total breech extraction. For those in cephalic presentation, internal version was permitted at the discretion of the attending physician.
The hospitals with a PVD policy performed cesarean deliveries when FHR was abnormal during labor, when labor was protracted, and in cases of chorioamnionitis before labor. Units with a PCD policy limited vaginal deliveries to cases in which labor progressed too rapidly to perform a cesarean delivery.
Neonatal medical records were reviewed to identify the following outcomes: neonatal death; birth trauma, including subdural hematoma, spinal cord injury, basal skull fracture, peripheral-nerve injury, or clinically significant genital injury; 5 minute Apgar score less than 7, which has been linked to adverse neonatal outcomes in preterm infants ; bronchopulmonary dysplasia, defined by the need for oxygen at 28 days of life; intraventricular hemorrhage grades 3 and 4, according to the cranial ultrasound findings and Papile’s criteria ; periventricular leukomalacia, defined as hyperechoic lesions persisting to day 7 of life ; neonatal sepsis, defined by clinical findings suggesting infection and a positive result on a blood or cerebrospinal fluid culture or the need for vasopressors to reverse sepsis-induced hypotension until discharge from the neonatal intensive care unit ; necrotizing enterocolitis grades II/III according to Bell’s classification ; and the duration of intensive care.
The primary outcome was severe neonatal morbidity. A composite outcome measure was based on the composite measures used in other trials of severe preterm birth and was defined by any one of the following: intraventricular hemorrhage grades 3 and 4, periventricular leukomalacia, necrotizing enterocolitis, bronchopulmonary dysplasia, and hospital death.
Continuous variables were analyzed by an analysis of variance, and categorical variables were compared with χ 2 or Fisher exact tests. The independent effect of the planned mode of delivery according to the center’s management policy was tested and quantified with a 2-level multivariable logistic regression with a random intercept to take into account the hierarchical structure of the data, with women clustered in maternity units.
We built a pragmatic model by including in a core model all available potential explanatory factors identified from the literature and prior hypotheses: gestational age at birth (by weeks of gestation), antenatal corticotherapy (yes/no), fetal sex (male/female), geographic origin (white, sub-Saharan Africa and the West Indies, North Africa, Asia and others), and for PPROM (preterm labor with intact membranes/PPROM).
We used Stata version 12 software for all analyses (StataCorp, College Station, TX).
Results
From 1999 to 2010, there were 191,931 births in the 6 university tertiary hospital centers. The total number of very preterm singletons and twins born during this time period was, respectively, 1907 and 1334 (667 twin pairs), for an overall incidence of 1.0% of preterm births and 0.3% of preterm twin births. After excluding 356 twin pairs (for breech or transverse presentation of the first twin [n = 102], hypertensive disease of pregnancy or fetal growth restriction [n = 73], previous cesarean [n = 46], monochorionic pregnancy [n = 13], twin-twin-transfusion syndrome [n = 63], prelabor fetal death [n = 44], severe congenital disease or malformation (n = 13), and delayed birth of the second twin [n = 2]), there were 311 twin pairs (622 twins) available for analysis.
The PVD group included 248 women and the PCD group 63. The rates of very preterm twins among all very preterm births did not differ significantly between the PVD and PCD groups either before (35.0% [527 of 1505] vs 33.0% [140 of 424]; P = .48) or after exclusions (16.5% [248 of 1505] vs 14.9% [63 of 424]; P = 0.45).
As shown in Table 1 , maternal and obstetric characteristics did not differ between the 2 groups except for the presentation of the second twin, the mode of delivery, and the intertwin delivery interval, which was longer in hospitals with a PVD policy. In the PCD group, 13 women (20.6%) had a vaginal delivery because the first stage of labor concluded or labor progressed too rapidly to perform a cesarean delivery ( Table 1 ). In the PVD group, 12 women (4.8%) had a cesarean delivery before labor for abnormal FHR or chorioamnionitis, 213 (85.9%) delivered both twins vaginally, and 23 (9.3%) had a cesarean delivery during labor, including 4 (1.6%) who had a cesarean delivery for the second twin after vaginal birth of the first twin. The intertwin delivery interval was less than 10 minutes in at least 90% of cases for both groups.
| Characteristic | PVD (n = 248) | PCD (n = 63) | P value |
|---|---|---|---|
| Maternal characteristics | |||
| Maternal age, y | 30.5 (4.8) | 29.4 (5.8) | .15 |
| Geographic origin | .11 | ||
| White | 167 (67.3) | 50 (79.4) | |
| Sub-Saharan Africa and the West Indies | 18 (7.3) | 1 (1.6) | |
| North Africa | 39 (15.7) | 5 (7.9) | |
| Asia and others | 24 (9.7) | 7 (11.1) | |
| Parity >0 | 81 (32.7) | 21 (33.3) | .92 |
| Iatrogenic twins | 103 (41.5) | 26 (41.3) | .91 |
| Chorionicity (n = 309) a | .76 | ||
| Dichorionic diamniotic twins | 207 (84.1) | 52 (82.5) | |
| Monochorionic diamniotic twins | 39 (15.9) | 11 (17.5) | |
| Antenatal corticosteroid therapy | 237 (95.6) | 60 (95.2) | .91 |
| PPROM | 117 (47.2) | 35 (55.6) | .24 |
| Gestational age at PPROM | 28.2 (2.9) | 28.0 (4.5) | .63 |
| Histological chorioamnionitis (n = 230) a | 61 (30.3) | 8 (27.6) | .76 |
| Obstetric characteristics | |||
| Gestational age at delivery, wks | .68 | ||
| 26–27 | 50 (20.2) | 10 (15.9) | |
| 28–29 | 81 (32.7) | 20 (31.7) | |
| 30–31 | 117 (47.1) | 33 (52.4) | |
| Presentation of the second twin (n = 303) a | .03 | ||
| Cephalic | 111 (45.9) | 17 (27.9) | |
| Breech | 81 (33.5) | 30 (49.2) | |
| Transverse | 50 (20.6) | 14 (22.9) | |
| Mode of delivery | < .001 | ||
| Vaginal | 213 (85.9) | 13 (20.6) | |
| Cesarean during labor | 23 (9.3) | 42 (66.7) | |
| Cesarean before labor | 12 (4.8) | 8 (12.7) | |
| Cesarean for second twin | 4 (1.6) | 3 (4.8) | .13 |
| Intertwin delivery interval (n = 303) a | < .001 | ||
| <5 min | 143 (58.6) | 53 (89.8) | |
| 5–10 min | 77 (31.6) | 2 (3.4) | |
| >10 min | 24 (9.8) | 4 (6.8) |
Severe neonatal morbidity and mortality did not differ between the PCD and PVD groups for first or second twins after either univariate or multivariable analysis ( Tables 2 and 3 ).
| Neonatal characteristic | PVD (n = 248) | PCD (n = 63) | OR (95% CI) | aOR (95% CI) a |
|---|---|---|---|---|
| Male b | 140 (56.5) | 34 (54.0) | — | — |
| Birthweight, g, median (range) b | 1312 (790–2070) | 1392 (700–2140) | — | — |
| Apgar score <7 at 5 min | 17 (6.9) | 4 (6.3) | 1.09 (0.35–3.35) | 0.77 (0.23–2.56) |
| Duration of hospitalization in ICU (d), mean ±SD b | 26 ± 51 | 20 ± 19 | ||
| Traumatism | 0 | 0 | — | — |
| Convulsions | 0 | 1 | — | — |
| Neonatal sepsis | 24 (9.7) | 13 (20.6) | 0.43 (0.20–0.89) | 0.39 (0.18–0.88) |
| Necrotizing enterocolitis | 10 (4.0) | 3 (4.8) | 0.84 (0.22–3.14) | 0.86 (0.22–3.40) |
| Bronchopulmonary dysplasia | 38 (15.3) | 9 (14.3) | 1.09 (0.49–2.38) | 1.1 (0.29–4.36) |
| Intraventricular hemorrhage grades 3–4 | 12 (4.8) | 5 (7.9) | 0.60 (0.21–1.79) | 0.41 (0.12–1.41) |
| Periventricular leukomalacia | 6 (2.4) | 3 (4.8) | 0.50 (0.12–2.09) | 0.47 (0.11–2.01) |
| Death in the delivery room | 0 | 0 | — | — |
| Neonatal death | 13 (5.2) | 4 (6.3) | 0.83 (0.26–2.65) | 0.78 (0.17–3.68) |
| Severe neonatal composite morbidity | 78 (31.5) | 22 (34.9) | 0.86 (0.48–1.53) | 0.71 (0.36–1.44) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
