The incidence of twin gestations has increased during the past decades worldwide and now accounts for approximately 3% of all deliveries,1,2 but up to 20% and 30% of preterm and very preterm births (ie, before 37 and 32 weeks of gestation), respectively.1–3 The optimal mode of delivery of preterm twins remains debatable.4–6 Several large retrospective population-based studies have shown an association between vaginal delivery and increased neonatal mortality and morbidity, mainly for the second twin, suggesting that a planned cesarean delivery may be the best option for preterm second twins.7–10 However, these studies suffer the major drawback of selection bias as a result of the lack of information about maternal–fetal condition at admission or the clinical condition leading to a cesarean or a vaginal delivery.11,12 Moreover, most of these studies were unable to compare planned vaginal delivery with planned cesarean delivery7,8,10 and none has assessed mid- or long-term neonatal outcome.13
As a result of recruiting difficulties, it is likely that a randomized study will never be conducted.14 Consequently, national, observational, prospective, population-based cohort studies such as the Epidémiologique sur les Petits Ages Gestationnels 2 cohort, which focused on preterm births, provide a unique opportunity to assess the perinatal risks associated with the planned mode of delivery in an unselected population of women with preterm twin pregnancies.15 Therefore, we investigated both neonatal and 2-year outcomes for preterm twins with the first twin in cephalic presentation by comparing planned vaginal and planned cesarean deliveries in the Epidémiologique sur les Petits Ages Gestationnels 2 cohort sample using propensity score to ensure comparability of the study groups.
Data came from the Epidémiologique sur les Petits Ages Gestationnels 2 study, a population-based cohort study that included all preterm children born between 22 and 34 weeks of gestation (termination of pregnancies, stillbirths, and live births) in 546 maternity units in 25 French administrative regions (all but one) in 2011.15 The Epidémiologique sur les Petits Ages Gestationnels 2 study was implemented to describe the short- and long-term outcomes of preterm neonates in terms of birth circumstances, medical practices, and organization of care. The Epidémiologique sur les Petits Ages Gestationnels 2 study protocol has been published previously.16 Children born at 22–26 weeks, 27–31 weeks, and 32–34 weeks of gestation were included for 8 months, 6 months, and 5 weeks, respectively. Different stages of follow-up are planned up to age 12 years. Maternal, obstetric, and neonatal data were collected prospectively from the obstetric and neonatal records following a standardized protocol. At 2 years of corrected age, for children included in the follow-up, a detailed neurologic and sensory examination was performed by the referring physician. Details on protocol, questionnaires, data collection, and data quality control can be found elsewhere.16 The appropriate ethics committees, specifically the consultative committee on the treatment of information on personal health data for research purposes (No. 10.626) and the committee for the protection of people participating in biomedical research (No. CPP SC-2873), approved this study as did the National Data Protection Authority (CNIL no. 911009).
This study was a planned analysis from the Epidémiologique sur les Petits Ages Gestationnels 2 cohort. The study population included all women with cephalic first twin pregnancies delivered after preterm labor (defined as contractions associated with cervical dilation and rupture of membranes less than 24 hours before birth) or preterm prelabor rupture of membranes (PROM) (rupture of membranes more than 24 hours before birth) from 24 0/7 to 34 6/7 weeks of gestation. There were multiple exclusion criteria (Fig. 1).
Exposure was the planned route of delivery. Planned vaginal delivery was defined as vaginal delivery or cesarean delivery performed during labor for an abnormal fetal heart rate, failure to progress, or for the second twin only. Planned cesarean delivery was one performed during labor for the indication of preterm twins or before labor whatever the indication. Women with a history of two or more cesarean deliveries were allocated to planned cesarean delivery whatever the actual route of delivery because French guidelines recommend performing a cesarean delivery for such women.17
The primary outcome was survival, defined as the number of neonates discharged alive from the hospital relative to the number of fetuses alive at the beginning of labor or at the time of the decision to perform cesarean delivery before labor. The secondary outcome was survival to discharge without severe neonatal morbidity defined as survival without any of the following16: severe intraventricular hemorrhage defined as intraventricular hemorrhage associated with ventricular dilatation (grade III intraventricular hemorrhage) and intraparenchymal hemorrhage (ie, large unilateral parenchymal hyperdensity or large unilateral porencephalic cyst)18; cystic periventricular leukomalacia (ie, periventricular white matter echolucencies at ultrasonography); stages II and III necrotizing enterocolitis according to Bell’s staging19; stage 3 or greater retinopathy of prematurity according to the international classification (International Committee for the Classification of Retinopathy of Prematurity),20 laser treatment, or both; and severe bronchopulmonary dysplasia defined as requiring oxygen for at least 28 days plus the need for 30% or more oxygen, mechanical ventilator support, or both or continuous positive airway pressure at 36 weeks of gestation.21 The third outcome was survival without severe or moderate neuromotor or sensory disabilities at 2 years of corrected age, defined as survival without cerebral palsy, blindness, deafness, or all of these.22 Cerebral palsy was diagnosed if the child had permanent disorders of movement, posture and disorders of motor function (not walk without aids) as a result of a nonprogressive lesion located in the developing brain, or both.23 The following variables related to neonatal outcomes during the hospital stay were also included in the analysis: endotracheal intubation in the labor ward, attempt of continuous passive airway pressure in the first 24 hours of life, thrombocytopenia before day 7, transfusion of blood products, phototherapy, and median length of hospital stay (among survivors).
At a pregnancy level, we described the following variables: mother’s age, nationality, social insurance, socioeconomic status, marital status, body mass index, parity, history of previous cesarean delivery, chorionicity, spontaneous onset of labor before admission, antenatal steroids, tocolysis treatment, and suspicion of chorioamnionitis.25 Spontaneous labor at admission was defined as direct admission to the delivery room with cervix dilation greater than 2 cm, an interval of less than 10 hours from admission to birth, or both. We did not report the use of magnesium sulphate for neuroprotection, which was only recommended after our study period in women with spontaneous preterm labor. At a fetal or neonatal level, we investigated the diagnosis or suspicion of fetal growth restriction, defined by estimated fetal weight below the 10th percentile according to the health care providers’ reference curves, the presence of a significant abnormal heart rate during the last prenatal monitoring, the presentation of the second twin, the gestational age of delivery, sex of the neonate, birth weight, Apgar score at 5 minutes, and umbilical artery pH. Gestational age was determined as the best obstetric estimate combining last menstrual period and ultrasonography assessment.
Demographic and clinical characteristics and outcomes were first described (frequencies and percentages) and then compared between planned vaginal delivery and planned cesarean delivery groups by Rao-Scott χ2 for categorical variables to account for differences in sampling process between gestational age groups. Medians of quantitative variables were compared by Kruskal-Wallis tests. Outcomes were reported separately for the first and second twins. We also explored outcomes at the pregnancy level using a four-category variable: no twin survived, first twin survived, second twin survived, and both twins survived. Descriptive analyses were weighted to account for the sampling scheme of the cohort and to ensure representativeness: weights were 1.0 for births at 24–26 weeks of gestation, 1.34 at 27–31 weeks of gestation, and 7.0 at 32–34 weeks of gestation.15,16,22,25 We used marginal model (generalized estimating equations) to take into account the correlation between twins.
We used a propensity score analysis to minimize indication bias in planned delivery route. The propensity score reflects the likelihood of planning a cesarean delivery rather than a vaginal delivery depending on the woman’s baseline characteristics. It was estimated with a logistic regression model with planned cesarean delivery as the dependent variable regressed by baseline characteristics selected from the literature and clinical considerations: mother’s age, mother’s country of birth, parents' socioeconomic status, parity, antenatal steroids, tocolysis, cause of preterm labor or preterm PROM, suspicion of clinical chorioamnionitis, spontaneous labor at admission, chorionicity, gestational age, fetal growth restriction for at least one twin, presentation of the second twin, and birth in a level 3 maternity unit. Inverse probability of treatment weighting based on estimated propensity scores was then used to obtain a synthetic population in which planned delivery route was independent of measured baseline covariates, as confirmed by balance diagnostics. Standardized differences between groups in the inverse probability of treatment weighting cohort were less or equal than 10% for all variables included in the propensity score, except for birth in level 3 maternity units (SD 11%) (Appendix 2, available online at http://links.lww.com/AOG/B222). Finally, the association of planned delivery route and the three outcomes was quantified by odds ratios and 95% CIs estimated from a marginal model (generalized estimating equation) with further adjustments on gestational age alone and gestational age, fetus sex, and birth weight as supplementary analyses.
Missing data were imputed by chained equations using the SAS “MI” procedure. The imputation model involved use of variables and outcomes of the propensity score model with a logistic regression imputation model for binary variables and a multinomial imputation model for categorical variables. Propensity scores were estimated for each of the 50 independent imputed data sets generated with 30 iterations each, and results were pooled in a single estimate according to Rubin’s rules.26 Results from survival at discharge and survival at discharge without severe morbidity are reported for both complete cases and all cases after multiple imputation. At 2 years of corrected age, 15% of surviving children eligible were lost to follow-up; their disability statuses were unknown. To account for selective dropouts, survival at 2 years without disabilities was based only on data using multiple imputations.
We also performed different sensitivity analyses of births at 26 0/7 to 31 6/7 weeks of gestation: women who were not in labor at admission and women with no previous cesarean deliveries.
SAS 9.4 was used for data analysis. Statistical significance was set at two-tailed P<.05.
Among the 932 women with twin pregnancy included in the Etude Epidémiologique sur les Petits Ages Gestationnels 2 study, 549 women (59%) were excluded (Fig. 1). Thus, 383 women (766 fetuses) met our inclusion criteria and were available for analysis. The planned vaginal delivery group included 276 (72.1%) women and the planned cesarean delivery group 107 (Fig. 1).
Rates of previous cesarean delivery, tocolysis, suspicion of fetal growth restriction for the first twin, chorioamnionitis, and significant abnormal heart rate for first and second twins were significantly more frequent in the planned cesarean delivery group. Planned vaginal delivery was more frequent for nulliparas, women in spontaneous labor at admission, and women who delivered at more advanced gestational age (Table 1). In the planned vaginal delivery group, 207 (80.2%) women delivered both twins vaginally, and 68 (19.8%) had a cesarean delivery including 16 (2.9%) who had a cesarean delivery for the second twin after vaginal birth of the first twin. In the planned cesarean delivery group, only one neonate (0.4%) was delivered vaginally because labor progressed too rapidly to perform a cesarean delivery (Table 1).
There were no neonatal deaths in the delivery room in the planned cesarean delivery group; there was one death of a first twin (0.1%) and three deaths (0.4%) of second twins in the planned vaginal delivery group (Tables 2 and 3). However, first and second twins with planned vaginal delivery and planned cesarean delivery did not differ in survival at discharge (97.7% vs 98.3%; P=.57 and 95.6% vs 97.1%; P=.39, respectively), survival at discharge without severe morbidity (91.6% vs 88.8%; P=.42 and 90.3% vs 84.5%; P=.18, respectively), or survival at 2 years without neurosensory impairment (96% vs 95.4%; P=.82 and 92.9% vs 90.8%; P=.59, respectively) (Tables 2 and 3). Planned cesarean delivery did not improve these three neonatal outcomes when the vital twin status was taken into account (Appendix 3, available online at http://links.lww.com/AOG/B222).
After applying propensity scores and assigning inverse probability of treatment weighting, as compared with planned vaginal delivery, planned cesarean delivery was not associated with improved outcomes (Table 4).
Analyses of these outcomes in the inverse probability of treatment weighting cohort with complete cases (Appendix 4, available online at http://links.lww.com/AOG/B222) or adjusted on gestational age alone or with fetus sex and birth weight produced similar results (Appendix 5, available online at http://links.lww.com/AOG/B222). Sensitivity analysis of women who were not in labor at admission and women who delivered at 26–31 weeks of gestation or with no previous cesarean deliveries provided also consistent nonsignificant results (Appendix 6, available online at http://links.lww.com/AOG/B222).
In this study, planned cesarean delivery for women in preterm labor or with preterm PROM was not associated with improved neonatal outcomes or 2-year outcomes. It is difficult to compare our results with the literature because previous studies assessing mode of delivery in preterm twin pregnancies are all retrospective, with their inherent biases.7–11,27 Nevertheless, it is interesting to underline that there are in accordance with a previous multicenter retrospective study that aimed to compare neonatal mortality and morbidity in hospitals with a policy of planned vaginal delivery (n=248) and those with a policy of planned cesarean delivery (n=63). Using a two-level multivariable logistic regression with a random intercept to take into account the hierarchical structure of the data with women clustered in maternity units, planned vaginal delivery had no independent effect on either newborn hospital morbidity or severe neonatal composite morbidity.11 Moreover, in the JUmeaux MODe d’Accouchement (JUMODA) study, a French national prospective population-based cohort study including women with twin pregnancies and their neonates born at or after 32 weeks of gestation with a cephalic first twin, recruited in 176 maternity units in France from February 2014 to March 2015, the propensity-matched analyses conducted on the subgroup of women who delivered between 32 0/7 weeks of gestation and 34 6/7 weeks of gestation, showed that planned cesarean delivery was associated with higher composite neonatal mortality and morbidity.12 Consequently, it seems unlikely that planned cesarean delivery has a protective effect on neonatal outcome in twin pregnancies when the first twin is cephalic.
Our study has several strengths. First, outcomes were assessed by the planned mode of delivery, rather than the actual one, the latter of which may lead to classification bias.11,13,27 Data including details relevant to clinical decision13,27 came from a large population-based cohort design and prospective enrollment of neonates born prematurely, which ensured external validity. To control for the indication bias, we performed propensity score analysis using different models to minimize the likelihood of incorrectly attributing any association to planned cesarean delivery. These analyses including sensitivity analysis also showed nonsignificant results and therefore confirm the robustness of statistical models. Furthermore, we also used a marginal model to take into account the nonindependency of observations within twin pairs. We followed children until 2 years of corrected age.
Nonetheless, our study has limitations; the main is that our study is a not a randomized controlled trial that would be theoretically the best study design to determine the safest mode of delivery for preterm twins. However, its feasibility of a randomized controlled trial seems difficult to perform in the context of (especially) extreme prematurity.11,16 Indeed, similarly, three studies were attempted to compare planned vaginal and cesarean delivery in cases of preterm breech and none was completed as a result of their lack of inclusion.28–30 Moreover, our findings are only generalizable to maternity units accustomed to active management of second twin delivery, as recommended4 and performed11,12,31,32 in France.
Attributing a case to planned vaginal delivery or planned cesarean delivery was performed a posteriori and misclassification cannot be totally excluded. In particular, we cannot exclude that a subgroup of vaginal deliveries should have been classified as planned cesarean delivery; for example, if labor progressed too quickly to perform a cesarean delivery and resulted in vaginal delivery. This classic misclassification can result in bias in either direction.11–13,27 The exclusion of women admitted after the beginning of active labor did not modify our findings, so, if any bias existed, it would be weak. Although our study included women (766 fetuses) from a large population-based cohort study, it still lacked statistical power to detect potentially clinically meaningful differences between planned modes of delivery. Finally, our results are not generalizable to all preterm twin pregnancies but only to those that met our inclusion criteria (ie, 41% of preterm twin pregnancies in the present study).
These limitations notwithstanding, our results showed that a planned cesarean delivery is not associated with improved neonatal outcomes for pregnancies in which the first twin is cephalic and birth occurs between 24 and 34 weeks of gestation after preterm labor or preterm PROM. Our results support the continued use of planned vaginal delivery in these situations for obstetricians experienced with vaginal twin deliveries.
1. Blondel B, Kermarrec M. Enquête nationale périnatale. Available at: http://sante.gouv.fr/IMG/pdf/Les_naissances_en_2010_et_leur_evolution_depuis_2003.pdf
. Retrieved September 1, 2018.
2. Hamilton BE, Martin JA, Osterman MJ, Curtin SC, Mathews TJ. Births: final data for 2014. Natl Vital Stat Rep 2015;64:1–64.
3. Larroque B, Ancel PY, Marret S, Marchand L, André M, Arnaud C, et al. Neurodevelopmental disabilities and special care of 5-year-old children born before 33 weeks of gestation (the EPIPAGE study): a longitudinal cohort study. Lancet 2008;371:813–20.
4. Vayssière C, Benoist G, Blondel B, Deruelle P, Favre R, Gallot D, et al. Twin pregnancies: guidelines for clinical practice from the French College of Gynaecologists and obstetricians (CNGOF). Eur J Obstet Gynecol Reprod Biol 2011;156:12–7.
5. Barrett JF, Ritchie WK. Twin delivery. Best Pract Res Clin Obstet Gynaecol 2002;16:43–56.
6. Sentilhes L, Sénat MV, Ancel PY, Azria E, Benoist G, Blanc J, et al. Prevention of spontaneous preterm birth: guidelines for clinical practice from the French College of Gynaecologists and Obstetricians (CNGOF). Eur J Obstet Gynecol Reprod Biol 2016;210:217–24.
7. Wen SW, Fung Kee Fung K, Oppenheimer L, Demissie K, Yang Q, Walker M. Neonatal mortality in second twin according to cause of death, gestational age, and mode of delivery. Am J Obstet Gynecol 2004;191:778–83.
8. Meyer MC. Translating data to dialogue: how to discuss mode of delivery with your patient with twins. Am J Obstet Gynecol 2006;195:899–906.
9. Armson BA, O’Connell C, Persad V, Joseph KS, Young DC, Baskett TF. Determinants of perinatal mortality and serious neonatal morbidity in the second twin. Obstet Gynecol 2006;108:556–64.
10. Herbst A, Källén K. Influence of mode of delivery on neonatal mortality in the second twin, at and before term. BJOG 2008;115:1512–7.
11. Sentilhes L, Oppenheimer A, Bouhours AC, Normand E, Haddad B, Descamps P, et al. Comparison of neonatal outcomes in cases a policy of planned vaginal versus cesarean delivery for severe preterm cephalic first twins. Am J Obstet Gynecol 2015;213:73.e1–7.
12. Schmitz T, Prunet C, Azria E, Bohec C, Bongain A, Chabanier P, et al. A national prospective population-based cohort study of the planned mode of delivery in twin pregnancies. Obstet Gynecol 2017;129:986–95.
13. Sentilhes L, Bouhours AC, Biquard F, Gillard P, Descamps P, Kayem G. Delivery of twins [in French]. Gynecol Obstet Fertil 2009;37:432–41.
14. Alfirevic Z, Milan SJ, Livio S. Caesarean section versus vaginal delivery for preterm birth in singletons. The Cochrane Database of Systematic Reviews 2013, Issue 9. Art No.: CD000078.
15. Ancel PY, Goffinet F, EPIPAGE-2 Writing Group, Kuhn P, Langer B, Matis J, et al. Survival and morbidity of preterm children born at 22 through 34 weeks' gestation in France in 2011: results of the EPIPAGE-2 cohort study. JAMA Pediatr 2015;169:230–8.
16. Ancel PY, Goffinet F; EPIPAGE 2 Writing Group. EPIPAGE 2: a preterm birth cohort in France in 2011. BMC Pediatr 2014;14:97.
17. Sentilhes L, Vayssière C, Beucher G, Deneux-Tharaux C, Deruelle P, Diemunsch P, et al. Delivery for women with a previous cesarean: guidelines for clinical practice from the French College of Gynaecologists and Obstetricians (CNGOF). Eur J Obstet Gynecol Reprod Biol 2013;170:25–32.
18. Volpe JJ. Brain injury in premature infants: a complex amalgam of destructive and developmental disturbances. Lancet Neurol 2009;8:110–24.
19. Bell MJ, Ternberg JL, Feigin RD, Keating JP, Marshall R, Barton L, et al. Neonatal necrotizing enterocolitis. Therapeutic decisions based upon clinical staging. Ann Surg 1978;187:1–7.
20. International Committee for the Classification of Retinopathy of Prematurity. The International Classification of Retinopathy of Prematurity revisited. Arch Ophthalmol 2005;123:991–9.
21. Jobe AH, Bancalari E. Bronchopulmonary dysplasia. Am J Respir Crit Care Med 2001;163:1723–9.
22. Pierrat V, Marchand-Martin L, Arnaud C, Kaminski M, Resche-Rigon M, Lebeaux C, et al. Neurodevelopmental outcome at 2 years for preterm children born at 22 to 34 weeks’ gestation in France in 2011: EPIPAGE-2 cohort study. BMJ 2017;358:j3448.
23. Surveillance of Cerebral Palsy in Europe. Surveillance of cerebral palsy in Europe: a collaboration of cerebral palsy surveys and registers. Surveillance of Cerebral Palsy in Europe (SCPE). Dev Med Child Neurol 2000;42:816–24.
24. Ego A, Prunet C, Lebreton E, Blondel B, Kaminski M, Goffinet F, et al. Customized and non-customized French intrauterine growth curves. I—methodology [in French]. J Gynecol Obstet Biol Reprod (Paris) 2016;45:155–64.
25. Delorme P, Goffinet F, Ancel PY, Foix-L’Hélias L, Langer B, Lebeaux C, et al. Cause of preterm birth as a prognostic factor for mortality. Obstet Gynecol 2016;127:40–8.
26. Mitra R, Reiter JP. A comparison of two methods of estimating propensity scores after multiple imputation. Stat Methods Med Res 2016;25:188–204.
27. Dagenais C, Lewis-Mikhael AM, Grabovac M, Mukerji A, McDonald SD. What is the safest mode of delivery for extremely preterm cephalic/non-cephalic twin pairs? A systematic review and meta-analyses. BMC Pregnancy Childbirth 2017;17:397.
28. Locatelli A, Vergani P, Ghidini A, Assi F, Bonardi C, Pezzullo JC, et al. Duration of labor and risk of cerebral white-matter damage in very preterm infants who are delivered with intrauterine infection. Am J Obstet Gynecol 2005;193:928–32.
29. Viegas OA, Ingemarsson I, Sim LP, Singh K, Cheng M, Ratnam SS, et al. Collaborative study on preterm breeches: vaginal delivery versus cesarean section. Asia Oceania J Obstet Gynaecol 1985;11:349–55.
30. Zlatnik FJ. The Iowa premature breech trial. Am J Perinatol 1993;10:60–3.
31. Schmitz T, Carnavalet Cde C, Azria E, Lopez E, Cabrol D, Goffinet F. Neonatal outcomes in twin pregnancy according to the planned mode of delivery. Obstet Gynecol 2008;111:695–703.
32. Schmitz T, Korb D, Battie C, Cordier AG, de Carne Carnavalet C, Chauleur C, et al. Neonatal morbidity associated with vaginal delivery of noncephalic second twins. Am J Obstet Gynecol 2018;218:449.e1–13.