Neonatal and Maternal Outcomes in Spontaneously-Conceived Twin Pregnancies According to Mode of Delivery

Introduction: Appropriate intrapartum conduct in a twin delivery remains a challenging aspect of obstetric practice. The objective of this study was to compare neonatal and maternal outcomes in twin pregnancies according to mode of delivery. Materials and methods: This is a single centre retrospective cohort study of all consecutive spontaneously-conceived twin deliveries (≥ 24 weeks, estimated fetal weight ≥ 500 grams) over a nine-year period between 01/01/2007 31/12/2016 at a tertiary-level centre. Neonatal outcomes included survival, APGAR score, prematurity-associated pathology (PAP), admission to the neonatal intensive care unit (NICU) and length of stay (LOS). Maternal outcomes included postpartum complications and LOS. Statistical analysis comprised Chi-square test with subsequent p-value and odds-ratio with 95% confidence interval. Statistical significance was set at p <0.05. Results: A total of 173 consecutive women with spontaneously-conceived twin deliveries were enrolled in this study, 129 (74.6%) women delivered by caesarean section (CS). The success rate of vaginal delivery (VD) was 93.6% (44/47). A strong statistical correlation was identified between CS and NICU admission; 53.2% vs. 1.5% (p=0.0001). Neonatal LOS in the NICU was significantly longer in the CS group. Prematurity-associated pathology (PAP) was noted in 75 pairs of twins (75/173); 61 pairs were delivered by CS, bearing strong statistical significance (p<0.0001). Postpartum complications occurred in 14.7% of CS compared to 13.6% of VDs. Conclusion: Neonates delivered by CS had a higher rate of PAP, NICU admission, lower birth weight and longer LOS. This study showed that VD is safe, especially when the first twin is in cephalic presentation.


INTRODUCTION
The prevalence of multiple pregnancies, especially twin pregnancies, continues to rise due to wider availability of in-vitro fertilisation (IVF). 1 In contrast, spontaneous conception of twins is a rarity compared to IVF conception, however, there is evidence that high consumption of red meat and a maternal history of twin gestation are associated with spontaneous conception of twins. 2 Conduct in a twin delivery continues to be one of the most challenging aspects of obstetric practice 3,4 and the optimal mode of delivery remains elusive. 5 There is an abundance of literature demonstrating that women carrying a twin pregnancy experience higher rates of complications during pregnancy and delivery compared to their counterparts carrying a singleton pregnancy. [6][7][8][9][10] This increase in complications is largely attributable to the higher incidence of premature delivery in twin pregnancies and its associated morbidity. 7 Prematurity-induced morbidity encompasses respiratory distress syndrome, transient tachypnea of the newborn, neonatal seizures, persistent fetal circulation, necrotising enterocolitis and intraventricular haemorrhage, even when birth weight exceeds 2500 grams. 11,12 Twin pregnancies often culminate in caesarean section, fuelling the on-going debate regarding the optimal mode of twin delivery. 9,13,14 The incidence of both short and long term maternal morbidity is increased following CS compared to vaginal delivery (VD). The main short-term complications include intra or postpartum haemorrhage and infection, while long-term complications of a scarred uterus comprise morbidly adherent placental disorders and uterine rupture. 4,[15][16][17] No randomised controlled trials were successful in identifying the optimal mode and timing of delivery in twin pregnancies until 2013. Since then, the Twin Birth Study (TBS) provided pivotal evidence that neonatal outcomes following VD were superior to CS when the leading twin was in vertex presentation, at a minimum 32 weeks gestation and an experienced obstetrician was present at delivery. 3 Although this study provided highly sought-after evidence, only a fraction of professional obstetrics societies worldwide have since included planned VD in twin pregnancy into their guidelines. 18,19 The TBS also concluded that planned CS is safer than VD when the leading twin is breech as it reduces the risk of severe neonatal morbidity. 3 The JUMODA study, a national prospective analysis of 5,915 women, concluded that composite neonatal morbidity and mortality was higher in the planned CS group. 20 Rossi et al. 's meta-analysis found that there was no difference in outcome of the second twin when both twins were in vertex presentation. 21 The objectives of this study were to compare neonatal and maternal outcomes in twin deliveries according to mode of delivery.

MATERIALS AND METHODS
Over a nine-year period, between 2007 -2016, the medical records of all spontaneously-conceived twin deliveries (24+0 -40+0 weeks gestation) at a single tertiary referral centre, were reviewed. Data comprising baseline maternal characteristics and maternal and neonatal outcomes were collected according to mode of delivery. Baseline characteristics encompassed demographics, obstetric history, gestational age, medical and pregnancy induced comorbidities, pregnancy-associated pathology and fetal characteristics (chorionicity, amnionicity, difference in estimated fetal weight (EFW). Neonatal outcomes were centred around neonatal survival, APGAR score, prematurity-associated pathology, birth weight, admission to the neonatal intensive care unit (NICU) and length of stay (LOS). On the other hand, maternal outcomes focused primarily on early postpartum complications, which occurred whilst in hospital and LOS. Primary outcomes were neonatal and maternal morbidity. Statistical analysis encompassed Chi-square test with subsequent p-value and odds ratio with 95% confidence interval, carried out using GraphPad Prism 6 ® (GraphPad Inc., USA). Statistical significance was set at p <0.05. All procedures were performed in compliance with relevant laws and institutional guidelines. Ethical approval of this study was waived since this analysis used existing records, based on information routinely collected, and subjects represented a de-identified data set.

RESULTS
Over the nine-year study period, a total of 279 women carrying a twin pregnancy were admitted. Of these 279 women, 106 were excluded; 41 conceived through in-vitro fertilisation, 10 experienced an abortion, while 55 delivered at another facility, yielding a final study population of 173 women, depicted in Fig. 1.
A total of 346 neonates were delivered; 334 live births, 11 cases of intrauterine fetal death (IUFD) and 8 neonatal deaths. One-hundred and twenty-nine (74.6%) women   Table 2 depicts the lack of association between mode of delivery and any specific week of gestational age (p>0.05).
Baseline characteristics are displayed in Table 3.
Fetal presentation at delivery is outlined in Table 4; both twins in vertex presentation was most frequent, accounting for 27.7% of all twins.
Seventy-five (43.3%) of the 173 twin pairs suffered from prematurity-associated pathology (PAP), of whom 61 (81.3%) were delivered by CS, demonstrating a strong statistical correlation between CS and PAP (OR: 18.98, 95% CI: 8.34-43. 16, p<0.0001). Despite the overall higher rate of PAP in neonates delivered by CS, only apnoea crises achieved statistical significance (p=0.03). Moreover, all types of PAP persisted longer in neonates delivered by CS, as depicted in Table 7.
A total of 25 postpartum complications occurred in 24

DISCUSSION
Appropriate intrapartum conduct in a twin delivery remains a challenging aspect of obstetric practice. 3,4,19 This cohort study found that composite neonatal and maternal outcomes were superior with VD. Prior to 2013, the tendency was to deliver twins by elective CS, mainly due to the high rate of malpresentation, featuring interlocked twins, as well as the risk of acute hypoxia in the second twin from decreased placental circulation and trauma to the premature fetal brain upon passage through the birth canal. 22 The success rate of planned VD ascertained in this study (93.6%) was higher compared to the rate reported in the Twin Birth Study (TBS): 56.2% 3 and Ylilehto et al. 's study of 495 twin deliveries: 81% 23 , however, it paralleled the success rate of Sadeh et al's study of 411 twin deliveries: 91% 24 .
Of the 33 pairs of twins with the leading twin in breech and the second twin in either breech or vertex presentation, only two pairs (6.1%) were delivered vaginally. This result was most likely influenced by TBS which found that elective CS decreased neonatal morbidity from 5% to 1.6% when the leading twin was in non-vertex presentation. 3 Of the eight cases of neonatal mortality, five (62.5%) occurred in neonates delivered by CS (p=0.42). This finding is consistent with the TBS's results 3 , however contradictory to Jhaveri et al. 's findings who reported three neonatal deaths of the second twin in premature twins delivered vaginally between 28-32 weeks gestation with extremely low birth weights. The proportion of low and critically-low APGAR scores in neonates delivered by CS was similar: 11.6% compared to 11.3% following VD. These findings coincide with those reported by Jhaveri et al. whose rate of neonatal morbidity in the planned vaginal group amounted to 23.8%, compared to 22.6% in the CS group. 14 Birth weight was significantly lower among neonates delivered by CS (p=0.0017). A likely explanation for this phenomenon is the rationale of fetal protection against intracerebral haemorrhage during passage through the birth canal. 25 The TBS found that twins weighing ≥ 1500 g or above 32 weeks gestation delivered by planned VD did not exhibit an increased rate of perinatal morbidity compared to their counterparts delivered by CS. Zamarfand et al. elaborate that the impact of gestational age on composite perinatal outcome at gestations < 33 weeks as prematurity-induced morbidity may mask the effect of delivery mode compared to a term delivery. 4 The authors also discuss the protective effect that VD exerts on twins delivered at earlier gestations than 33 weeks, resulting in a lower need for mechanical ventilation. 4 A strong statistical correlation was identified between CS and neonatal intensive care unit (NICU) admission: 53.2% vs. 1.5% (p=0.0001). Moreover, neonatal LOS in the NICU was significantly longer in the CS group; range: 1-68 days, average: 22.78 days compared to the VD group: 1-30, 12.05 days). Prematurity-associated pathology (PAP) was noted in 75 pairs of twins (43.4%); 61 pairs (81.3%) were delivered by CS, bearing strong statistical significance (p<0.0001).
Postpartum complications occurred in 14.7% of CS compared to 13.6% of VDs. The rate obtained in this study was higher than that of the TBS (7.3% vs. 8.5%). Maternal LOS was longer in women who delivered by CS; range: 4-83 days, average: 15.65 compared to 3-83, 12.13 following VD. The LOS reported in the current study is significantly longer compared to published rates as parturients are offered the choice of remaining in hospital during their twins' NICU stay.
There is a clear rise in rates of elective CS for twins worldwide despite the recommendations of the TBS. 3,14 The risks and adverse outcomes associated with CS, mainly complications and extended hospital stay, should be carefully weighed up in the case of planned CS not indicated due to fetal distress. In the short-term, women who deliver via CS have a higher incidence of haemorrhage and infection, compared to morbidly adherent placenta and uterine rupture with a scarred uterus in the long term. 4,[15][16][17] Moreover, Reitter et al. are concerned that the rising trend of CS delivery of low-risk twins pregnancies worldwide will minimise obstetricians' exposure and skills in a twin VD, especially when there is no evident benefit to the mother or fetuses. 18 With regard to long-term paediatric outcomes in twins, Fox et al. published a prospective observational study comprising a cohort of 354 twins, aged 6, which showed that there was no difference in long-term paediatric outcomes between VD and CS delivery. 26 Management of delivery of the second twin remains a considerable challenge in obstetric practice. The inter-twin interval has a higher rate of complications including decreased placental perfusion, cord prolapse, and more rarely, placental abruption. 22,27 Benito et al. concluded that second twins delivered after an inter-twin interval exceeding 10 minutes had poorer APGAR scores and higher rates of blood pH <7.15. 28 Current guidance advises that the inter-twin interval should not exceed 30 minutes as this causes poor perinatal outcomes secondary to acute hypoxia due to decreased placental perfusion. 27,29 However, rare cases of extremely delayed inter-delivery intervals up to 131 days have been reported in the interest of delaying extremely premature deliveries. 3 This study featured a case of intentional delayed delivery of the second twin 50 days after the leading twin. The patient was a gravida 4, para 2 (with a previous VD) whose delivery of dichorionic diamniotic twins was complicated by preterm premature rupture of membranes at 27 weeks. The leading twin was delivered in breech presentation, weighing 920 g, receiving an APGAR score of 6. The delivery of the second twin was delayed by 50 days, until 33 weeks, during which time a CS was performed due to redistribution phenomena. The second twin weighed 1870 g and received an APGAR score of 8.
The limitations of this study comprised a retrospective design and a small sample size due to a high number of women who delivered at another facility as well as an unequal proportion of the modes of delivery. Hence, further larger prospective multicentric studies are needed to conclude whether VD is safer than CS when the leading twin is in vertex presentation. Another limitation comprised the lack of data on neonatal blood pH as this is not a departmental practice for neonates with satisfactory APGAR scores.

CONCLUSION
The success rate of VD was 93.6%. Neonates delivered by CS had a higher rate of NICU admission and longer length of stay, prematurity induced morbidity and lower birth weight. There were no cases of intrapartum complications or intrapartum fetal deaths. To conclude, this study showed that VD is safe, especially when the leading twin is in vertex presentation. As such, women with a twin pregnancy > 32 weeks gestation should be counselled that a planned VD where the leading twin is in vertex presentation is safe and yields favourable perinatal outcomes.