Induction of labor doubled the rate of CD in our cohort. Nulliparity and induction of labor were identified as risk factors for CD. No risk factors were identified for emergency CD for the second twin. However, emergency CD for the second twin was associated with a lower Apgar score and pH for the affected twin.
Delivery in twin gestations
The Twin Birth Study [28], compared the risk for fetal or neonatal death or serious morbidity in twin pregnancies between 32 and 38 GW in either planned CD or planned vaginal delivery. Around 56% of women, with planned vaginal birth, delivered vaginally, while 40% had a CD for both twins and 4% experienced an emergency CD for the second twin [28]. These numbers are comparable to those in our cohort.
In a further large retrospective cohort study by de Castro et al. [29], 883 women with twin gestations underwent a trial of labor. The rate of vaginal delivery in this study was almost 87% [29]. One noticeable difference to our cohort is that their number of labor induction was 33%, while ours was 55%, this might be a reason why the numbers are difficult to compare between ours and their study. However, in line with our results, they found nulliparity to be a risk factor for CD in twin gestations [29].
Induction in twin gestations
A retrospective observational cohort study was conducted by Lopian et al. [20] to determine and compare the safety and efficacy of induction of labor in twins. They compared women with a twin gestation > 32 GW that either underwent induction of labor (N = 268) or went into labor spontaneously (N = 450). Primary outcome was the CD rate. They found a twofold increased risk of CD in the group with induction of labor, while the method of induction (prostaglandins vs. oxytocin) did not make a difference [20]. Like in our cohort, induction of labor was associated with a higher rate of CD.
Another cohort study, from two University Hospitals in Sweden, investigated the association between induction of labor and CD [19]. They compared twin pregnancies with induction of labor (N = 220) and spontaneous onset of labor (N = 242) in twin pregnancies. Again, they found a significant higher rate of CD in the induction group (21 vs. 12%) [19]. In their cohort 80% of women with induced labor delivered vaginally. These results clearly differ from ours since the rate of vaginal deliveries after induction was only 52%.
However, comparison of the indications for induction and for secondary CD is difficult and might add to some of this difference. For example, how many women wish to proceed with delivery by CD instead of continuing with induction is unknown. Additionally, 68% of the inductions in the Swedish cohort were started by amniotomy, a method that we do apply however not that often [19].
Two studies, including a total of 435 twin pregnancies, compared expectant management with induction of labor and both found no increased rate in CD [16, 30].
To conclude, data about induction and rate of CD in twin pregnancies remains somewhat controversial. Ultimately, this demands for valid indications for induction but also for CD. When counseling patients regarding the mode of delivery in twin pregnancies, risks of each method need to be evaluated and discussed.
Emergency CS for the second twin
A study of Aviram et al. [31] examined the incidence, risk factors and outcomes of CD for the second twin (combined delivery) [31]. Their overall rate of a vaginal delivery in the first twin, in women who planned a vaginal birth, was slightly higher than in our cohort (60.4% vs. 51%). In their cohort, in 7% the second twin was delivered via CD, which is comparable to our data. The most common reasons for CD in the second twin were malpresentation, fetal compromise and cord presentation / prolapse, which are similar to our findings, where fetal compromise and malpresentation were the most common reasons for CD in the second twin. They also reported a higher neonatal morbidity (5-minute Apgar score < 7, neonatal intensive care unit admission, abnormal level of consciousness and prolonged assisted ventilation) after CD for the second twin compared to vaginal delivery in both twins [31].
Intertwin interval, pH and Apgar score
After delivering the first twin, several problems (such as insufficient contractions, malpresentation, placental disruption or prolapse of the umbilical cord) can difficult the delivery of the second twin, and therefore expose the second twin to a higher risk of hypoxia. In a study from McGrail et al. [32], 144 twin deliveries were analyzed which showed that both umbilical cord arterial and venous pH declined for the second twin as the interval between their delivery increased (arterial pH 0–15 min: 7.25 ± 0.06, 16–30 min: 7.22 ± 0.07). This is in accordance with our findings. However, neither their nor our findings are clinically strong enough to influence or plea for a change of management. This is undermined by a retrospective study from Schneuber et al. which included 207 twin pairs. Their results did not show a negative impact on the second twin if the interval from delivering twin one to twin two exceeds 15 minutes [33].
Very strict intertwin intervals of 10 minutes as well as more ease intervals of 30 minutes have been looked at as well [34]. However, a clear definition of the ideal intertwin interval cannot be deduced from literature. Probably other factors, such as fetal wellbeing and course of delivery are more important than a precise time definition.
Taken together, vaginal delivery in twin births bears risks for the second twin. Moreover, women who choose to deliver vaginally need to be informed about the risks, especially for the second twin and about the possibility of an emergency CD for the second twin.
Limitations and strengths
One limitation is the retrospective cohort study design. Another is the high rate of exclusions due to missing informed consent. A strength of this study is that data of one single center for 20 years were analyzed including clear reasons for induction of labor in twins, with documented chorionicity in almost all cases.