Planned caesarean section for women with a twin pregnancy

Outcome for second twin compared with first twin

In a study of 1305 twin pairs at 1500 g or more birthweight delivered between 1988 and 1999 in Nova Scotia, the risk of adverse perinatal outcome (intrapartum fetal death, neonatal death, moderate-severe respiratory distress syndrome (RDS), asphyxia, trauma and complications of prematurity) was significantly increased for second-born compared with first-born twins (RR 2.1; 95% CI 1.4 to 3.1) (Persad 2001b). However, it should be noted that in the Term Breech Trial, short-term neonatal morbidity was not associated with any long-term impairment (Whyte 2004).

There is also evidence that the second twin is at greater risk of adverse perinatal outcome, compared to the first twin if delivery is vaginal but the same has not been shown if delivery is by CS. Arnold and colleagues undertook a matched case-control study of preterm twin pairs (Arnold 1987). The risk of RDS was increased for the second twin compared to the first if delivery was vaginal (odds ratio (OR) 14.2; 95% CI 2.5 to 81.1) but not if delivery was by CS (OR 0.90; 95% CI 0 to 17.8). This could be due to a greater protective effect of VB for the leading twin.

Outcomes for twins born vaginally versus by CS

Higher rates of adverse perinatal outcome have been reported for the twin pregnancy at or near term if birth is vaginal versus by CS (Barrett 2004). In the Kiely 1990 review, for twins in vertex presentation weighing more than 3000 g at birth, the perinatal mortality rate was 12.3/1000 versus 2.9/1000 (RR 4.22) if birth was vaginal versus by CS. However, comparisons of actual as opposed to planned method of delivery are subject to considerable bias. In a retrospective study of women with previous CS and current twin pregnancy, perinatal mortality was more common following trial of labour than repeat CS, but there was no difference in outcome after adjusting for confounding variables (Aaronson 2010).

There is increasing evidence for perinatal benefits related to vaginal birth. In a cross-sectional study of 6929 inborn infants without congenital anomalies, with gestational ages from 37 to 41 (6/7) weeks with vertex presentation, non-urgent caesarean delivery under regional anaesthesia compared to vaginal delivery under local or no anaesthesia increased the risk of bag and mask ventilation (OR 1.42, 95% CI 1.07 to 1.89) when adjusted for number of gestations, maternal hypertension and birthweight (De Almeida 2010). However, another retrospective study found no significant increase in transient tachypnoea of the newborn associated with increasing rates of elective CS for twin pregnancies over time (Suzuki 2010).

In a retrospective study of twin births at 37 or more weeks’ gestation, elective but not emergency CS was associated with an increased risk of transfusion in the neonate (Suzuki 2008).

Outcomes for twins delivered by planned VB (actual VB plus emergency CS) versus planned CS

A systematic review of non-randomised studies that compared the policies of planned VB and planned CS for the delivery of twins weighing at least 1500 g or reaching at least 32 weeks’ gestation (Hogle 2002) found four small studies that were eligible for inclusion in the review (Blickstein 2000; Grisaru 2000; Rabinovici 1987; Wells 1991). A meta-analysis of the data from the four studies did not find significant differences between the two approaches to delivery in terms of mortality or neonatal morbidity, although low Apgar score at five minutes was reduced with a policy of CS. This finding, however, was confined to twins in which the first twin presented as a breech. A subsequent retrospective comparison of planned vaginal birth versus planned CS for twin pregnancies with a leading breech found no significant differences in perinatal or maternal morbidity, other than an increase in thromboembolic disease in the planned CS group (Sentilhes 2007). Two large retrospective studies of women with previous CS and current twin pregnancy comparing planned vaginal birth with elective CS (Ford 2006; Varner 2005), found no difference in maternal or perinatal outcome, other than an incidence of uterine rupture of 0.9% in the planned vaginal birth group in one study (Ford 2006).

A comparison of outcomes in two Danish counties with high (57%) and low (28%) caesarean rates for twin pregnancy found no difference in perinatal outcomes (Henriksen 1994).

The previously published Cochrane review (Crowther 1996) found only one randomised controlled trial of planned CS versus planned VB for twins, in which 60 pairs of twins were enrolled (Rabinovici 1987). There were no perinatal deaths or cases of serious neonatal morbidity in either group. The sample size was too small to provide evidence of the better approach to delivery.

In low-resourced settings with limited facilities and personnel for CSs, high post-CS complication rates and uncertain access for the mother to CS facilities in future pregnancies, the benefit to the mother of avoiding CS is greater than in well-resourced settings.

Outcomes for second twin born by CS following vaginal birth of twin one, compared with CS birth of both twins

A prospective multicentre cohort study compared the outcome for second twins born by CS following vaginal birth of twin one, compared with CS for both twins. There was no difference in low Apgar scores, low cord blood pH nor neonatal encephalopathy. Neonatal infection was increased in the group with CS following vaginal birth of twin one, but this difference was not statistically significant after logistic regression analysis (Alexander 2008). In a large retrospective study of planned vaginal birth for twin pregnancies, CS for the second twin occurred in 9% of cases and was more common when the presentation of the second twin was nonvertex, and in pregnancies with other complications, and less common when the vaginal birth of the first twin was operative (Wen 2004a).

Importance of fetal presentation

If the first twin presents as breech, the current trend in well-resourced health systems, as for singleton breech presentation, is to recommend CS as being safer for the babies. This approach may be influenced by extrapolation of evidence from randomised trials of planned CS for singleton breech presentation (Hofmeyr 2003), though the interpretation of this evidence is the subject of considerable debate.The approach to the delivery of vertex/nonvertex twins is controversial (ACOG 2002; Barrett 2000; Crowther 1996). For twins presenting vertex/vertex, most clinicians recommend planned VB (ACOG 2002; Crowther 1996). However, planned CS may benefit twins in which the first twin is presenting vertex for a number of reasons. As many as 20% of vertex presenting second twins will change presentation spontaneously after the first twin is delivered (Houlihan 1996). A substantial number of those presenting vertex/vertex will present with serious acute intrapartum problems following the delivery of the first twin (for example, conversion to transverse lie, cord prolapse, prolonged interval to delivery of the second twin), which may lead to emergency CS, perinatal death, and neonatal morbidity. Lastly, if there are benefits to avoiding labour, both twins regardless of presentation should benefit.

Maternal outcomes

A policy of planned VB for women with a twin pregnancy in a hospital setting is associated with a 30% to 40% rate of emergency CS. Among those twins in which the first twin is born vaginally, there is still a risk of emergency CS for the birth of the second twin, ranging from 4% (Suzuki 2009) to 7% (Persad 2001a). In a retrospective case-control study, the risk of CS for the second twin after vaginal birth of the first was increased in association with a history of infertility therapy, gestational age equal to or greater than 39 weeks, non-vertex presentation, operative delivery of the first twin and intertwin delivery time interval greater than 30 minutes (Suzuki 2009). As the risk of maternal death is highest if delivery is by emergency CS, lowest following a VB, and intermediate following an elective CS (Hall 1999), increasing numbers of emergency CS may reduce the mortality benefits of planned VBs.

A relatively small case control study found no increased maternal morbidity for CS for twin compared with singleton pregnancy (Simoes 2007). However, in a large population-based, matched case-control study using the United Kingdom Obstetric Surveillance System, the risk of peripartum hysterectomy was increased for women with twin pregnancy (OR 6.30, 95% CI 1.73 to 23.0) (Knight 2008), while another study found twin pregnancy to be a risk factor for post CS wound sepsis (Schneid-Kofman 2005). There is growing evidence of an association between VB and urinary incontinence, particularly if the VB requires forceps or vacuum extraction (Farrell 2001; Meyer 1998; Wilson 1996). Urinary incontinence identified in the postpartum period has been shown to have long-lasting effects, with a high risk of urinary incontinence five years later (Viktrup 2001). Use of the supine position for birth has been cited as a possible contributor to incontinence following vaginal birth. Faecal incontinence and incontinence off latus have been reported to be associated with VB, particularly if forceps are used, and if there are lacerations involving the anal sphincter (Eason 2002; Zetterström 1999).

In the Term Breech Trial, there was no significant difference in maternal mortality or immediate serious maternal morbidity between planned CS and planned VB for singleton breech presentation at term (3.9% versus 3.2%, P = 0.35) (Hannah 2002). At three months postpartum, the women in the planned CS group reported a lower incidence of urinary incontinence (4.5% versus 7.3%, P = 0.02). Although the incidence of incontinence of flatus was not different between groups, if incontinence of flatus was reported, it was significantly less of a problem in the planned CS group (P = 0.006). At two years, the only significant difference found was less constipation in the planned vaginal birth group (Hofmeyr 2003). However, long-term risks of CS, particularly risks in subsequent pregnancies such as placenta praevia, placenta accreta, repeat CS and uterine rupture, as well as reduced fertility, have not been evaluated in randomised trials. Thus from a maternal perspective, the relative benefits and risks of planned CS versus planned VB are not clear-cut. Any evaluation of risks versus benefits must take into account the likelihood of serious long-term adverse effects of CS. Data from a large Canadian cohort indicated that even in women with previous CS, maternal mortality was significantly higher with CS than with VB after CS Wen 2004b.

Given that in well-resourced environments suicide is becoming an important contributor to maternal mortality, emotional consequences of birth such as depression need to be given importance when weighing up benefits and harms of different approaches to care. Even when emergency CS is performed, some women may gain self-esteem from having at least attempted to achieve VB. The sense of control from participation in the decision-making process may also promote self-esteem.

A more detailed account of evidence related to the benefits and risks of caesarean section is given in Lavender 2006.

For these reasons, a systematic review of the information from randomised trials is required to determine the relative benefits and risks of planned CS for twin pregnancy, for the infants and the mother.

Types of studies

All comparisons of intention to perform caesarean section and intention to birth vaginally, subject to a management protocol, for women with a twin pregnancy at or before term; random allocation to treatment and control groups, with adequate allocation concealment; violations of allocated management and exclusions after allocation not sufficient to materially affect outcomes.

Types of participants

Women with viable twin pregnancy considered suitable for vaginal birth, excluding women with known serious fetal anomaly.

Types of interventions

Planned caesarean section compared with planned vaginal birth subject to the requirements of a management protocol.

Types of outcome measures

Electronic searches

We contacted the Trials Search Co-ordinator to search the Cochrane Pregnancy and Childbirth Group’s Trials Register (30 September 2011).

The Cochrane Pregnancy and Childbirth Group’s Trials Register is maintained by the Trials Search Co-ordinator and contains trials identified from:

1. quarterly searches of the Cochrane Central Register of Controlled Trials (CENTRAL);

2. weekly searches of MEDLINE;

3. weekly searches of EMBASE;

4. handsearches of 30 journals and the proceedings of major conferences;

5. weekly current awareness alerts for a further 44 journals plus monthly BioMed Central email alerts.

Details of the search strategies for CENTRAL, MEDLINE and EMBASE, the list of handsearched journals and conference proceedings, and the list of journals reviewed via the current awareness service can be found in the ‘Specialized Register’ section within the editorial information about the Cochrane Pregnancy and Childbirth Group.

Trials identified through the searching activities described above are each assigned to a review topic (or topics). The Trials Search Co-ordinator searches the register for each review using the topic list rather than keywords.

We did not apply any language restrictions.

Selection of studies

Two review authors independently assessed for inclusion all the potential studies we identified as a result of the search strategy. We planned to resolve any disagreement through discussion or, if required, we would have consulted a third person.

Data extraction and management

We designed a form to extract data. For eligible studies, two review authors extracted the data using the agreed form. We would resolve discrepancies through discussion or, if required, we would consult a third person. We entered data into Review Manager software (RevMan 2011) and checked for accuracy.

When information regarding any of the above was unclear, we would attempt to contact authors of the original reports to provide further details.

Assessment of risk of bias in included studies

Two review authors independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We would resolve any disagreement by discussion or by involving a third assessor.

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

For included studies, we noted levels of attrition. We would explore the impact of including studies with high levels of missing data in the overall assessment of treatment effect by using sensitivity analysis.

For all outcomes, we carried out analyses, as far as possible, on an intention-to-treat basis, i.e. we attempted to include all participants randomised to each group in the analyses, and all participants were analysed in the group to which they were allocated, regardless of whether or not they received the allocated intervention. The denominator for each outcome in each trial was the number randomised minus any participants whose outcomes were known to be missing.

Assessment of heterogeneity

We would assess statistical heterogeneity in each meta-analysis using the T², I² and Chi² statistics. We would regard heterogeneity as substantial if I² was greater than 30% and either T² was greater than zero, or there was a low P value (less than 0.10) in the Chi² test for heterogeneity.

Assessment of reporting biases

If there had been 10 or more studies in the meta-analysis we would have investigated reporting biases (such as publication bias) using funnel plots. We would assess funnel plot asymmetry visually, and use formal tests for funnel plot asymmetry. For continuous outcomes we would use the test proposed by Egger 1997, and for dichotomous outcomes we would use the test proposed by Harbord 2006. If we detected asymmetry in any of these tests or by a visual assessment, we would perform exploratory analyses to investigate it.

Data synthesis

We carried out statistical analysis using the Review Manager software (RevMan 2011). We would use fixed-effect meta-analysis for combining data where it was reasonable to assume that studies were estimating the same underlying treatment effect: i.e. where trials were examining the same intervention, and we judged the trials’ populations and methods to be sufficiently similar. If there was clinical heterogeneity sufficient to expect that the underlying treatment effects differ between trials, or if substantial statistical heterogeneity was detected, we would use random-effects meta-analysis to produce an overall summary if an average treatment effect across trials was considered clinically meaningful. We would treat the random-effects summary as the average range of possible treatment effects and we would discuss the clinical implications of treatment effects differing between trials. If the average treatment effect was not clinically meaningful we would not combine trials. If we used random-effects analyses, we would present the results as the average treatment effect with 95% confidence intervals, and the estimates of T² and I².

Subgroup analysis and investigation of heterogeneity

If we identified substantial heterogeneity, we would investigate it using subgroup analyses and sensitivity analyses. We would consider whether an overall summary was meaningful, and if it was, use random-effects analysis to produce it.

We planned to carry out the following subgroup analyses.

1. Settings with high (more than 20 per 1000) versus low (less than 20 per 1000) perinatal mortality versus perinatal mortality mixed or unknown.

2. Leadin twin cephalic versus non-cephalic versus mixed presentation or unknown.

3. Leading twin cephalic and second twin cephalic versus non-cephalic versus mixed presentation or unknown.

4. Gestational age term versus preterm versus mixed or unknown gestational age.

We would use all outcomes in subgroup analysis.

For fixed-effect Mantel-Haenzel meta-analyses we would assess differences between subgroups by interaction tests. For random-effects and fixed-effect meta-analyses using methods other than Mantel-Haenzel, we would assess differences between subgroups by inspection of the subgroups’ confidence intervals; non-overlapping confidence intervals would indicate a statistically significant difference in treatment effect between the subgroups.

Sensitivity analysis

We would undertake sensitivity analyses to assess the effect of exclusion of studies with high risk of bias.

Included studies

We included one trial (Rabinovici 1987), involving 60 women with cephalic/non-cephalic twin pregnancies in labour at 35 or more weeks’ gestation who were allocated to vaginal delivery (33) or caesarean section (CS) (27). The trial was conducted in Israel. In the vaginal delivery group, two women were delivered by CS, and in four the second twin changed to cephalic presentation. These six women were excluded from the published data analysis. We have included the categorical neonatal outcomes in this review to conform to an intention-to-treat analysis. We were not able to include the continuous neonatal variables, and the maternal outcomes for the six excluded women were not reported, so we have not included these outcomes in this review. The reported maternal febrile morbidity (excluding the six women) was increased in the CS group (11/27 versus 3/27).

Allocation

The allocation procedure was described as being changed randomly by a non-involved person without prior notice on a time basis. A 20% difference in group sizes was not accounted for (27 versus 33). The possibility of inadequate allocation concealment is therefore high.

Blinding

Blinding not feasible and therefore the trial was rated as ‘high risk’ of bias. There was no mention about whether neonatal assessments were blinded.

Incomplete outcome data

The trial was assessed as ‘high risk’ for attrition bias. Six women allocated to planned vaginal birth were excluded from primary analysis for delivery (this was not in accordance with the protocol) (2 CS and 4 vertex vaginal births). Analysis was not conducted on an intention-to-treat basis.

Selective reporting

No pre-published protocol available to check predefined outcome reporting.

Other potential sources of bias

We assessed Rabinovici 1987 as being at ‘high risk’ for other bias because of a baseline imbalance: CS n = 27 versus vaginal n = 33.