Vaginal birth after two caesarean sections (VBAC-2)—a systematic review with meta-analysis of success rate and adverse outcomes of VBAC-2 versus VBAC-1 and repeat (third) caesarean sections


Dr S Tahseen, 20 Malthouse Green, Luton LU2 8SN, UK. Email


Background  Trial of vaginal birth after Caesarean (VBAC) is considered acceptable after one caesarean section (CS), however, women wishing to have trial after two CS are generally not allowed or counselled appropriately of efficacy and complications.

Objective  To perform a systematic review of literature on success rate of vaginal birth after two caesarean sections (VBAC-2) and associated adverse maternal and fetal outcomes; and compare with commonly accepted VBAC-1 and the alternative option of repeat third CS (RCS).

Search strategy  We searched MEDLINE, EMBASE, CINAHL, Cochrane Library, Current Controlled Trials, HMIC Database, Grey Literature Databases (SIGLE, Biomed Central), using search terms Caesarean section, caesarian, C*rean, C*rian, and MeSH headings ‘Vaginal birth after caesarean section’, combined with second search string two, twice, second, multiple.

Selection criteria  No randomised studies were available, case series or cohort studies were assessed for quality (STROBE), 20/23 available studies included.

Data collection and analysis  Two independent reviewers selected studies and abstracted and tabulated data and pooled estimates were obtained on success rate, uterine rupture and other adverse maternal and fetal outcomes. Meta-analyses were performed using RevMan-5 to compare VBAC-1 versus VBAC-2 and VBAC-2 versus RCS.

Main results  VBAC-2 success rate was 71.1%, uterine rupture rate 1.36%, hysterectomy rate 0.55%, blood transfusion 2.01%, neonatal unit admission rate 7.78% and perinatal asphyxial injury/death 0.09%. VBAC-2 versus VBAC-1 success rates were 4064/5666 (71.1%) versus 38 814/50 685 (76.5%) (P < 0.001); associated uterine rupture rate 1.59% versus 0.72% (P < 0.001) and hysterectomy rates were 0.56% versus 0.19% (P = 0.001) respectively. Comparing VBAC-2 versus RCS, the hysterectomy rates were 0.40% versus 0.63% (P = 0.63), transfusion 1.68% versus 1.67% (P = 0.86) and febrile morbidity 6.03% versus 6.39%, respectively (P = 0.27). Maternal morbidity of VBAC-2 was comparable to RCS. Neonatal morbidity data were too limited to draw valid conclusions, however, no significant differences were indicated in VBAC-2, VBAC-1 and RCS groups in NNU admission rates and asphyxial injury/neonatal death rates (Mantel–Haenszel).

Conclusions  Women requesting for a trial of vaginal delivery after two caesarean sections should be counselled appropriately considering available data of success rate 71.1%, uterine rupture rate 1.36% and of a comparative maternal morbidity with repeat CS option.


Caesarean section (CS) rates have risen worldwide. Performance of elective repeat caesarean is one of the main reasons for the rise in caesarean rates, together with fetal distress, dystocia and breech presentation.1,2 In UK, CS rate in women with a previous caesarean is 67% as compared to 24% in primigravid women according to the results of The National Sentinel Caesarean Section Audit.1 Multiple caesarean sections are associated with placental adherence to scar (placenta increta/praevia)3 which is a potential surgical challenge and a cause of maternal morbidity and mortality. A trial of labour after previous caesarean delivery has been accepted as a way to reduce the overall caesarean rate and also to allow women choice for mode of delivery. Many studies have supported the efficacy and safety of vaginal birth after caesarean (VBAC) after one caesarean section and reliable figures of success rate and complications are available for counselling women for VBAC after one caesarean section.4–6 While clinicians [supported by guidelines (Society of Obstetricians & Gynaecologists of Canada7 and American College of Obstetricians & Gynaecologists)8] generally recommend or offer a trial of vaginal birth after one caesarean section, a trial of labour is generally not offered after two CS.

Although outcomes of trial of vaginal delivery after two caesarean sections9–31 have been published over last two decades, the subject of VBAC-2 has not received its due consideration among obstetricians. Women requesting for such trials are generally not allowed or counselled appropriately and may receive conflicting advice, an issue likely to be of considerable importance for many women. Successful VBAC-2 may also reduce overall caesarean section rate and associated complications of multiple caesareans.


To assess the success rate and associated major complications of trial of vaginal birth after two caesarean sections by means of systematic review; and to provide the relevant figures for patient counselling for such trials.

Search strategy

The peer-reviewed protocol for this review was prepared a priori, detailing specific objectives, criteria for study selection and approach to assessing quality, outcomes and statistical methods. The article was prepared in accordance with the Meta-analysis of Observational Studies in Epidemiology (MOOSE) Statement.32 We used published deidentified data and thus the present study was exempt from Local Research & Ethics Committee (LREC) approval.

Searches were performed on the following electronic bibliographic databases; Medline (from 1966), Cumulative Index to Nursing and Allied Health Literature CINAHL (from 1982), The Cochrane Library (2008: Issue 3), Current Controlled Trials, HMIC database, National research register, Research Findings Electronic Register (ReFER). Additionally Grey Literature databases searched were SIGLE (from 1980) and Biomed Central. Targeted internet searching of key organisation websites included; National Institute for Clinical Excellence (NICE), Royal College of Obstetricians & Gynaecologist (RCOG), American College of Obstetrician & Gynaecologist (ACOG), Society of Obstetricians & Gynaecologists of Canada (SOGC), Cochrane Library Issue 3 2006 and National Electronic Library for Health (NeLH).

The search terms comprised first search string including; Caesarean section, caesarean, caesarian, C*rean and C*rian, combined with a second search string which included; two, twice, second, multiple. Searches were applied, in turn, restricting to key words, title and then abstract. This search strategy yielded a large number of non-relevant articles. Use of the MeSH heading ‘Vaginal birth after caesarean section’ was then applied combined with the second string search words (not restricted to title or abstract) which yielded relevant papers Relevance of articles was further determined by the titles and/or abstracts. There were no language restrictions on the searches. Electronic searching was supplemented by hand searching of the reference lists. We attempted to contact authors where additional information was needed.

Study selection

There were no controlled trials available on the subject. Publications were either case reports, small case series or major cohorts. Individual case reports, duplicate publications and publications commenting on other papers were excluded. In case of publications reflecting the same cohort, only the study with most up-dated, complete and relevant data was used. Four foreign language papers were identified, after translation two were confirmed to be case reports,10,33 the third was a duplicate publication19,20 which was included and the last21 study was considered but excluded due to poor methodology. We used the appraisal tools from STROBE34 to assess methodological quality of identified studies (study selection process and targeted searching for guidelines is shown in Figure 1). Characteristics of each study are summarised in Table 1 (20 studies). All but the three studies14,21,23 were deemed to be of reasonable quality by STROBE criteria to be included. Jamelle23 described experience of unplanned VBAC in unbooked women (ten cases), largely or entirely labouring unsupervised, presenting to a tertiary centre usually in advanced labour or already with complications. Similarly, Emembolu14 described women presenting in advanced labour with unplanned VBAC (139 cases), these scenarios do not reflect planned trial of labour and would introduce bias to the results. Guettier21 reported 17 women with two previous uterine scars, one of which may be a previous myomectomy/hysterotomy scar. Nine women (53%) delivered vaginally including two women fully dilated on admission and one home delivery. These three studies were not included in the analysis.

Figure 1.

 Study selection process for the systematic review of success rates and complications of trial of vaginal birth after two caesarean sections.

Table 1.   Details of studies included in the systematic review of success rate and complications of VBAC-2
Study referenceStudy populationMethodsLabour managementSuccess rateMaternal outcomeNeonatal outcome
  1. VBAC, vaginal birth after caesarean; RCS, repeat (third) caesarean section.

Macones25 (North America)
Cohort study, comparing VBAC-2 with VBAC-1 and RCS
1082 VBAC-2
2888 RCS and 12,535 VBAC-1
Previous classical CS excluded
27% of subjects with previous two CS had a trial and 73% repeat CSVBAC-2 vs VBAC-1 IOL 30% vs 29%
Syntocinon augmentation 34% vs 34%
74.6% VBAC-2 success vs 75.5% VBAC-1.
Women with prior vaginal delivery were more likely to undergo trial
Uterine rupture 1.8% vs 0.9%
Transfusion 0.92% vs 0.68% and 1.18% in RCS.
Fever 12.7% RCS vs 8.8% in VBAC-2
Birthweight 3347 vs 3349 g
Landon24 (North America)
Prospective cohort study, comparing VBAC-2 vs VBAC-1 and VBAC-2 vs RCS
975 cases VBAC-2 (including 84 cases with three previous CS and 20 with four previous CS) vs 16 915 VBAC1 and 6035 RCS after two CS Previous classical CS excluded14% of subjects with previous two CS had a trial, 86% RCSVBAC-2 vs VBAC-1 IOL 23% vs 26%
Syntocinon augmentation 25% vs 32%
Epidural 58% vs 71%
66% (648/975) success in group with multiple CS, 74% (12490/16915) in VBAC-1 group
Women with prior vaginal delivery were more likely to undergo trial
Uterine rupture 0.9% (9/975) in VBAC-2 vs 0.7% in VBAC-1 (115/16 915)
Hysterectomy 0.6% vs 0.2% respectively, transfusion 3.2% vs 1.6%
Maternal morbidity comparable with RCS
Term NICU admission 11% vs 9%
Term intrapartum stillbirth 0% vs 0.01%, term NND0.15% vs 0.08%, term HIE 0% vs 0.1%
Garg18 (Saudi Arabia)
Case series
100 cases in VBAC-2 trial vs 71 elective CS and 34 non-trial emergency CS. Only previous low transverse uterine scars included100 cases had trial (48% of women with previous two CS)No IOL or augmentation66% successNo uterine rupture or hysterectomy in either group, no data for morbidity providedNo difference in fetal outcome reported although no data were provided
Spaans31 (The Netherlands)
Cohort data from caesarean registry/delivery records
59 cases VBAC-2 vs 187 RCS
Data on previous scar type not available
26% of women with previous two CS had trialIOL 24%83% (49/59) had a vaginal birth
Similar success rate with history of vaginal delivery
Transfusion 4/59 (6.7%) vs 19/187 (10%) in labour uterine rupture 1/59 (1.7%), and hysterectomy 1/59 (1.7%), minor maternal morbidity comparable in VBAC and RCSApgar <7 at 5 min 7%
NICU admission
Bretelle11 (France)
Case series
96 cases were allowed a trial of labour, 84 had elective repeat CS
Only previous low transverse uterine scars included
VBAC-2 (52%) and elective CS (48%)Clinical/X-ray pelvimetry for allowing TOL. IOL12%, syntocinon augmentation 76%, pidural 76%Success 65.6%Uterine rupture 0%Blood transfusion 1%
Hysterectomy secondary to atony 1%, no morbidity data for RCS group available
72 infants had Apgar score >7 at first minute
 (North America)
 Caesarean registry data
134 cases VBAC-2 compared with 3757 VBAC-1
Previous classical scar excluded, others included
Comparative study, between VBAC-2 and VBAC-1 (proportion of women undergoing trial unknown)IOL 20% vs 22%
Syntocinon augmentation 38% vs 36%, epidural 72% vs 70%
Uterine rupture 3.7% vs 0.8% (prior vaginal delivery was protective of uterine rupture). Hysterectomy 1/134 (0.74%) vs 7/3757 (0.18%)Uterine rupture 3.7% vs 0.8% (prior vaginal delivery was protective of uterine rupture). Hysterectomy 1/134 (0.74%) vs 7/3757 (0.18%)Birthweight 3530 g vs 3506 g
 {5-min Apgar score <7.60%VBAC-2 vs 9.7%VBAC1 in uterine rupture group only}
Emembolu14 Nigeria
Case series and controls (? matched)
139 cases with previous two CS No information on previous type of scarProportion of women undergoing trial unknownWomen presented in labour (usually advanced) 2.2% vs 8.9% syntocinon augmentation33% (46/139) achieved vaginal delivery vs 62% in controlUterine rupture 1.4% vs 0.6%
Maternal death (not directly attributable) rate 7.2/1000 vs 5.9/1000, transfusion 35% vs 14%
Peri-natal death 12% vs 12%, low Apgar score <7 18% vs 17%
Jamelle23 (Pakistan)
Case series of 10
10 cases with previous low segment CS (9/10 unbooked, all unplanned vaginal delivery)
No information on previous scar
9/10 presented in advanced labour, one admitted postdeliveryApparent disproportion was excluded by vaginal examination. No syntocinon augmentationAll patients delivered vaginally1/10 scar rupture required laparotomy
1/10 septicaemia
One (1/10) stillbirth associated with scar rupture, one neonatal death caused by prematurity/septicaemia
Asakura9 (North America) Case series302 cases VBAC-2, 1110 VBAC-1
Previous low vertical and unknown uterine scars included
Compared VBAC-2 with VBAC-1
69% after two CS and 92% after 1CS had trial.
Unrestricted use of IOL, syntocinon augmentation and epidural, identical to those with an unscarred uterus.64% success VBAC-2, 77% in VBAC-1Uterine rupture 3/302 (1%) vs 5/1110 (0.45%), hysterectomy 0.33% vs 0%1 min Apgar <3 in failed VBAC-2 9.6%, failed VBAC-1 9.1%
Asphyxial injury caused by uterine rupture 0.33% vs 0.09%
 (Saudi Arabia)
 Case series
115 cases VBAC-2
1006 cases RCS
Only previous low transverse uterine scars included
Compared VBAC-2 (10%) and elective CS (90%).Only 115/230 requesting for a trial were permittedProstaglandin IOL 37/115 (32%)
Oxytocin augmentation 32/115(28%)
Success 103/115 (90%) Similar success rate with history of vaginal deliveryScar dehiscence (0.8% vs 0.7%) and hysterectomy (0.8% vs 1.4%) comparablePeri-natal mortality 2.6% vs 1%, deaths not considered directly related to mode of delivery
Granovsky-Grisaru20 (Israel)
Case series
26 cases VBAC-2
26 controls RCS
3/26 had three previous-CS
Only previous low transverse uterine scars included
Control group from elective CS, denominator data not givenOnly spontaneous labour
Syntocinon augmentation 54%
Epidural 80%
19/26 (73%) success
Ventouse delivery 4/26
No uterine scar rupture or dehiscence
Hospital stay average 3 days
Baby weight 2800–4600 g, 5-min Apgar scores = 7–9
No NICU admissions
 (North America)
Cohort data from caesarean registry/delivery records
1827 previous two CS (1586 previous two CS, 241 previous ≥3 CS), 10 880 previous one CS. Previous classical scars excluded but unknown scars included!Compared VBAC-2 with VBAC-1; 54% previous ≥2 CS had trial vs 80% previous 1CSNo information availableSuccess 75% previous two CS vs 83% previous 1CSUterine rupture 1.8% (vs 0.6%Rupture related peri-natal death .018% vs trials vs 0.05%
Guettier21 (France)
Case series
17/41 women with two previous uterine scars, ncluding myomectomy/hysterotomyA case series of 17Included one home delivery, 2 pt fully dilated on admission9/17 (53%) vaginal delivery, 3/9 actually had two CSNo relevant morbidityNo neonatal morbidity in 17 cases, but three major congenital abnormality
Flamm16 (North America)
Case series
5733 trial of labour after CS, including 245cases VBAC-2, previous classical scars excluded but unknown scars included!All women undergoing trial of labor (38% of previous CS) no separate denominator data for Pre two CS availableNo separate figures for VBAC-269% success (168/245) for VBAC-2, 75% for whole groupNo separate figures for VBAC-2No separate figuresf or VBAC-2
Hansell22 (North America)
Case series
35 cases VBAC-2
2/35 had previous three CS, 1/35 pre.4 CS. 135 cases non-trial CS
Low transverse scars
21% of women with ≥2 previous CS had trial, 135 cases without trial CSOnly spontaneous labour
No Oxytocin augmentation
Success 77% (27/35).
Higher success rate with history of vaginal delivery
0% uterine rupture and hysterectomy in trial group
Transfusion 1/35 (2.8%) vs 11/135 (8.1%) in without trial CS
5 min Apgar score <5, none in the two groups
Phelan28 (North America)
Case series
501 cases VBAC-2 and 587 elective CS
Previous unknown scars included
Compared VBAC-2 (46%) and elective CS (54%)Oxytocin used for induction (3.6%) augmentation 53%Success 69%.
Higher success rate with history of vaginal delivery
Uterine rupture 0% vs 0.2% on elective CS
Hysterectomy 0.2% vs 1.2% in elective CS
No perinatal deaths attributable to trial, 5 min Apgar <7 2.6% in VBAC-2 and 1.4% in Elective CS
Novas27 (North America)
Case series
36 cases having VBAC-2 (nine had previous three CS)
Previous unknown uterine scars included
Compared VBAC-2 36(52%) and non-trial CS, 33 (48%)Oxytocin 47%Success 80% (29/36)Uterine rupture 1/36–2.7%(H/O classical CSs revealed at laparotomy) vs 0/33
Hysterectomy 0/36 vs 2/33
Apgar scores comparable, no figures provided.
PND unrelated to trial
Pruett30 (North America)
Case series
55 cases VBAC-2, denominator data not available Previous unknown uterine scars includedSelected sample, pre2 CS in spontaneous labour
(indigent population, late booking), no comparator group
Only spontaneous labour
55% Oxytocin augmentation
45% success10/30 failed trial had endometritis, none in vaginal del.3/55 (5.4%) uterine rupture (two had previous vertical scar, one placenta accrete) and 2/55 (3.6%) hysterectomies. Transfusion 1/55 (1.8%)All Apgar scors >7 at 5 min
Farmakides15 (North America)
Case series
57 cases VBAC-2 (18 had three previous CS). Only low transverse uterine scars,
 denominator data not available
Selected sample, a report describing VBAC-2 spontaneous labour (? matched) controls (64) ERCSOnly spontaneous labour
Oxytocin augmentation 5/57 (8%)
Success 77% (44/57)No uterine ruptureNo significant peri-natal morbidity
Porreco29 (North America)
Case series
21 cases.Only low transverse uterine scars, denominator data not availableSelf-selected motivated patients. No comparator groupOnly spontaneous labour
33% Oxytocin,
19% epidural
81% successNo scar ruptureNo neonatal morbidity

Data collection and analysis

Data were abstracted independently by the two authors and any discrepancies were resolved by discussion. The following data items were collected if available from each paper; proportion of women undergoing trial (i.e. number of eligible women with previous two caesarean sections), success rate, uterine rupture rate, hysterectomy rate, blood transfusion, low Apgar scores, neonatal unit admission rate and perinatal asphyxial injury/death attributable to mode of delivery (judicious review of text to extrapolate this figure). The comparator groups [VBAC-1 or non-trial repeat (third) caesarean section] characteristics were noted if available.

Meta-analysis was performed following the guidelines proposed by the MOOSE Group.32 Interstudies heterogeneity, defined according to Higgins et al.35 as the percentage of total variation across studies because of heterogeneity rather than chance (I2), was tested with chi-square test for heterogeneity at a significance level of P = 0.10 and a random effect model was generated whenever the I2 statistics were >25% using Mantel–Haenszel analysis method. Categorical variables were examined with calculations of pooled odds ratios (ORs) with 95% confidence intervals (CI). Intergroup comparisons were considered statistically significant at an alpha level of two-tailed P < 0.05, if CIs did not encompass 1.0. Meta-analysis was performed with RevMan (Revision Manager, version 5 for Windows, The Nordic Cochrane Centre, The Cochrane Collaboration, Copenaghen, Denmark 2008).


After exclusion of case reports33,36 and review articles,37,38 26 studies were assessed further. Four publications reflected the same cohort25,39–41 and two publications each reflected same cohort17–20,24,42 the most comprehensive publications relevant to our subject were chosen from each cohort. The 20 studies considered in detail are summarised in Table 1. Three studies were excluded due to poor quality. Seventeen studies were included in data abstraction and analysis including 5666 subjects undergoing labour (mostly as planned trial of labour) after two or more caesarean sections. Six studies reported outcomes using a comparator group of VBAC-1 (50 685 subjects in VABC-1 versus 4565 in VBAC-2 group). Eight studies used a comparator group of repeat third caesarean sections (RCS) (nonlabour and/or elective), the subjects included in this subset were 2829 in VBAC-2 versus 10 897 in repeat (third) caesarean sections. In two recent large studies,24, 25 data were available for both the comparator groups. Comparison with VBAC-1 were carried out because of its wide acceptability among patients and clinicians and to compare the level of risk associated with VBAC-1 versus VBAC-2.

Success rates of trial of labour were available in all included studies. Uterine rupture rates were given in all studies except one;16 reported trial of scar data after single and multiple previous caesarean sections, and although success rates for single and multiple caesarean sections were specified, only a combined uterine rupture rate of 0.17% (10/5733) for the whole study population was given. Asymptomatic scar dehiscence found incidentally at caesarean section was disregarded and only symptomatic scar ruptures were included in analysis. Different maternal morbidity indicators were reported in studies. Febrile morbidity was reported mainly in comparison to caesarean sections in several studies, hospital stay was mentioned only in few.15,18,20 No maternal morbidity data except uterine rupture were available in some studies reporting on large cohorts from birth/caesarean registries, Miller26 reported 17 322 cases of trial of scar from a 10-year period and Flamm16 reported 5733 trial of scar cases over 5 year. Hysterectomy and blood transfusion rates were assessed, wherever available. Other maternal morbidities as operative injury or ITU admissions were variably classified24,25 and were too diverse for the purpose of pooled analysis.

Few studies have considered the neonatal outcomes, limiting the availability of neonatal data. Moreover, some studies reported neonatal morbidity only in cases of uterine rupture not the whole study population12,26 included this data would skew the adverse neonatal effects therefore only studies describing neonatal outcomes for all subjects were included (Table 2). No perinatal outcomes were reported by Porreco29 and Chattopadhyay13. Apgar scores considered in few studies were variably reported as 1-minute score or 1-, 5- and/or 10-minute scores; 5-min Apgar scores below 7 (reported by Spaan,31 Gravnovsky-Grisarau,20 Phelan28 and Pruett30), 1-minute Apgar score below 39 and 5-minute Apgar score below 5.22 Bretelle11 reported the neonatal morbidity of their study as ‘72/96 newborns had Apgar scores superior to 7 at first minute’, this statement was un-informative to assess neonatal outcomes in their study. We considered Apgar scores to be too diverse for aggregate comparison. Asphyxial injury/HIE/perinatal death and neonatal unit admission rates, where reported were pooled to assess neonatal outcome.

Table 2.   Outcome of VBAC-2 for all included studies
Study referenceNumbers VBAC-2VBAC successUterine RuptureTransfusionHysterectomyAsphyxial injury/PND**NNU
  1. Values in parenthesis are expressed in percentage.

  2. *2/3 uterine ruptures diagnosed by palpation after successful vaginal delivery.

  3. **Prelabour stillbirths and postnatal deaths unrelated to mode of delivery (e.g. prematurity related) were not included.

Macones251082807 (74.6)20 (1.8)10 (0.92)   
Landon24975648 (66)9 (0.9)31 (3.2)6 (0.6)1 (0.15)75 (11.2)
Garg1810066 (66)0 (0.0)00  
Spaan315949 (83)1 (1.8)4 (6.7)1 (1.7)0 (0.0)15 (34.9)
Bretelle119663 (65.6)0 (0.0)2 (2.1)1 (1.04)0 (0.0)0 (0.0)
Caughey1213483 (62)5 (3.7) 1 (0.75)  
Asakura9302194 (71)3 (1) 1 (0.33)1 (0.33) 
Chattopadhyay13115103 (89)1 (0.7) 1 (0.89)  
Granovsky-Grisarau202619 (73)0 (0.0)00 (0.0)0 (0.0)0 (0.0)
Miller2618271376 (75)32 (1.8)  1 (0.05) 
Flamm16245168 (69%)NA    
Hansell223527 (77)0 (0.0)1 (2.8)0 (0.0)  
Phelan28501346 (69)0 1 (0.2)  
Novas273629 (80)      (0.0) 0  
Pruett305525 (45)       3* (5.4)1 (1.8)2 (3.6)  
Farmakides155744 (77)0 (0.0) 0  
Porreco292117 (81)0 (0.0) 0  
Total56664064 (71.7)74/5421 (1.36)49/2428 (2.01)14/2512 (0.55)3/3285 (0.09)90/1156 (7.78)

The calculation of percentages were based only on the data from the papers reporting the relevant outcomes. If studies used different definitions of outcomes, a pooled analysis was not obtained. Table 2 shows the outcomes in all included studies. Table 3 shows comparison of VBAC-1 versus VBAC-2 reported in the same cohorts (six studies) and only the paired available outcomes are tabulated. Table 4 shows comparison of VBAC-2 with non-trial repeat (third) caesarean section within same cohorts (eight studies), again only the paired outcomes are listed.

Table 3.   Outcome of VBAC-2 versus VBAC-1: only outcomes with paired data available are included
Study referenceGroupNumberSuccessUterine ruptureTransfusionHysterectomyPND/Asphyxial injury*NNU
  1. Values in parenthesis are expressed in percentage.

  2. *Prelabour stillbirths and postnatal deaths unrelated to mode of delivery (e.g. prematurity related) were not included.

Macones25VBAC-21082807 (74.6)20 (1.8)10 (0.92)   
VBAC-112 5359464 (75.5)113 (0.9)85 (0.68)   
Landon24VBAC-2975648 (66)9 (0.9)31 (3.2)6 (0.6)1 (0.15)75 (11.2)
VBAC-116 91512 490 (74)115 (0.7)273 (1.6)35 (0.2)14 (0.09)1321 (9)
Caughey12VBAC-213483 (62)5 (3.7) 1 (0.75)0 (0.0) 
VBAC-137572818 (75)31 (0.8) 7 (0.19)1 (0.02) 
Asakura9VBAC-2302194 (71)3 (1.04) 1 (0.33)1 (0.33) 
VBAC-11110856 (64)5 (0.45) 0 (0.0)1 (0.09) 
Miller26VBAC-218271376 (75)32 (1.8)  1 (0.05) 
VBAC-110 8809063 (83)63 (0.6)  2 (0.018) 
Flamm16VBAC-2245168 (69)     
VBAC-154884123 (75)     
TotalVBAC-245653276 (71.7)69 (1.59)41 (1.99)8 (0.56)3 (0.09)75 (11.2)
VBAC-150 68538 814 (76.5)327 (0.72)358 (1.21)42 (0.19)17 (0.05)1321 (9)
Table 4.   Outcome of studies comparing VBAC-2 versus Repeat (third) Caesarean Section (RCS)
Study reference No. casesUterine ruptureTransfusionHysterectomyFeverPeri-natal deathNNU admission
  1. Values in parenthesis are expressed in percentage.

Macones25VBAC-2108219 (1.76)10 (0.92) 96 (8.8)  
CS28881 (0.03)34 (1.18) 366 (12.7)  
Landon24VBAC-29759 (0.9)31 (3.2)6 (0.6)30 (3.1)1 (0.1)75 (11.2)
CS60350 (0%)93 (1.5)27 (0.4)129 (2.1)1 (0.02)514 (9.1)
Spaan31VBAC-2591 (1.7)4 (6.7)1 (1.7)5 (8.4) 15 (35)
CS1872 (1.06)19 (10)0 (0)17 (9) 39 (23)
Chattopadhyay13VBAC-21151 (0.8)1 (0.8)1 (0.8)   
CS10067 (0.7)15 (1.4)15 (1.4)   
Granovsky-Grisarau20VBAC-2260 (0)0 (0)0 (0)2 (7.7)0 (0)0 (0)
CS260 (0)0 (0)0 (0)5 (19)0 (0)0 (0)
Hansell22VBAC-2350 (0)1 (2.8)0 (0)4 (11.4)  
CS1351 (0.7)11 (8)0 (0)39 (28)  
Phelan28VBAC-25010 (0) 1 (0.2)55 (11)  
CS5871 (0.17) 7 (1.2)74 (12)  
Novas27VBAC-2361 (2.7) 0 (0)   
CS330 (0) 2 (6)   
TotalVBAC-2282931 (1.09)47 (1.68)9 (0.40)192 (6.03)1 (0.09)90 (8.49)
CS10 89712 (0.11)172 (1.67)51 (0.63)630 (6.39)1 (0.01)553 (8.85)


Successful vaginal delivery was achieved in 4064/5666 (71.1%) as shown in Table 2, ranging in studies from 45% to 89%. The comparable rate in VBAC-1 group was higher 38 814/50 685 (76.5%—Table 4); meta-analysis showed a significant difference between the two groups with OR = 1.48 of higher success rate in VBAC-1 group versus VBAC-2, CI 1.23–1.78 (Figure 2, P < 0.0001, Z = 4.18).

Figure 2.

 Success rate of VBAC-2 versus VBAC-1.

Adverse maternal outcomes

Uterine rupture rate after VBAC-2 was reported in all studies except by Flamm16. The pooled uterine rupture rate of 16 studies was 1.36% (74/5421) (Table 2), ranging 0–5.4% within studies. Subgroup comparative analysis with VBAC-1 in five studies (Table 3), revealed rupture rates 0.72% in VBAC-1 versus 1.59% in VBAC -2; meta-analysis showed pooled OR = 0.42 of a uterine rupture in VBAC-1 group versus VBAC-2, CI 0.29–0.60 (Figure 3, P < 0.0001, Z = 4.65).

Figure 3.

 Uterine rupture rates in VABC-2 versus VBAC-1 groups.

A lower risk of uterine rupture with history of prior vaginal delivery was indicated; Macones25 in a large birth registry cohort reported a uterine rupture rate of 1.8% in VBAC-2 versus 0.9% in VBAC-1. Previous vaginal delivery appeared protective for uterine rupture as 0.5% compared to 2.4% rupture noted, respectively, with and without a previous vaginal delivery. Similarly Caughey12 reported subjects with previous vaginal delivery were one-fourth as likely to have a uterine rupture as those without.

The rate of hysterectomy was reported in eight studies, pooled average was 0.55% in VBAC-2 group, ranging within studies 0–3.6% (Table 2). Considering hysterectomy figures in the comparison of VBAC-2 and VBAC-1 (Table 3), the rates were 0.56% versus 0.19%; meta-analysis of three studies reporting hysterectomy rates comparing VBAC-1 versus VBAC-2, showed OR = 0.29 of hysterectomy in VBAC-1 versus VBAC-2 groups, CI 0.13–0.61 (P = 0.001, Z = 3.22, Figure 4). In the subset of data comparing VBAC-2 with repeat CS (Table 4), the hysterectomy rates were similar, 0.40% and 0.63%, respectively; meta-analysis of seven studies reporting the paired outcome showed OR = 0.75 of hysterectomy in VBAC-2 group versus RCS, CI 0.23–2.43 (Figure 5, P = 0.63, Z = 0.48). Rate of major haemorrhage was not specified in any of the papers but numbers needing blood transfusion were given in eight studies, average 2.01% range 0–6.7% (Table 2). Transfusion rates were lower in the VBAC-1 group 1.21% versus 1.99% in VBAC-2 (Table 3); meta-analysis showed OR = 0.56 of having a blood transfusion in VBAC-1 versus VBAC-2, CI 0.40–0.77 (Figure 6, P = 0.0004, Z = 3.52) and similar in repeat CS and VBAC-2 groups (1.67% and 1.68% respectively Table 4). Meta-analysis of six studies reporting paired transfusion rates showed OR = 0.94 of having a blood transfusion in VBAC-2 versus RCS, CI 0.45–1.96 (Figure 7, P = 0.86, Z = 0.17).

Figure 4.

 Hysterectomy rates in VABC-2 versus VBAC-1 groups.

Figure 5.

 Hysterectomy rates in VBAC-2 versus RCS.

Figure 6.

 Blood transfusion rates in VBAC-2 versus VBAC-1 groups.

Figure 7.

 Blood transfusion rates in VBAC-2 versus Repeat (third) CS (RCS) groups.

Febrile morbidity was reported particularly in comparison with repeat caesarean sections in six studies, 6.03% in VBAC-2 and 6.39% in RCS group Table 4. Meta-analysis of six studies reporting the paired outcome showed OR = 0.81 of febrile morbidity in VBAC-2 versus RCS, CI 0.55–1.18 (Figure 8, P = 0.27, Z = 1.11).

Figure 8.

 Febrile morbidity rates in VBAC-2 versus Repeat(third) CS (RCS) groups.

Non-specific reassuring statements regarding maternal morbidity were given in some studies as ‘no maternal complications occurred’,29‘no febrile morbidity noted’,30‘outcomes were similar to hospital’s general obstetric population’.27

Adverse neonatal outcomes

Neonatal Apgar scores are not analysed further due to variable reporting within studies. Asphyxial injury or perinatal death attributed to mode of delivery occurred in 0.09% (range 0–0.33%), Table 2. Neonatal unit admission rates only were reported in some studies as an index of neonatal morbidity rather than Apgar scores. Pooled NNU admission rate was 7.7% (range 0–34.9%), Table 2. The neonatal outcome of five studies comparing VBAC-1 and VBAC-2 had similar rates for neonatal asphyxial injury/perinatal death (attributable to mode of delivery) after VBAC-2 (0.09%) versus VBAC-1 (0.05%) (P = 0.35, Mantel–Haenszel). The NNU admission rates were also comparable (P = 0.89, Mantel–Haenszel). Subgroup analysis of VBAC-2 versus repeat (third) caesarean sections revealed rates of perinatal death/asphyxial injury, 0.09% with VBAC-2 versus 0.01% with repeat caesarean sections (P = 0.14, Mantel–Haenszel), although the difference does not reach statistical significance but may be clinically different. The NNU admission rates were similar as well (8.85 versus 8.49%P = 0.57, Mantel–Haenszel).

Non-specific reassuring statements regarding neonatal morbidity were given in some studies; Farmakides15‘there was no significant perinatal morbidity’; Novas27‘no significant differences were observed between the two groups in gestational age, Apgar scores and birthweights and no rupture related perinatal death’; Garg18‘no difference in fetal outcome’; no neonatal data were extracted from these studies.


This review shows that trial of vaginal delivery after two caesarean sections is associated with a reasonable success rate (71.7%), although lower than VBAC-1 (76.5%). The adverse maternal outcomes of a trial of vaginal delivery after two previous caesareans are comparable to repeat (third) caesarean sections, with similar hysterectomy (P = 0.63), blood transfusion (P = 0.86) and febrile morbidity (P = 0.27) rates. The adverse maternal outcomes rates of VBAC-2 are higher than VBAC-1, but the absolute rates are small. The neonatal outcome data are limited, however, the available data does not indicate a significant difference (assessed by neonatal death/asphyxial injury and Neonatal Unit admission rates) between VBAC-2, RCS or VBAC-1. Although our comparison was carried out with VBAC-1 group because of its wide acceptance, pragmatic and rational comparison of maternal morbidity of VBAC-2 is with RCS, as previous multiple operations would have a higher background risk43and the available alternative choice to women who already had two CS is a RCS.

The reluctance to offer a trial after two caesarean sections is likely to stem from concerns regarding scar rupture. Scar rupture is a rare event and individual studies are limited by size making uterine rupture risk a difficult outcome to assess, pooled data analysis provides more reliable figures. The rupture rate in the pooled analysis was 1.36% (Table 2). All included studies provided figures for scar rupture, slightly varying but clinically meaningful definitions of rupture have been used in different studies as ‘disruption of uterine muscle and peritoneum—or disruption of muscle and extension to bladder or broad ligament’ in Landon24 (2006, rupture rate 0.9%); ‘Signs and symptoms of intra-peritoneal bleeding—or disruption of uterine scar immediately preceded by non-reassuring fetal heart rate pattern (rupture rate 1.8%25); ‘….and full thickness dehiscence found at caesarean section performed for acute fetal distress (rupture rate 1.8%26). Earlier studies which contributed a heavy weight in the systematic review included patients with unknown uterine scars (with possibility of unknown proportion of lower vertical/classical scars) and reported higher uterine rupture rates (3.6%12; 2.7%27; 5.4%30; 1.8%26). A lower risk of scar rupture is indicated in the subgroup in Table 4 (with data from more recent years), 1.09% as compared to a rate of 1.59% in subgroup in Table 3 where earlier data (subjects26) contributed a higher proportion. Where manual scar exploration after a successful vaginal delivery was routinely performed and full thickness uterine wall defects repaired,26,30 a higher scar rupture rate was reported. Manual scar exploration is now rarely carried out and this together with a known lower segment uterine scar may be a reason for lower scar rupture rates in more recent studies.

Overall hysterectomy rate was 0.55%. A lower rate noted after VBAC-1 (0.19%), however, rate after RCS (0.63%, Table 4) was similar within subgroup analysis, which probably reflects the higher surgical risks associated with previous multiple caesarean sections including placenta accrete/praevia. Noteworthy, a lower threshold for hysterectomy in Pruett’s paper30 (3.6%) was because of women’s desire for tubal ligation, when scar defects were detected on manual scar palpation after successful vaginal delivery.

The blood transfusion rates were similar as with RCS (1.68% versus 1.67%), as well as febrile morbidity (6.03% versus 6.39%). Higher febrile morbidity after a failed trial of labour and in repeat caesarean sections (33%30, 28%22, 19.2%20) reported in earlier studies has significantly reduced after advent of broad spectrum antibiotics and concern regarding postoperative infectious morbidity is not a major issue in selecting mode of delivery.

The proportion of women undergoing a trial after two caesarean sections is variable between studies which may indicates variable selection criteria; from 9.2% by Landon24 to 69% by Asakura9. A more selective approach may be associated with higher success and/or lower uterine rupture rate,24 but a clear pattern does not emerge.

Neonatal morbidity with VBAC-2 assessed by neonatal unit admissions was comparable to the RCS group (Table 4). The more important measure of neonatal morbidity, hypoxic neonatal brain injury/death attributable to mode of delivery was reported in six studies (3285 subjects), the pooled rates were 0.09% in VBAC-2 as compared 0.01% in RCS group (P = 0.14) and 0.05% in VBAC-1 group (P = 0.34). The publications involved a time period of one or two decades ago with less-advanced neonatal facilities, hence the neonatal morbidity figures may not be representative of current practice given considerable advances in neonatal care in recent years, moreover, the available data does not indicate a significant difference in VBAC-2 and RCS.

It has been identified that women who have had a prior vaginal birth in addition to one prior caesarean delivery are more likely to have a successful VBAC attempt compared with women without a prior vaginal birth,41 and for the subset of women with prior vaginal birth as well as a caesarean section, a trial of labour as opposed to an elective repeat caesarean delivery is associated with a decreased rate of major maternal morbidities, postpartum fever and need for blood transfusions.41 In our analysis considering two prior caesarean sections, a similar trend of higher success and lower uterine rupture rates are indicated in women who have had a prior successful vaginal delivery; Macones25, Phelan28 and Hansell22 noted better prospects for successful VBAC-2 (as for VBAC-1) for women with a previous vaginal birth, interestingly Chattopadhay13 and Spaan31 found no advantage to this group; bearing in mind the latter two studies are much smaller as compared to the studies indicating favourable outcome with a prior vaginal delivery. Of note is the American College of Obstetricians and Gynecologist’s recommendation that for women with two prior caesarean deliveries, only those with a prior vaginal delivery should be considered candidates for a trial of labour.8 Moreover, Macones et al.25 concluded that as the majority of ruptures occurred in women whose labour was either induced or augmented and in those who had not had a previous vaginal delivery; it would seem reasonable to particularly target VBAC attempts in this lower risk group (i.e. those with a previous vaginal delivery) and avoidance of interventions as labour induction or augmentation to reduce the risk of rupture. However, a planned secondary analysis by the same group further commented40 that the associations (lack of prior vaginal delivery, induction and augmentation) do not lend themselves to prediction of scar rupture; hence clinicians will be justified in offering a trial to wider group. Landon24 reached the same conclusion that a requirement of prior vaginal delivery with multiple caesarean deliveries to be considered candidates for trial of labour seems unwarranted given the low risk of uterine rupture and adverse outcomes in this population.

Women who already have had two prior caesareans may choose to have a trial of vaginal birth or elect to have another caesarean section; such a choice may impact her future pregnancies as risk of abnormal placentation (placenta praevia/increta) increase with further surgery on uterus. A successful VBAC-2 attempt would improve prospects of future pregnancies in terms of future success and need for further caesareans.

Heterogeneous populations in included studies merit judicious interpretation of results. Ideally, the comparison should be within cases eligible to have a trial and then assignment to either elective CS or a trial of labour policy (prospective randomised controlled studies) to control the effect of confounding variables. However, the substantial body of accumulated evidence presented, especially large cohort studies12,24,25 aiming to report on all women with previous multiple CS receiving trial or otherwise, showing concordant results, provides more reliable evidence of efficacy and risks of trial of vaginal birth after two CS. In our view, to completely exclude such patients from a trial is an arbitrary decision, as although the relative risk of uterine rupture and of major complications is higher with a VBAC attempt than with repeat third caesarean delivery the absolute risk of such complications is small and overall maternal morbidity is comparable. Patients keen to have a trial should be counselled to pursue a carefully monitored trial of labour. The risks involved should be considered in perspective keeping in mind risks/morbidity associated with policy of repeat elective caesarean sections.


The included studies span over time period of more than two decades; in late 1980s and early 1990s in the US were times of high rates of trial of labour after caesarean. In the current decade, VBAC rate has fallen. Thus, it is not surprising that there may be differences seen in the studies by different time periods that may reflect not that the old studies had inappropriately high rates, but that the current studies represent biased populations in that only good/self motivated candidates are encouraged to consider a trial of labour after caesarean. Moreover, patients may have been managed differently in labour for varying amounts of time. Proportion of women having a trial after two CS is variable from 9% to 69% among studies which may indicate heterogeneous selection. Women having more than two CS were included in some studies, although the numbers were small, some studies included patients with unknown uterine scars (lower vertical/classical) which might have led to higher complications risk. Method of prior uterine incision has also not been described. The control groups, if chosen, were variable as well, some studies compared with VBAC-1 group and others with women having non-trial CS, which led to the sub-group analyses. There is possibility of over-representation of favourable results within published literature, an inherent limitation of meta-analysis of observational studies, however, in the absence of randomised trials on the subject, such analyses present reasonable available evidence of efficacy and risks.


This pooled data analysis indicates that a trial of labour by women with a history of previous two caesarean sections is associated with success rate of 71.1% with associated risk of scar rupture 1.36%. Success rate is higher for women with a prior history of vaginal delivery. More importantly, the maternal morbidity is comparable with RCS. Although neonatal morbidity information is limited, the available data does not indicate excessive risk. This information is important for women and healthcare providers who are making choices about the type of delivery. Women requesting to have a trial after two caesarean sections should be counselled appropriately considering available evidence; and should have the option of a carefully monitored vaginal delivery available to them.

Disclosure of interest

No financial disclosures or funding was involved in manuscript preparation.

Contribution to authorship

MG conceived the idea and conducted literature searches. Both MG and ST were involved in designing the study questions, reviewed studies for quality, abstracted, tabulated and analysed data as well as preparation of manuscript.

Details of ethical approval

Published de-identified data were used in preparation of this manuscript, hence no ethical approval was required.


No funding from any sources is involved in preparation of this manuscript.


None to declare.