1. DOI: 10.1111/j.1471-0528.2009.02351.x
Systematic review
www.bjog.org
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
S Tahseen,a M Griffithsb
a
Leeds University Hospitals NHS Trust, Leeds, UK b Luton & Dunstable Hospital NHS Foundation Trust, Luton UK
Correspondence: Dr S Tahseen, 20 Malthouse Green, Luton LU2 8SN, UK. Email stjavaid@yahoo.co.uk
Accepted 19 July 2009. Published Online 14 September 2009.
Background Trial of vaginal birth after Caesarean (VBAC) is RevMan-5 to compare VBAC-1 versus VBAC-2 and VBAC-2
considered acceptable after one caesarean section (CS), however, versus RCS.
women wishing to have trial after two CS are generally not
Main results VBAC-2 success rate was 71.1%, uterine rupture rate
allowed or counselled appropriately of efficacy and
1.36%, hysterectomy rate 0.55%, blood transfusion 2.01%, neonatal
complications.
unit admission rate 7.78% and perinatal asphyxial injury/death
Objective To perform a systematic review of literature on success 0.09%. VBAC-2 versus VBAC-1 success rates were 4064/5666
rate of vaginal birth after two caesarean sections (VBAC-2) and (71.1%) versus 38 814/50 685 (76.5%) (P < 0.001); associated
associated adverse maternal and fetal outcomes; and compare with uterine rupture rate 1.59% versus 0.72% (P < 0.001) and
commonly accepted VBAC-1 and the alternative option of repeat hysterectomy rates were 0.56% versus 0.19% (P = 0.001)
third CS (RCS). respectively. Comparing VBAC-2 versus RCS, the hysterectomy
rates were 0.40% versus 0.63% (P = 0.63), transfusion 1.68% versus
Search strategy We searched MEDLINE, EMBASE, CINAHL,
1.67% (P = 0.86) and febrile morbidity 6.03% versus 6.39%,
Cochrane Library, Current Controlled Trials, HMIC Database,
respectively (P = 0.27). Maternal morbidity of VBAC-2 was
Grey Literature Databases (SIGLE, Biomed Central), using search
comparable to RCS. Neonatal morbidity data were too limited to
terms Caesarean section, caesarian, C*rean, C*rian, and MeSH
draw valid conclusions, however, no significant differences were
headings ‘Vaginal birth after caesarean section’, combined with
indicated in VBAC-2, VBAC-1 and RCS groups in NNU admission
second search string two, twice, second, multiple.
rates and asphyxial injury/neonatal death rates (Mantel–Haenszel).
Selection criteria No randomised studies were available, case
Conclusions Women requesting for a trial of vaginal delivery after
series or cohort studies were assessed for quality (STROBE), 20/23
two caesarean sections should be counselled appropriately
available studies included.
considering available data of success rate 71.1%, uterine rupture
Data collection and analysis Two independent reviewers selected rate 1.36% and of a comparative maternal morbidity with repeat
studies and abstracted and tabulated data and pooled estimates CS option.
were obtained on success rate, uterine rupture and other adverse
Keywords Complications, uterine rupture, vaginal birth after two
maternal and fetal outcomes. Meta-analyses were performed using
caesarean sections.
Please cite this paper as: Tahseen S, Griffiths M. 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. BJOG 2010;117:5–19.
reasons for the rise in caesarean rates, together with fetal
Background
distress, dystocia and breech presentation.1,2 In UK, CS rate
Caesarean section (CS) rates have risen worldwide. Perfor- in women with a previous caesarean is 67% as compared
mance of elective repeat caesarean is one of the main to 24% in primigravid women according to the results of
ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology 5

2. Tahseen, Griffiths
The National Sentinel Caesarean Section Audit.1 Multiple tute for Clinical Excellence (NICE), Royal College of
caesarean sections are associated with placental adherence Obstetricians & Gynaecologist (RCOG), American College
to scar (placenta increta/praevia)3 which is a potential sur- of Obstetrician & Gynaecologist (ACOG), Society of Obste-
gical challenge and a cause of maternal morbidity and mor- tricians & Gynaecologists of Canada (SOGC), Cochrane
tality. A trial of labour after previous caesarean delivery has Library Issue 3 2006 and National Electronic Library for
been accepted as a way to reduce the overall caesarean rate Health (NeLH).
and also to allow women choice for mode of delivery. The search terms comprised first search string including;
Many studies have supported the efficacy and safety of vag- Caesarean section, caesarean, caesarian, C*rean and C*rian,
inal birth after caesarean (VBAC) after one caesarean sec- combined with a second search string which included; two,
tion and reliable figures of success rate and complications twice, second, multiple. Searches were applied, in turn,
are available for counselling women for VBAC after one restricting to key words, title and then abstract. This search
caesarean section.4–6 While clinicians [supported by guide- strategy yielded a large number of non-relevant articles.
lines (Society of Obstetricians & Gynaecologists of Canada7 Use of the MeSH heading ‘Vaginal birth after caesarean
and American College of Obstetricians & Gynaecologists)8] section’ was then applied combined with the second string
generally recommend or offer a trial of vaginal birth after search words (not restricted to title or abstract) which
one caesarean section, a trial of labour is generally not yielded relevant papers Relevance of articles was further
offered after two CS. determined by the titles and/or abstracts. There were no
Although outcomes of trial of vaginal delivery after two language restrictions on the searches. Electronic searching
caesarean sections9–31 have been published over last two was supplemented by hand searching of the reference lists.
decades, the subject of VBAC-2 has not received its due We attempted to contact authors where additional infor-
consideration among obstetricians. Women requesting for mation was needed.
such trials are generally not allowed or counselled appro-
priately and may receive conflicting advice, an issue likely
Study selection
to be of considerable importance for many women. Suc-
cessful VBAC-2 may also reduce overall caesarean section There were no controlled trials available on the subject. Pub-
rate and associated complications of multiple caesareans. lications were either case reports, small case series or major
cohorts. Individual case reports, duplicate publications and
publications commenting on other papers were excluded. In
Objectives
case of publications reflecting the same cohort, only the
To assess the success rate and associated major complica- study with most up-dated, complete and relevant data was
tions of trial of vaginal birth after two caesarean sections used. Four foreign language papers were identified, after
by means of systematic review; and to provide the relevant translation two were confirmed to be case reports,10,33 the
figures for patient counselling for such trials. 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
Search strategy
STROBE34 to assess methodological quality of identified
The peer-reviewed protocol for this review was prepared studies (study selection process and targeted searching for
a priori, detailing specific objectives, criteria for study selec- guidelines is shown in Figure 1). Characteristics of each
tion and approach to assessing quality, outcomes and sta- study are summarised in Table 1 (20 studies). All but the
tistical methods. The article was prepared in accordance three studies14,21,23 were deemed to be of reasonable quality
with the Meta-analysis of Observational Studies in Epide- by STROBE criteria to be included. Jamelle23 described expe-
miology (MOOSE) Statement.32 We used published deiden- rience of unplanned VBAC in unbooked women (ten cases),
tified data and thus the present study was exempt from largely or entirely labouring unsupervised, presenting to a
Local Research & Ethics Committee (LREC) approval. tertiary centre usually in advanced labour or already with
Searches were performed on the following electronic bib- complications. Similarly, Emembolu14 described women
liographic databases; Medline (from 1966), Cumulative presenting in advanced labour with unplanned VBAC (139
Index to Nursing and Allied Health Literature CINAHL cases), these scenarios do not reflect planned trial of labour
(from 1982), The Cochrane Library (2008: Issue 3), Cur- and would introduce bias to the results. Guettier21 reported
rent Controlled Trials, HMIC database, National research 17 women with two previous uterine scars, one of which
register, Research Findings Electronic Register (ReFER). may be a previous myomectomy/hysterotomy scar. Nine
Additionally Grey Literature databases searched were SIGLE women (53%) delivered vaginally including two women
(from 1980) and Biomed Central. Targeted internet search- fully dilated on admission and one home delivery. These
ing of key organisation websites included; National Insti- three studies were not included in the analysis.
6 ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology

3. Success rate & adverse outcomes of vaginal birth after two caesarean sections
Study selection flow chart
273 potentially relevant articles
identified and screened Targeted searching for
guidelines
-Society of Obstetricians &
Gynecologists of Canada
-American College of
243 articles excluded on
Obstetricians &
basis of title or abstract
Gynecologists
-Royal College of
Obstetricians &
Gynecologists-UK
26 studies, 3 case reports and -NICE (National Institute
one review article were of Clinical Excellence) UK
potentially relevant -Cochrane Library
6 studies reflected same
cohort and 3 case reports
and one review article were
not considered further
20 studies were appraised for
quality, 3 studies excluded
due to poor methodology, 17
studies included
Figure 1. Study selection process for the systematic review of success rates and complications of trial of vaginal birth after two caesarean sections.
not encompass 1.0. Meta-analysis was performed with
Data collection and analysis
RevMan (Revision Manager, version 5 for Windows, The
Data were abstracted independently by the two authors and Nordic Cochrane Centre, The Cochrane Collaboration,
any discrepancies were resolved by discussion. The follow- Copenaghen, Denmark 2008).
ing data items were collected if available from each paper;
proportion of women undergoing trial (i.e. number of eli-
Results
gible women with previous two caesarean sections), success
rate, uterine rupture rate, hysterectomy rate, blood transfu- After exclusion of case reports33,36 and review articles,37,38
sion, low Apgar scores, neonatal unit admission rate and 26 studies were assessed further. Four publications reflected
perinatal asphyxial injury/death attributable to mode of the same cohort25,39–41 and two publications each reflected
delivery (judicious review of text to extrapolate this figure). same cohort17–20,24,42 the most comprehensive publications
The comparator groups [VBAC-1 or non-trial repeat relevant to our subject were chosen from each cohort. The
(third) caesarean section] characteristics were noted if 20 studies considered in detail are summarised in Table 1.
available. Three studies were excluded due to poor quality. Seventeen
Meta-analysis was performed following the guidelines studies were included in data abstraction and analysis
proposed by the MOOSE Group.32 Interstudies heterogene- including 5666 subjects undergoing labour (mostly as
ity, defined according to Higgins et al.35 as the percentage planned trial of labour) after two or more caesarean sec-
of total variation across studies because of heterogeneity tions. Six studies reported outcomes using a comparator
rather than chance (I2), was tested with chi-square test for group of VBAC-1 (50 685 subjects in VABC-1 versus 4565
heterogeneity at a significance level of P = 0.10 and a ran- in VBAC-2 group). Eight studies used a comparator group
dom effect model was generated whenever the I2 statistics of repeat third caesarean sections (RCS) (nonlabour and/or
were >25% using Mantel–Haenszel analysis method. Cate- elective), the subjects included in this subset were 2829 in
gorical variables were examined with calculations of pooled VBAC-2 versus 10 897 in repeat (third) caesarean sections.
odds ratios (ORs) with 95% confidence intervals (CI). In- In two recent large studies,24, 25 data were available for
tergroup comparisons were considered statistically signifi- both the comparator groups. Comparison with VBAC-1
cant at an alpha level of two-tailed P < 0.05, if CIs did were carried out because of its wide acceptability among
ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology 7

4. 8
Table 1. Details of studies included in the systematic review of success rate and complications of VBAC-2
Study Study Methods Labour Success rate Maternal Neonatal
reference population management outcome outcome
Macones25 1082 VBAC-2 27% of subjects with VBAC-2 vs VBAC-1 74.6% VBAC-2 Uterine rupture Birthweight 3347
Tahseen, Griffiths
(North America) 2888 RCS and previous two CS IOL 30% vs 29% success vs 1.8% vs 0.9% vs 3349 g
Cohort study, 12,535 VBAC-1 had a trial Syntocinon 75.5% VBAC-1. Transfusion 0.92%
comparing Previous classical and 73% repeat CS augmentation Women with vs 0.68% and 1.18%
VBAC-2 with VBAC-1 CS excluded 34% vs 34% prior vaginal in RCS.
and RCS delivery were Fever 12.7% RCS vs
more likely 8.8% in VBAC-2
to undergo trial
Landon24 975 cases VBAC-2 14% of subjects VBAC-2 vs VBAC-1 66% (648/975) Uterine rupture 0.9% Term NICU admission
(North America) (including with previous IOL 23% vs 26% success in (9/975) in VBAC-2 11% vs 9%
Prospective cohort study, 84 cases with two CS had a Syntocinon group with multiple CS, vs 0.7% in VBAC-1 Term intrapartum stillbirth
comparing VBAC-2 three previous trial, 86% RCS augmentation 74% (12490/16915) (115/16 915) 0% vs 0.01%, term NND
vs VBAC-1 CS and 20 with 25% vs 32% in VBAC-1 group Hysterectomy 0.6% 0.15% vs 0.08%,
and VBAC-2 vs RCS four previous Epidural 58% vs 71% Women with vs 0.2% respectively, term HIE 0% vs 0.1%
CS) vs 16 915 VBAC1 and prior vaginal transfusion 3.2% vs 1.6%
6035 RCS after delivery were Maternal morbidity
two CS Previous more likely comparable with RCS
classical CS excluded to undergo trial
Garg18 (Saudi Arabia) 100 cases in VBAC-2 100 cases had No IOL or 66% success No uterine rupture or No difference in fetal
Case series trial vs 71 trial (48% augmentation hysterectomy in either outcome reported
elective CS and 34 non-trial of women with group, no data for although no data
emergency CS. previous two CS) morbidity provided were provided
Only previous
low transverse uterine
scars included
Spaans31 59 cases VBAC-2 26% of women with IOL 24% 83% (49/59) had a Transfusion 4/59 (6.7%) Apgar <7 at 5 min 7%
(The Netherlands) vs 187 RCS previous two vaginal birth vs 19/187 (10%) in NICU admission
Cohort data Data on previous scar CS had trial Similar success rate with labour uterine rupture 34%
from caesarean type not available history of vaginal delivery 1/59 (1.7%), and
registry/delivery hysterectomy 1/59
records (1.7%), minor maternal
morbidity comparable
in VBAC and RCS
Bretelle11 (France) 96 cases were VBAC-2 (52%) and Clinical/X-ray pelvimetry Success 65.6% Uterine rupture 0% 72 infants had Apgar
Case series allowed a trial elective CS (48%) for allowing TOL. IOL12%, Blood transfusion 1% score >7 at
of labour, 84 had syntocinon augmentation Hysterectomy first minute
elective repeat CS 76%, pidural 76% secondary
Only previous to atony 1%, no
low transverse morbidity data for
uterine scars included RCS group available
ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology

5. Table 1. (Continued)
Study Study Methods Labour Success rate Maternal Neonatal
reference population management outcome outcome
Caughey12 134 cases VBAC-2 Comparative study, IOL 20% vs 22% Uterine rupture 3.7% vs Uterine rupture 3.7% Birthweight 3530 g
(North America) compared between VBAC-2 Syntocinon augmentation 0.8% (prior vs 0.8% (prior vaginal vs 3506 g
Caesarean registry data with 3757 VBAC-1 and VBAC-1 (proportion of 38% vs 36%, vaginal delivery delivery was protective {5-min Apgar score
Previous classical women undergoing epidural 72% vs 70% was protective of uterine rupture). <7.60%VBAC-2
scar excluded, trial unknown) of uterine Hysterectomy 1/134 vs 9.7%VBAC1 in
others included rupture). Hysterectomy (0.74%) vs 7/3757 uterine rupture
1/134 (0.74%) vs (0.18%) group only}
7/3757 (0.18%)
Emembolu14 Nigeria 139 cases with Proportion of women Women presented in 33% (46/139) achieved Uterine rupture Peri-natal death 12%
Case series and previous two undergoing labour (usually advanced) vaginal delivery vs 1.4% vs 0.6% vs 12%, low Apgar
controls (? matched) CS No information trial unknown 2.2% vs 8.9% syntocinon 62% in control Maternal death (not score <7 18% vs 17%
on previous type of scar augmentation directly attributable) rate
7.2/1000 vs 5.9/1000,
transfusion 35% vs 14%
Jamelle23 (Pakistan) 10 cases with previous 9/10 presented Apparent disproportion All patients 1/10 scar rupture One (1/10) stillbirth
Case series of 10 low segment in advanced was excluded by delivered vaginally required laparotomy associated with scar
CS (9/10 unbooked, all labour, one vaginal examination. 1/10 septicaemia rupture, one neonatal
unplanned vaginal delivery) admitted postdelivery No syntocinon death caused by
No information on augmentation prematurity/septicaemia
previous scar
Asakura9 302 cases VBAC-2, Compared VBAC-2 Unrestricted use of 64% success Uterine rupture 1 min Apgar <3 in
(North America) 1110 VBAC-1 with VBAC-1 IOL, syntocinon VBAC-2, 77% 3/302 (1%) vs 5/1110 failed VBAC-2 9.6%,
Case series Previous low vertical and 69% after two CS and augmentation and in VBAC-1 (0.45%), hysterectomy failed VBAC-1 9.1%
unknown uterine 92% after 1CS had trial. epidural, identical 0.33% vs 0% Asphyxial injury caused
scars included to those with an by uterine
unscarred uterus. rupture 0.33%
vs 0.09%
Chattopadhyay13 115 cases VBAC-2 Compared VBAC-2 Prostaglandin IOL Success 103/115 (90%) Scar dehiscence Peri-natal mortality 2.6%
(Saudi Arabia) 1006 cases RCS (10%) and 37/115 (32%) Similar success rate with (0.8% vs vs 1%, deaths not
ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology
Case series Only previous elective CS (90%). Oxytocin augmentation history of vaginal delivery 0.7%) and considered directly
low transverse Only 115/230 32/115(28%) hysterectomy related to mode
uterine scars included requesting for a trial (0.8% vs 1.4%) of delivery
were permitted comparable
Granovsky-Grisaru20 26 cases VBAC-2 Control group from Only spontaneous labour 19/26 (73%) success No uterine scar Baby weight 2800–4600 g,
(Israel) 26 controls RCS elective CS, denominator Syntocinon Ventouse delivery 4/26 rupture or dehiscence 5-min Apgar scores = 7–9
Case series 3/26 had three previous-CS data not given augmentation 54% Hospital stay No NICU admissions
Only previous Epidural 80% average 3 days
low transverse
uterine scars included
Success rate & adverse outcomes of vaginal birth after two caesarean sections
9

6. 10
Table 1. (Continued)
Study Study Methods Labour Success rate Maternal Neonatal
Tahseen, Griffiths
reference population management outcome outcome
Miller26 1827 previous two CS Compared VBAC-2 No information Success 75% Uterine rupture Rupture related
(North America) (1586 previous two CS, 241 with VBAC-1; available previous two 1.8% (vs 0.6% peri-natal death
Cohort data previous ‡3 CS), 10 880 54% previous CS vs 83% .018% vs trials
from caesarean previous one CS. Previous ‡2 CS had previous 1CS vs 0.05%
registry/delivery classical scars excluded but trial vs 80%
records unknown scars included! previous 1CS
Guettier21 (France) 17/41 women with two A case series of 17 Included one home 9/17 (53%) vaginal No relevant No neonatal morbidity
Case series previous uterine scars, delivery, 2 pt fully delivery, 3/9 actually morbidity in 17 cases, but three
ncluding myomectomy/ dilated on admission had two CS major congenital
hysterotomy abnormality
Flamm16 5733 trial of labour All women undergoing No separate figures 69% success (168/245) No separate No separate figures
(North America) after CS, including trial of labor (38% of for VBAC-2 for VBAC-2, 75% figures for VBAC-2 f or VBAC-2
Case series 245cases VBAC-2, previous CS) no separate for whole group
previous classical scars denominator data for
excluded but Pre two CS available
unknown scars included!
Hansell22 35 cases VBAC-2 21% of women with ‡2 Only spontaneous labour Success 77% (27/35). 0% uterine rupture 5 min Apgar score <5,
(North America) 2/35 had previous previous CS had trial, No Oxytocin Higher success rate and hysterectomy none in the two
Case series three CS, 1/35 135 cases without trial CS augmentation with history of in trial group groups
pre.4 CS. 135 cases vaginal delivery Transfusion 1/35 (2.8%)
non-trial CS vs 11/135 (8.1%) in
Low transverse scars without trial CS
Phelan28 501 cases VBAC-2 Compared VBAC-2 Oxytocin used for Success 69%. Uterine rupture 0% vs No perinatal deaths
(North America) and 587 elective CS (46%) and elective induction (3.6%) Higher success rate with 0.2% on elective CS attributable to trial,
Case series Previous unknown CS (54%) augmentation 53% history of Hysterectomy 0.2% vs 5 min Apgar <7 2.6%
scars included vaginal delivery 1.2% in elective CS in VBAC-2 and 1.4%
in Elective CS
Novas27 36 cases having VBAC-2 Compared VBAC-2 36(52%) Oxytocin 47% Success 80% (29/36) Uterine rupture 1/ Apgar scores
(North America) (nine had previous and non-trial CS, 33 (48%) 36–2.7%(H/O classical comparable, no figures
Case series three CS) CSs revealed at provided.
ERCS 33 laparotomy) vs 0/33 PND unrelated to trial
Previous unknown Hysterectomy 0/36
uterine scars included vs 2/33
ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology

7. Success rate & adverse outcomes of vaginal birth after two caesarean sections
patients and clinicians and to compare the level of risk
No significant peri-natal
No neonatal morbidity
associated with VBAC-1 versus VBAC-2.
All Apgar scors >7
Neonatal Success rates of trial of labour were available in all
outcome included studies. Uterine rupture rates were given in all
studies except one;16 reported trial of scar data after single
morbidity
at 5 min
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%
2/55 (3.6%) hysterectomies.
previous vertical scar, one
(10/5733) for the whole study population was given.
uterine rupture (two had
Transfusion 1/55 (1.8%)
vaginal del.3/55 (5.4%)
placenta accrete) and Asymptomatic scar dehiscence found incidentally at caesar-
endometritis, none in
10/30 failed trial had
ean section was disregarded and only symptomatic scar
Maternal
outcome
No uterine rupture
ruptures were included in analysis. Different maternal mor-
No scar rupture
bidity 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 rup-
ture were available in some studies reporting on large
Success rate
Success 77%
45% success
81% success
cohorts from birth/caesarean registries, Miller26 reported
(44/57)
17 322 cases of trial of scar from a 10-year period and
Flamm16 reported 5733 trial of scar cases over 5 year. Hys-
terectomy and blood transfusion rates were assessed, wher-
Only spontaneous labour
Only spontaneous labour
Only spontaneous labour
ever available. Other maternal morbidities as operative
Oxytocin augmentation
injury or ITU admissions were variably classified24,25 and
management
Labour
were too diverse for the purpose of pooled analysis.
augmentation
33% Oxytocin,
55% Oxytocin
19% epidural
Few studies have considered the neonatal outcomes, limit-
5/57 (8%)
ing 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
spontaneous labour (? matched)
describing neonatal outcomes for all subjects were included
(Table 2). No perinatal outcomes were reported by Porreco29
Selected sample, pre2 CS
Selected sample, a report
(indigent population, late
and Chattopadhyay13. Apgar scores considered in few studies
in spontaneous labour
Self-selected motivated
Methods
describing VBAC-2
controls (64) ERCS
comparator group
comparator group
were variably reported as 1-minute score or 1-, 5- and/or
VBAC, vaginal birth after caesarean; RCS, repeat (third) caesarean section.
10-minute scores; 5-min Apgar scores below 7 (reported by
booking), no
patients. No
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
unknown uterine scars included
first minute’, this statement was un-informative to assess neo-
denominator data not available
57 cases VBAC-2 (18 had three
55 cases VBAC-2, denominator
natal outcomes in their study. We considered Apgar scores
21 cases.Only low transverse
data not available Previous
uterine scars, denominator
to be too diverse for aggregate comparison. Asphyxial injury/
transverse uterine scars,
previous CS). Only low
population
HIE/perinatal death and neonatal unit admission rates,
Study
data not available
where reported were pooled to assess neonatal outcome.
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
Table 1. (Continued)
all included studies. Table 3 shows comparison of VBAC-1
(North America)
(North America)
(North America)
versus VBAC-2 reported in the same cohorts (six studies)
Farmakides15
and only the paired available outcomes are tabulated.
Case series
Case series
Case series
reference
Porreco29
Pruett30
Table 4 shows comparison of VBAC-2 with non-trial repeat
Study
(third) caesarean section within same cohorts (eight stud-
ies), again only the paired outcomes are listed.
ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology 11

8. Tahseen, Griffiths
Table 2. Outcome of VBAC-2 for all included studies
Study reference Numbers VBAC Uterine Transfusion Hysterectomy Asphyxial NNU
VBAC-2 success Rupture injury/PND**
Macones25 1082 807 (74.6) 20 (1.8) 10 (0.92)
Landon24 975 648 (66) 9 (0.9) 31 (3.2) 6 (0.6) 1 (0.15) 75 (11.2)
Garg18 100 66 (66) 0 (0.0) 0 0
Spaan31 59 49 (83) 1 (1.8) 4 (6.7) 1 (1.7) 0 (0.0) 15 (34.9)
Bretelle11 96 63 (65.6) 0 (0.0) 2 (2.1) 1 (1.04) 0 (0.0) 0 (0.0)
Caughey12 134 83 (62) 5 (3.7) 1 (0.75)
Asakura9 302 194 (71) 3 (1) 1 (0.33) 1 (0.33)
Chattopadhyay13 115 103 (89) 1 (0.7) 1 (0.89)
Granovsky-Grisarau20 26 19 (73) 0 (0.0) 0 0 (0.0) 0 (0.0) 0 (0.0)
Miller26 1827 1376 (75) 32 (1.8) 1 (0.05)
Flamm16 245 168 (69%) NA
Hansell22 35 27 (77) 0 (0.0) 1 (2.8) 0 (0.0)
Phelan28 501 346 (69) 0 1 (0.2)
Novas27 36 29 (80) (0.0) 0
Pruett30 55 25 (45) 3* (5.4) 1 (1.8) 2 (3.6)
Farmakides15 57 44 (77) 0 (0.0) 0
Porreco29 21 17 (81) 0 (0.0) 0
Total 5666 4064 (71.7) 74/5421 (1.36) 49/2428 (2.01) 14/2512 (0.55) 3/3285 (0.09) 90/1156 (7.78)
Values in parenthesis are expressed in percentage.
*2/3 uterine ruptures diagnosed by palpation after successful vaginal delivery.
**Prelabour stillbirths and postnatal deaths unrelated to mode of delivery (e.g. prematurity related) were not included.
Table 3. Outcome of VBAC-2 versus VBAC-1: only outcomes with paired data available are included
Study reference Group Number Success Uterine rupture Transfusion Hysterectomy PND/Asphyxial injury* NNU
Macones25 VBAC-2 1082 807 (74.6) 20 (1.8) 10 (0.92)
VBAC-1 12 535 9464 (75.5) 113 (0.9) 85 (0.68)
Landon24 VBAC-2 975 648 (66) 9 (0.9) 31 (3.2) 6 (0.6) 1 (0.15) 75 (11.2)
VBAC-1 16 915 12 490 (74) 115 (0.7) 273 (1.6) 35 (0.2) 14 (0.09) 1321 (9)
Caughey12 VBAC-2 134 83 (62) 5 (3.7) 1 (0.75) 0 (0.0)
VBAC-1 3757 2818 (75) 31 (0.8) 7 (0.19) 1 (0.02)
Asakura9 VBAC-2 302 194 (71) 3 (1.04) 1 (0.33) 1 (0.33)
VBAC-1 1110 856 (64) 5 (0.45) 0 (0.0) 1 (0.09)
Miller26 VBAC-2 1827 1376 (75) 32 (1.8) 1 (0.05)
VBAC-1 10 880 9063 (83) 63 (0.6) 2 (0.018)
Flamm16 VBAC-2 245 168 (69)
VBAC-1 5488 4123 (75)
Total VBAC-2 4565 3276 (71.7) 69 (1.59) 41 (1.99) 8 (0.56) 3 (0.09) 75 (11.2)
VBAC-1 50 685 38 814 (76.5) 327 (0.72) 358 (1.21) 42 (0.19) 17 (0.05) 1321 (9)
Values in parenthesis are expressed in percentage.
*Prelabour stillbirths and postnatal deaths unrelated to mode of delivery (e.g. prematurity related) were not included.
OR = 1.48 of higher success rate in VBAC-1 group versus
Effectiveness VBAC-2, CI 1.23–1.78 (Figure 2, P < 0.0001, Z = 4.18).
Successful vaginal delivery was achieved in 4064/5666
(71.1%) as shown in Table 2, ranging in studies from 45% Adverse maternal outcomes
to 89%. The comparable rate in VBAC-1 group was higher
38 814/50 685 (76.5%—Table 4); meta-analysis showed a Uterine rupture rate after VBAC-2 was reported in all stud-
significant difference between the two groups with ies except by Flamm16. The pooled uterine rupture rate of
12 ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology

9. Success rate & adverse outcomes of vaginal birth after two caesarean sections
Table 4. Outcome of studies comparing VBAC-2 versus Repeat (third) Caesarean Section (RCS)
Study reference No. cases Uterine rupture Transfusion Hysterectomy Fever Peri-natal death NNU admission
Macones25 VBAC-2 1082 19 (1.76) 10 (0.92) 96 (8.8)
CS 2888 1 (0.03) 34 (1.18) 366 (12.7)
Landon24 VBAC-2 975 9 (0.9) 31 (3.2) 6 (0.6) 30 (3.1) 1 (0.1) 75 (11.2)
CS 6035 0 (0%) 93 (1.5) 27 (0.4) 129 (2.1) 1 (0.02) 514 (9.1)
Spaan31 VBAC-2 59 1 (1.7) 4 (6.7) 1 (1.7) 5 (8.4) 15 (35)
CS 187 2 (1.06) 19 (10) 0 (0) 17 (9) 39 (23)
Chattopadhyay13 VBAC-2 115 1 (0.8) 1 (0.8) 1 (0.8)
CS 1006 7 (0.7) 15 (1.4) 15 (1.4)
Granovsky-Grisarau20 VBAC-2 26 0 (0) 0 (0) 0 (0) 2 (7.7) 0 (0) 0 (0)
CS 26 0 (0) 0 (0) 0 (0) 5 (19) 0 (0) 0 (0)
Hansell22 VBAC-2 35 0 (0) 1 (2.8) 0 (0) 4 (11.4)
CS 135 1 (0.7) 11 (8) 0 (0) 39 (28)
Phelan28 VBAC-2 501 0 (0) 1 (0.2) 55 (11)
CS 587 1 (0.17) 7 (1.2) 74 (12)
Novas27 VBAC-2 36 1 (2.7) 0 (0)
CS 33 0 (0) 2 (6)
Total VBAC-2 2829 31 (1.09) 47 (1.68) 9 (0.40) 192 (6.03) 1 (0.09) 90 (8.49)
CS 10 897 12 (0.11) 172 (1.67) 51 (0.63) 630 (6.39) 1 (0.01) 553 (8.85)
Values in parenthesis are expressed in percentage.
16 studies was 1.36% (74/5421) (Table 2), ranging 0–5.4% The rate of hysterectomy was reported in eight studies,
within studies. Subgroup comparative analysis with VBAC- pooled average was 0.55% in VBAC-2 group, ranging
1 in five studies (Table 3), revealed rupture rates 0.72% in within studies 0–3.6% (Table 2). Considering hysterectomy
VBAC-1 versus 1.59% in VBAC -2; meta-analysis showed figures in the comparison of VBAC-2 and VBAC-1
pooled OR = 0.42 of a uterine rupture in VBAC-1 group (Table 3), the rates were 0.56% versus 0.19%; meta-analy-
versus VBAC-2, CI 0.29–0.60 (Figure 3, P < 0.0001, sis of three studies reporting hysterectomy rates comparing
Z = 4.65). VBAC-1 versus VBAC-2, showed OR = 0.29 of hysterec-
A lower risk of uterine rupture with history of prior vag- tomy in VBAC-1 versus VBAC-2 groups, CI 0.13–0.61
inal delivery was indicated; Macones25 in a large birth reg- (P = 0.001, Z = 3.22, Figure 4). In the subset of data com-
istry cohort reported a uterine rupture rate of 1.8% in paring VBAC-2 with repeat CS (Table 4), the hysterectomy
VBAC-2 versus 0.9% in VBAC-1. Previous vaginal delivery rates were similar, 0.40% and 0.63%, respectively;
appeared protective for uterine rupture as 0.5% compared meta-analysis of seven studies reporting the paired out-
to 2.4% rupture noted, respectively, with and without a come showed OR = 0.75 of hysterectomy in VBAC-2
previous vaginal delivery. Similarly Caughey12 reported group versus RCS, CI 0.23–2.43 (Figure 5, P = 0.63,
subjects with previous vaginal delivery were one-fourth as Z = 0.48). Rate of major haemorrhage was not specified in
likely to have a uterine rupture as those without. any of the papers but numbers needing blood transfusion
Figure 2. Success rate of VBAC-2 versus VBAC-1.
ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology 13

10. Tahseen, Griffiths
Figure 3. Uterine rupture rates in VABC-2 versus VBAC-1 groups.
Figure 4. Hysterectomy rates in VABC-2 versus VBAC-1 groups.
were given in eight studies, average 2.01% range 0–6.7% OR = 0.94 of having a blood transfusion in VBAC-2 versus
(Table 2). Transfusion rates were lower in the VBAC-1 RCS, CI 0.45–1.96 (Figure 7, P = 0.86, Z = 0.17).
group 1.21% versus 1.99% in VBAC-2 (Table 3); meta- Febrile morbidity was reported particularly in compari-
analysis showed OR = 0.56 of having a blood transfusion son with repeat caesarean sections in six studies, 6.03% in
in VBAC-1 versus VBAC-2, CI 0.40–0.77 (Figure 6, P = VBAC-2 and 6.39% in RCS group Table 4. Meta-analysis
0.0004, Z = 3.52) and similar in repeat CS and VBAC-2 of six studies reporting the paired outcome showed
groups (1.67% and 1.68% respectively Table 4). Meta-anal- OR = 0.81 of febrile morbidity in VBAC-2 versus RCS, CI
ysis of six studies reporting paired transfusion rates showed 0.55–1.18 (Figure 8, P = 0.27, Z = 1.11).
Figure 5. Hysterectomy rates in VBAC-2 versus RCS.
14 ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology

11. Success rate & adverse outcomes of vaginal birth after two caesarean sections
Figure 6. Blood transfusion rates in VBAC-2 versus VBAC-1 groups.
VBAC-2 RCS Odds ratio Odds ratio
Study or subgroup Events Total Events Total Weight M-H, random, 95% CI M-H, random, 95% CI
Chattopadhyay 1994 1 115 15 1006 9.7% 0.58 [0.08, 4.43]
Granovsky-Grisarau 1994 0 26 0 26 Not estimable
Hansall 1990 1 35 11 135 9.4% 0.33 [0.04, 2.66]
Landon 2006 31 975 93 6035 33.3% 2.10 [1.39, 3.17]
Macones 2005 10 1082 34 2888 27.7% 0.78 [0.39, 1.59]
Spann 2003 4 59 19 187 20.0% 0.64 [0.21, 1.97]
Total (95% CI) 2292 10277 100.0% 0.94 [0.45, 1.96]
Total events 47 172
Heterogeneity: τ² = 0.38; χ² = 11.05, df = 4 (P = 0.03); I² = 64%
0.01 0.1 1 10 100
Test for overall effect: Z = 0.17 (P = 0.86) Blood transfusion VBAC-2Blood transfusion RCS
Figure 7. Blood transfusion rates in VBAC-2 versus Repeat (third) CS (RCS) groups.
VBAC-2 RCS Odds ratio Odds ratio
Study or subgroup Events Total Events Total Weight M-H, random, 95% CI M-H, random, 95% CI
Granovsky-Grisarau 1994 2 26 5 26 4.2% 0.35 [0.06, 2.00]
Hansall 1990 4 35 39 135 8.7% 0.32 [0.11, 0.96]
Landon 2006 30 975 129 6035 23.9% 1.45 [0.97, 2.18]
Macones 2005 96 1082 366 2888 28.9% 0.67 [0.53, 0.85]
Phelan 1989 55 501 74 587 24.9% 0.85 [0.59, 1.24]
Spann 2003 5 59 17 187 9.5% 0.93 [0.33, 2.63]
Total (95% CI) 2678 9858 100.0% 0.81 [0.55, 1.18]
Total events 192 630
Heterogeneity: τ² = 0.12; χ² = 14.38, df = 5 (P = 0.01); I² = 65%
0.01 0.1 1 10 100
Test for overall effect: Z = 1.11 (P = 0.27)
Fever VABC-2 Fever RCS
Figure 8. Febrile morbidity rates in VBAC-2 versus Repeat(third) CS (RCS) groups.
Non-specific reassuring statements regarding maternal death attributed to mode of delivery occurred in 0.09%
morbidity were given in some studies as ‘no maternal compli- (range 0–0.33%), Table 2. Neonatal unit admission rates
cations occurred’,29 ‘no febrile morbidity noted’,30 ‘outcomes only were reported in some studies as an index of neonatal
were similar to hospital’s general obstetric population’.27 morbidity rather than Apgar scores. Pooled NNU admis-
sion rate was 7.7% (range 0–34.9%), Table 2. The neonatal
outcome of five studies comparing VBAC-1 and VBAC-2
Adverse neonatal outcomes
had similar rates for neonatal asphyxial injury/perinatal
Neonatal Apgar scores are not analysed further due to vari- death (attributable to mode of delivery) after VBAC-2
able reporting within studies. Asphyxial injury or perinatal (0.09%) versus VBAC-1 (0.05%) (P = 0.35, Mantel–Haens-
ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology 15

12. Tahseen, Griffiths
zel). The NNU admission rates were also comparable pattern (rupture rate 1.8%25); ‘….and full thickness dehis-
(P = 0.89, Mantel–Haenszel). Subgroup analysis of VBAC-2 cence found at caesarean section performed for acute fetal
versus repeat (third) caesarean sections revealed rates of distress (rupture rate 1.8%26). Earlier studies which con-
perinatal death/asphyxial injury, 0.09% with VBAC-2 versus tributed a heavy weight in the systematic review included
0.01% with repeat caesarean sections (P = 0.14, Mantel– patients with unknown uterine scars (with possibility of
Haenszel), although the difference does not reach statistical unknown proportion of lower vertical/classical scars) and
significance but may be clinically different. The NNU reported higher uterine rupture rates (3.6%12; 2.7%27;
admission rates were similar as well (8.85 versus 8.49% 5.4%30; 1.8%26). A lower risk of scar rupture is indicated
P = 0.57, Mantel–Haenszel). in the subgroup in Table 4 (with data from more recent
Non-specific reassuring statements regarding neonatal years), 1.09% as compared to a rate of 1.59% in subgroup
morbidity were given in some studies; Farmakides15 ‘there in Table 3 where earlier data (subjects26) contributed a
was no significant perinatal morbidity’; Novas27 ‘no signifi- higher proportion. Where manual scar exploration after a
cant differences were observed between the two groups in successful vaginal delivery was routinely performed and full
gestational age, Apgar scores and birthweights and no rup- thickness uterine wall defects repaired,26,30 a higher scar
ture related perinatal death’; Garg18 ‘no difference in fetal rupture rate was reported. Manual scar exploration is now
outcome’; no neonatal data were extracted from these rarely carried out and this together with a known lower
studies. segment uterine scar may be a reason for lower scar rup-
ture rates in more recent studies.
Overall hysterectomy rate was 0.55%. A lower rate noted
Discussion
after VBAC-1 (0.19%), however, rate after RCS (0.63%,
This review shows that trial of vaginal delivery after two Table 4) was similar within subgroup analysis, which prob-
caesarean sections is associated with a reasonable success ably reflects the higher surgical risks associated with previ-
rate (71.7%), although lower than VBAC-1 (76.5%). The ous multiple caesarean sections including placenta accrete/
adverse maternal outcomes of a trial of vaginal delivery praevia. Noteworthy, a lower threshold for hysterectomy in
after two previous caesareans are comparable to repeat Pruett’s paper30 (3.6%) was because of women’s desire for
(third) caesarean sections, with similar hysterectomy tubal ligation, when scar defects were detected on manual
(P = 0.63), blood transfusion (P = 0.86) and febrile mor- scar palpation after successful vaginal delivery.
bidity (P = 0.27) rates. The adverse maternal outcomes The blood transfusion rates were similar as with RCS
rates of VBAC-2 are higher than VBAC-1, but the absolute (1.68% versus 1.67%), as well as febrile morbidity (6.03%
rates are small. The neonatal outcome data are limited, versus 6.39%). Higher febrile morbidity after a failed trial
however, the available data does not indicate a significant of labour and in repeat caesarean sections (33%30, 28%22,
difference (assessed by neonatal death/asphyxial injury and 19.2%20) reported in earlier studies has significantly
Neonatal Unit admission rates) between VBAC-2, RCS or reduced after advent of broad spectrum antibiotics and
VBAC-1. Although our comparison was carried out with concern regarding postoperative infectious morbidity is not
VBAC-1 group because of its wide acceptance, pragmatic a major issue in selecting mode of delivery.
and rational comparison of maternal morbidity of VBAC-2 The proportion of women undergoing a trial after two
is with RCS, as previous multiple operations would have a caesarean sections is variable between studies which may
higher background risk43and the available alternative choice indicates variable selection criteria; from 9.2% by Landon24
to women who already had two CS is a RCS. to 69% by Asakura9. A more selective approach may be
The reluctance to offer a trial after two caesarean sec- associated with higher success and/or lower uterine rupture
tions is likely to stem from concerns regarding scar rup- rate,24 but a clear pattern does not emerge.
ture. Scar rupture is a rare event and individual studies are Neonatal morbidity with VBAC-2 assessed by neonatal
limited by size making uterine rupture risk a difficult out- unit admissions was comparable to the RCS group
come to assess, pooled data analysis provides more reliable (Table 4). The more important measure of neonatal mor-
figures. The rupture rate in the pooled analysis was 1.36% bidity, hypoxic neonatal brain injury/death attributable to
(Table 2). All included studies provided figures for scar mode of delivery was reported in six studies (3285 sub-
rupture, slightly varying but clinically meaningful defini- jects), the pooled rates were 0.09% in VBAC-2 as compared
tions of rupture have been used in different studies as ‘dis- 0.01% in RCS group (P = 0.14) and 0.05% in VBAC-1
ruption of uterine muscle and peritoneum—or disruption group (P = 0.34). The publications involved a time period
of muscle and extension to bladder or broad ligament’ in of one or two decades ago with less-advanced neonatal
Landon24 (2006, rupture rate 0.9%); ‘Signs and symptoms facilities, hence the neonatal morbidity figures may not be
of intra-peritoneal bleeding—or disruption of uterine scar representative of current practice given considerable
immediately preceded by non-reassuring fetal heart rate advances in neonatal care in recent years, moreover, the
16 ª 2009 The Authors Journal compilation ª RCOG 2009 BJOG An International Journal of Obstetrics and Gynaecology