REVIEW

Safety of Vaginal Birth After Cesarean: A Systematic

Review
Jeanne-Marie Guise, MD, MPH, Michelle Berlin, MD, MPH, Marian McDonagh, PharmD,
Patricia Osterweil, Benjamin Chan, MS, and Mark Helfand, MD, MPH

OBJECTIVE: To evaluate the benefits and harms of vaginal deficiencies in the literature evaluating the relative safety of
birth after cesarean compared with repeat cesarean vaginal birth after cesarean compared with repeat cesar-
delivery. ean delivery are striking. The identification of high-risk
DATA SOURCES: The computerized databases MEDLINE, and low-risk groups of women and settings for morbidity
EMBASE, HealthSTAR, Cochrane CENTRAL, and Na- remains a key research priority. (Obstet Gynecol 2004;
tional Centre for Reviews and Dissemination Database of 103:420 –9. © 2004 by The American College of Obstetri-
Abstracts of Reviews of Effectiveness, along with reference cians and Gynecologists.)
lists and national experts, were used to conduct this review.
METHODS OF STUDY SELECTION: All studies that reported Pregnancy and childbirth are the most common indica-
data for maternal or infant outcomes in women with prior tions for hospitalization in the United States, represent-
cesarean delivery were eligible. Methodological quality ing 1 in 4 hospital stays for women.1 Cesarean deliveries
was evaluated for each study with the criteria of the United are at the highest level ever reported in U.S. history.2 In
States Preventive Services Task Force and the National contrast, the vaginal birth after cesarean (VBAC) rate
Health Service Centre for Reviews and Dissemination. continues to decline, from 28.3% in 1996 to 12.7% in
Twenty of 6,828 potentially relevant articles (55,506 pa- 2002.2
tients) were included in the analysis. The reasons for the decreasing VBAC rate and the
TABULATION, INTEGRATION, AND RESULTS: Two authors increasing rate of cesarean are not completely under-
independently abstracted information on study design, stood, but concerns for the safety of mother and infant
sample size, participant characteristics, and maternal and are considered to be prominent factors in decisions about
fetal health outcomes by using a standardized protocol. childbirth. To assess the evidence base of these concerns,
Rates of vaginal delivery in women undergoing a trial of
we conducted a systematic review of the health outcomes
labor ranged from 60% to 82%. There was no significant
experienced by women and their offspring after induced
difference in maternal deaths or hysterectomy between
trial of labor and repeat cesarean. Uterine rupture was or spontaneous trial of labor, VBAC, or elective repeat
more common in the trial-of-labor group, but rates of cesarean. The strength and suitability of the evidence
asymptomatic uterine dehiscence did not differ. Studies regarding the risks of major maternal and infant morbid-
conflicted on the effect of induction of labor on these out- ity and mortality associated with VBAC are the main
comes. Data regarding infant outcomes were poor. focuses of this article. This review is derived from infor-
CONCLUSION: Safety in childbirth for women with prior mation gathered for an evidence report on VBAC con-
cesarean is a major public health concern. Methodological ducted for the U.S. Agency for Healthcare Research and
Quality.3
From the Department of Obstetrics and Gynecology, Evidence-based Practice Center
and Department of Medical Informatics and Clinical Epidemiology, Oregon Health
& Science University, Portland, Oregon.
SOURCES
This study was conducted by the Oregon Health & Science University Evidence- The primary investigator and a medical librarian
based Practice Center under contract to the Agency for Healthcare Research and
searched the computerized databases MEDLINE,
Quality, Rockville, Maryland (contract 290-97-0018, task order no. 9). Support
for Dr. Guise was also provided by the Agency for Healthcare Research and Quality HealthSTAR, and EMBASE from 1980 to 2002, using
grant no. 1 K08 HS11338-01. search terms such as “vaginal birth after cesarean,” “trial
of labor/labour,” “trial of scar,” “labor,” “delivery,”
The authors of this article are responsible for its contents, including any clinical or
treatment recommendations. No statement in this article should be construed as an “vaginal birth,” “vaginal delivery,” “cesarean,” “cesar-
official position of the U.S. Agency for Healthcare Research and Quality or the U.S. ean section,” “(labor) induction,” “augmentation,”
Department of Health and Human Services. “home child birth,” “natural child birth,” “obstetrical

VOL. 103, NO. 3, MARCH 2004

420 © 2004 by The American College of Obstetricians and Gynecologists. 0029-7844/04/$30.00
Published by Lippincott Williams & Wilkins. doi:10.1097/01.AOG.0000116259.41678.f1
extraction,” and “uterine rupture.” Only data published
in full-article form were included. We also searched the
online Cochrane Database of Systematic Reviews,
Cochrane Controlled Trials Register, and the National
Health Service Centre for Reviews and Dissemination
Database of Abstracts of Reviews of Effectiveness. Ad-
ditionally, articles were obtained by reviewing reference
lists of pertinent studies and reviews and from recom-
mendations from experts. The search was limited to
1980 or later based on the timing of the 1980 National
Institute of Health Consensus Development Conference
on Vaginal Birth after Cesarean,4 which established
VBAC as an acceptable delivery choice.

STUDY SELECTION
Controlled trials, cohort studies, case-control studies,
and case series with at least 10 cases were eligible for
inclusion in the systematic review if they 1) included
women with prior low-transverse cesarean delivery (or
unknown scar); 2) provided data regarding uterine rup-
ture, VBAC rates, and maternal or infant outcomes for
VBAC or repeat cesarean; and 3) were conducted in
developed countries.6 Studies were excluded if they fo-
cused on patients with particular medical (eg, gestational Figure 1. Study eligibility flow chart. *Excluded criteria
diabetes or human immunodeficiency virus) or obstetric related to: (1) no patients with prior low-transverse cesar-
(eg, preeclampsia, multiple gestation, breech presenta- ean or unknown scar; (2) no data regarding the following:
tion) conditions. uterine rupture, vaginal birth after cesarean rates, maternal
outcomes, and infant outcomes; (3) fewer than 10 patients
Two investigators (J.M.G. and M.M.) independently
in study; (4) patients not from developed country; (5)
reviewed a random set of titles and abstracts to establish patients with particular medical or obstetric conditions (eg,
reliable and reproducible inclusion criteria. Once an human immunodeficiency virus, gestational diabetes).
appropriate level of reliability was established (ie, ␬ of Guise. VBAC Safety: Systematic Review. Obstet Gynecol 2004.
0.80 or greater), the primary investigator reviewed the
remaining titles and abstracts for inclusion.
Two investigators independently rated each study’s significant heterogeneity was found, pooling was not
quality using criteria developed by the United States undertaken. Clinical heterogeneity was also assessed,
Preventive Services Task Force and the National Health and only studies of similar patients and interventions
Service Centre for Reviews and Dissemination.5 Dis- were considered for pooling. Both random and fixed
agreements were resolved by consensus between the 2 effects models were used in pooling; because the 2 meth-
reviewers and consultation with a third reviewer if nec- ods produced similar results, only the results of the
essary. Two reviewers (J.M.G. and M.B.) independently random effects models are presented. Statistical analyses
abstracted study design, setting, population demograph- were performed with StatsDirect software (CamCode;
ics, uterine rupture definition, rates, and predictors, ma- StatsDirect Ltd, Cheshire, United Kingdom).
ternal and fetal health outcomes, and methods for assess-
ing or adjusting for confounders.
Using a prespecified protocol for meta-analysis of
observational studies,6 we pooled eligible studies to esti- RESULTS
mate the likelihood of several outcomes of trial of labor We identified 6,828 citations, reviewed 614 full text
and elective repeat cesarean. To reduce potential bias, articles, and identified 20 fair-to-good quality studies,
only studies of fair or good quality were included in these including 2 randomized controlled trials and 18 observa-
analyses. For each outcome, we calculated the absolute tional studies (Figure 1). All non-English articles were
risk differences from each study and assessed statistical reviewed, and none provided additional information.
heterogeneity using the Q-statistic.7 When statistically Characteristics of these studies are described in Table 1.

VOL. 103, NO. 3, MARCH 2004 Guise et al VBAC Safety: Systematic Review 421
Table 1. Health Outcomes of Vaginal Birth After Cesarean Delivery
Study (quality) Population Reported outcomes
Randomized controlled trials
Rayburn et al,24 One prior low-transverse N ⫽ 294 Spontaneous labor ⫽ Induction ⫽ 143
1999; United cesarean, gestational age 151
States (fair) ⱖ38 wk, no labor, no Uterine rupture 0 0
fetal growth Maternal bleeding 0 0
abnormalities, reassuring
FHR tracings, and Maternal infection 7 8
unfavorable cervix Maternal death 0 0
(Bishop score ⱕ 6) Infant death 1 0
Lelaidier et al,25 Women with 1 prior N ⫽ 32 Spontaneous labor Induction ⫽ 16
1994; France cesarean, Bishop’s scores ⫽ 16
(fair) ⱕ3 Uterine rupture 1 1
Maternal infection 1 1
Maternal death 0 0
Infant death 0 0
Population-based database
McMahon et al,8 Singleton pregnancies with N ⫽ 6,138 Spontaneous labor Elective repeat
1996; Canada 1 prior low transverse (induction and/or cesarean ⫽
(good) cesarean augmentation) ⫽ 2,889
3,249
Vaginal delivery rate 1,962
Uterine rupture 10 1
Maternal bleeding 5 hysterectomies, 5 6 hysterectomies,
transfusions 39
transfusions
Maternal infection 171 fever, 43 185 fever, 64
wound infection wound
infection
Maternal death 0 0
Infant death 9 per 1,000 5 per 1,000
Smith et al,21 Women with prior N ⫽ 24,529 Spontaneous labor Elective repeat
2002; Scotland cesarean, singleton (induction and/or cesarean ⫽
(fair) pregnancies between 37– augmentation) ⫽ 9,014
42 weeks, cephalic, 15,515
without lethal congenital
anomalies Infant death 20 1
Prospective cohort
Duff et al,20 1988; All women with 1 prior N ⫽ 227 Spontaneous labor
United States low-transverse cesarean (induction and/or
(good) augmentation) ⫽
227
Vaginal delivery rate 167
Uterine rupture 1
Maternal bleeding No differences
Maternal infection 12 successful
VBAC, 11 failed
VBAC
Flamm et al,17 One or more prior low N ⫽ 1,776 Spontaneous labor Spontaneous
1988; United transverse cesareans; (no oxytocin) ⫽ labor
States (good) unknown scar allowed 1,291 (oxytocin) ⫽
485
Vaginal delivery rate 1,005 309
Uterine rupture Overall: 0 successful VBAC, 3 failed
VBAC (11 asymptomatic uterine
windows)
Maternal bleeding Overall: 1 successful VBAC, 1 failed
VBAC
Maternal death Overall: 0
Infant death Overall: 5 antepartum fetal deaths
(continued)

Results from the large population-based Nova Scotia mented) after prior low-transverse cesarean or unknown
study provided the most accurate representation of the scar in the general population. In that study, 3,249
frequency of vaginal delivery in women who undergo a (52.9%) of 6,138 women with 1 prior nonvertical cesar-
trial of labor (spontaneous onset, induced, and aug- ean chose a trial of labor, and 1,962 of them (60.4%)

422 Guise et al VBAC Safety: Systematic Review OBSTETRICS & GYNECOLOGY

Table 1. Health Outcomes of Vaginal Birth After Cesarean Delivery (continued)
Study (quality) Population Reported outcomes
13
Flamm et al, One or more prior low- N ⫽ 1,776 Spontaneous labor Induction and/or
1987; United transverse cesareans; ⫽ 1,291 augmentation
States (good) unknown scar allowed ⫽ 485
Vaginal delivery rate 1,005 309
Uterine rupture 1 2
Maternal bleeding 1 1
Maternal infection 35 18
Maternal death 0 0
Infant death 1 0
Flamm et al,26 All women with 1 or more N ⫽ 5,022 Spontaneous labor Induction and/or
1997; United prior cesareans; ⫽ 4,569 augmentation
States (fair) unknown scar allowed ⫽ 453
Vaginal delivery rate 3,513 233
Uterine rupture 33 6
Cowan et al,10 Women with prior low- N ⫽ 593 Spontaneous labor Induction ⫽ 67 Augmentation
1994; United transverse cesarean ⫽ 359 ⫽ 167
States (fair) choosing trial of labor; Vaginal delivery rate 315 46 117
unknown scar allowed (2 Uterine rupture 3 Spontaneous labor (induction and/
with vertical cesareans or augmentation): 2
entered)
Maternal major bleeding Spontaneous labor (induction and/or
augmentation)–successful VBAC 453 (95%), estimated
blood loss ⬍ 500 mL; 14 with estimated blood loss 501–
700 mL; 10 with estimated blood loss 701–1000 mL; 1
with estimated blood loss ⬎ 1,000 mL; UR average
estimated blood loss ⬎ 1,500 mL; 3/5 UR symptomatic
blood loss
Flamm et al,16 All women with 1 or more N ⫽ 7,229 Spontaneous labor Elective repeat
1994; United prior transverse cesarean; (induction and/or cesarean ⫽
States (fair) unknown scar allowed augmentation) ⫽ 2,207
5,022
Vaginal delivery rate 3,746
Uterine rupture 39
Maternal bleeding 3 hysterectomies, 36 5 hysterectomies,
transfusions 38
transfusions
Maternal infection 638 362
Maternal death 1
Infant death Overall: 7 per 1,000 live births
Blanco et al,22 Prior lower-segment N ⫽ 81 Spontaneous labor Induction and/or
1992; Unites cesarean delivery ⫽ 56 augmentation
States (fair) attempting trial of labor, ⫽ 25
medically indicated Vaginal delivery rate 18
delivery, an unfavorable Uterine rupture 0 0
cervix and a singleton,
vertex fetus Maternal infection 3
Maternal death 0 0
Infant death 0 0
Flamm et al,9 One or more prior low- N ⫽ 3,957 Induction and/or Elective repeat
1990; United transverse cesarean; augmentation ⫽ cesarean ⫽
States (fair) unknown scar allowed 1,201 2,756
Vaginal delivery rate 831
Uterine rupture 7 4
Maternal death 2
Infant death 1
Meehan et al,27 One prior cesarean without N ⫽ 719 Spontaneous labor Induction and/or Elective repeat
1989; Ireland a recurring indication for ⫽ 162 augmentation cesarean ⫽
(fair) cesarean ⫽ 127 430
Uterine rupture 0 1 0
Maternal death 0 0 0
Infant death 0 0 0
(continued)

delivered vaginally.8 Women attending tertiary care attending regional or community hospitals. The authors
hospitals were at least twice as likely to choose a trial of did not separately evaluate vaginal delivery rates for
labor and more likely to deliver vaginally than women women requiring medical augmentation or induction

VOL. 103, NO. 3, MARCH 2004 Guise et al VBAC Safety: Systematic Review 423
Table 1. Health Outcomes of Vaginal Birth After Cesarean Delivery (continued)
Study (quality) Population Reported outcomes
18
Phelan et al, Trial of labor 1982–3 1 N ⫽ 2,110 Spontaneous labor Elective repeat
1987; United prior cesarean 1983–4 (induction and/or cesarean ⫽
States (fair) 1–2 prior cesarean; low- augmentation) ⫽ 314
vertical, transverse, 1,796
unknown, allowed; 2nd Vaginal delivery rate 1,465
year more than 1 allowed Uterine rupture 34 dehiscence, 5 7 dehiscence or
rupture rupture
Maternal bleeding 5 Not reported
Maternal infection 53 successful 56
VBAC, 106 failed
VBAC
Maternal death Overall: 1
Infant death Overall: 17 fetal, 23 neonatal deaths; 11
weighing ⬍ 750 gm, 14 congenital
anomalies, 6 preterm
Stovall et al,11 Women with 1 or more N ⫽ 272 Spontaneous labor Induction and/or
1987; United prior low-vertical or low- ⫽ 139 augmentation
States (fair) transverse cesareans ⫽ 133
offered trial of labor, Vaginal delivery rate 116 98
unless medically Uterine rupture Overall: 1
contraindicated
Maternal death Overall: 0
Infant death Overall: 0
Paul et al,15 One prior low-vertical or N ⫽ 889 Spontaneous labor Induction ⫽ 32; Elective repeat
1985; United low-transverse cesarean; ⫽ 443 augmentation cesarean ⫽
States; unknown scar allowed ⫽257 157
overlapping Vaginal delivery rate 395 23; 177
data with Uterine rupture Spontaneous labor (induction and/or 4
Phelan et augmentation) ⫽ 11 successful VBAC,
al,18 1987 5 failed VBAC
(fair)
Maternal bleeding Spontaneous labor (induction and/or 5
augmentation) ⫽ 0 successful VBAC,
2 failed VBAC
Maternal infection Spontaneous labor (induction and/or 23
augmentation) ⫽ 14 successful VBAC,
37 failed VBAC
Maternal death Overall: 0
Infant death Overall: 7 fetal deaths (6 antepartum, 1 intrapartum), 9
neonatal deaths
Martin et al,19 One or more prior low- N ⫽ 717 Spontaneous labor Elective repeat
1983; United transverse or low-vertical (induction and/or cesarean ⫽
States (fair) cesarean augmentation) ⫽ 555
162
Vaginal delivery rate 101
Infant death 3 fetal and 0 3 fetal and 5
neonatal in neonatal
successful
VBAC, 1 fetal
and neonatal in
failed VBAC
Meier et al,23 Women attempting trial of N ⫽ 269 Spontaneous labor Elective repeat
1982; United labor had 1 or more (induction and/or cesarean ⫽ 62
States (fair) prior low-transverse augmentation) ⫽
cesarean with no 207
“obvious” cephalopelvic Uterine rupture 0 successful VBAC, 1
disproportion; the first 6 1 failed VBAC
undergoing elective Maternal bleeding Not reported 1
repeat with 1 prior
cesarean each month Maternal infection 2 11
were assessed Maternal death 0 0
Infant death 1 0
(continued)

from women who did not require medical assistance in women attempting trial of labor. These studies were
labor. largely conducted in university or tertiary care settings.
Eight prospective cohort studies provide additional The range across all studies was 60 – 82%, with a pooled
information regarding rates of vaginal delivery for rate of 75.9% (95% confidence interval [CI] 69.9%,

424 Guise et al VBAC Safety: Systematic Review OBSTETRICS & GYNECOLOGY

Table 1. Health Outcomes of Vaginal Birth After Cesarean Delivery (continued)
Study (quality) Population Reported outcomes
Retrospective cohorts
Raynor,12 1993; One or more prior low- N ⫽ 67 Spontaneous labor Induction and/or Elective repeat
United States transverse cesarean; ⫽ 26 augmentation cesarean ⫽
(fair) unknown scar allowed (2 ⫽ 25 16
verticals allowed) Vaginal delivery rate 17 14
Maternal bleeding Spontaneous labor (induction and/or
augmentation): 4 hemorrhages (2
required transfusions)
Infant death One at 28 wk,
spontaneous
rupture of
membranes,
polycystic
kidneys
Lao and One prior lower segment N ⫽ 666 Spontaneous labor Induction and/or Induction and/
Leung,14 cesarean ⫽ 529 augmentation or
1987; Hong (amniotomy ⫹ augmentation
Kong (fair) oxytocin) ⫽ (amniotomy
102 only) ⫽ 35
Vaginal delivery rate 436 26 86
Maternal bleeding 6 0
Maternal infection Not reported Not reported
Maternal death Not reported 0 0
Infant death 0 1
FHR⫽fetal heart rate; VBAC⫽vaginal birth after cesarean; UR ⫽ uterine rupture.

81.5%). On average, 80% (range 65– 89%) of women Although several studies provided information con-
with spontaneous onset of labor delivered vaginally, cerning hysterectomy, none specifically documented the
versus 68% (range 56 – 82%) of women who received indication for hysterectomy. The best evidence comes
oxytocin (not statistically significant).9 –16 There were from the Nova Scotia study8 that found no difference in
insufficient data to evaluate whether the dose of oxytocin hysterectomy rates in trial of labor (0.2%) versus elective
affected the likelihood of delivering vaginally. repeat cesarean (0.2%). Unlike the 2 prospective cohort
Maternal complications investigated included major studies reporting this outcome,15,16 McMahon et al8
maternal hemorrhage (requiring transfusion or hysterec- attempted to exclude “elective” repeat cesarean for med-
tomy), maternal infection (as manifested by endomyo- ical or obstetric indications such as placenta previa. The
metritis, wound infection, or postpartum/puerperal fe- 2 prospective cohort studies reported higher hysterec-
ver), uterine rupture, pelvic floor disturbance or tomy rates in repeat cesarean: 0.12% trial of labor com-
incontinence, and maternal death. Infant outcomes in- pared with 0.27% elective repeat cesarean16 and 0.27%
cluded infection, respiratory and neurologic sequelae, trial of labor compared with 3.2% in elective repeat
and perinatal death. cesarean.15 These provide weaker evidence because the
Ten fair-or good-quality observational studies com- cesarean group may have included women who had an
pared rates of maternal or infant complications with a indication for cesarean (ie, not truly elective cesarean)
trial of labor versus elective repeat cesarean.8 –11,15–21 and would not have been candidates for a trial of labor.
Two of these were large, retrospective, population-based In fact, in the latter study, Paul et al15 mention that only
studies.8,21 The other 8 were prospective cohort studies: 62 of the 157 patients in the “elective” repeat cesarean
3 large multicenter studies,9,16,17 1 large single-institution group were considered to be eligible for trial of labor.
study,15,18 1 small multicenter study,19 and 3 small sin- Thus, it is possible that the higher rates of hysterectomy
gle-institution studies.10,11,20 Only 3 of these studies9 –11 are attributable to medical or obstetric conditions (eg,
looked separately at the effect of oxytocin when used for hemorrhage secondary to placenta previa). Hysterec-
augmentation or induction within this larger population. tomy rates, which were reported in only 1 induction
Two good-quality cohort studies8,16 provided infor- study, were 0.2% in patients with induced trial of labor
mation concerning both transfusion and hysterectomy and 0.08% in patients with spontaneous labor.18
rates. Rates of maternal hemorrhage requiring transfu- Studies reporting maternal infection rates are of lim-
sion were 1.1% in the trial-of-labor group versus 1.3% for ited clinical applicability owing to a lack of explicit defi-
repeat cesarean in the large population-based study (not nitions or to the inclusion of many sources of infection.
statistically significant)8 and 0.72% versus 1.72% for the Two studies8,19 defined infection clearly and compared
prospective cohort study (P ⫽ .001).16 the incidence in trial-of-labor and elective repeat cesar-

VOL. 103, NO. 3, MARCH 2004 Guise et al VBAC Safety: Systematic Review 425
ean groups. Both studies combined puerperal infection of women with 1 prior low-transverse cesarean who
and abdominal wound infection within their definition of attempted trial of labor or repeat cesarean. However, no
infection. In the larger study,8 the rates were 5.3% for all details were provided on these deaths (eg, whether in-
women who attempted a trial of labor compared with fants with lethal anomalies were included), so it is not
6.4% for elective repeat cesarean. Among women who possible to determine whether these deaths were attrib-
attempted a trial of labor, those who did not delivery utable to mode of delivery or other etiologies. A more
vaginally (eg, failed trial of labor), had significantly recent population-based study from Scotland21 excluded
higher infection rates than women who were able to all perinatal deaths associated with lethal anomalies and
deliver vaginally (8% for failed trial of labor versus 3.5% medical conditions; however, the investigators did not
for successful trial of labor). This finding was reported clearly distinguish between trial of labor and elective
consistently among prospective cohort studies that per- repeat cesarean (ie, all emergent vaginal and cesarean
formed similar subgroup analyses (11–30% increased deliveries were classified as trial of labor). There were 20
risk of infection for failed trial of labor).15,18,20 The other deaths in 15,515 trials of labor for rate of perinatal death
study, a fair-quality prospective cohort,19 reported ma- of 12.9 per 10,000 (95% CI 7.9, 19.9) compared with 1 in
ternal infection rates of 6.79% in trial of labor compared 9,014 repeat cesareans for a pooled rate of 1.1 per 10,000
with 9.73% in elective repeat cesarean. Only 1 study of (95% CI 0.0, 6.1). Thirteen studies of induction agents
induction agents evaluated this outcome and found no reported perinatal death. Of these studies, 10 found no
infections in the induced group and 5% infection rate in deaths in any group studied.9 –12,18,22,24 –27 In the other 3
the spontaneous-labor group.22 studies, no consistent pattern emerged favoring sponta-
Six studies examined maternal death rates. The large neous or induced labor.13,14,21
population-based study of 6,138 women found no ma-
ternal deaths in either trial-of-labor or elective repeat
cesarean groups.8 In 5 prospective cohort studies involv- DISCUSSION
ing approximately 19,000 patients, there were 2 deaths The rate of cesarean delivery continues to rise, with
among women having a trial of labor and 2 among more than 1 million cesareans reported in 2002.2 In
women having a repeat cesarean.9,11,16 –18 No maternal contrast, rates of VBAC have experienced a 55% de-
deaths were mentioned in any studies of induction of crease since 1996. With almost 1 in 5 women (18%)
labor (n ⫽ 7,525). without a history of cesarean delivering by primary
One of the greatest concerns for patients, providers, cesarean in 2002, a 23% increase since 2001,2 women are
hospitals, and policymakers regarding VBAC is the po- increasingly facing questions regarding route of delivery
tential for uterine rupture and possible resultant devas- for future pregnancies. To make an informed decision,
tating outcomes. Terminology and definitions vary in women and clinicians need reliable estimates of the risks
usage among studies reporting uterine rupture. De- and benefits attributable to each delivery method.
tails of definitions, classification scheme, and findings of Although there is no direct evidence for the relative
individual studies can be found in a separate report.3 benefits and harms of VBAC, several cohort studies
Nine fair-to-good observational studies provide the best provide indirect evidence. The questions on the mind of
evidence for the frequency of symptomatic uterine rup- clinicians are what are the risks and how reliable are the
ture of the cesarean scar.8,9,11,16 –20,23 The pooled rate of data. Table 2 summarizes the results of our evidence
symptomatic uterine ruptures was greater for trial of review and provides an assessment of reliability of the
labor than for elective repeat cesarean (2.7 of 1,000; 95% data regarding the safety of VBAC. Reliability depends
CI 0.73, 4.7). not only on the precision of the estimate but also on the
Perinatal death was the only infant health outcome suitability of the study design to address the study ques-
measured among the 20 included studies. Although no tion and other factors that impact external validity.
study measured infant death relative to a mother’s choice Highly reliable data were only available for 1 outcome
of trial of labor versus repeat cesarean, 2 large, popula- studied: vaginal delivery rates. Regarding what is com-
tion-based studies attempted to measure whether trial of monly referred to as VBAC success rates, we found that
labor poses increased risk of infant death compared with more than 3 of 5 women who attempt trial of labor will
elective repeat cesarean.8,21 Each has important have a vaginal delivery. These data come from cohort
strengths and limitations. One study8 (n ⫽ 6,138) re- studies of fair-to-good quality, which provide good evi-
ported perinatal death rates of 9 per 1,000 in the trial-of- dence. However, there were no high-quality data to
labor group versus 5 per 1,000 in the repeat cesarean determine the effect of medical induction or augmenta-
group for women with 1 prior cesarean. The strength of tion of labor. With more than 1 in 5 births being the
this study was its ability to identify a conceptual cohort product of induced labor (more than double since 1989),

426 Guise et al VBAC Safety: Systematic Review OBSTETRICS & GYNECOLOGY

Table 2. Summary of Evidence
Suitability
Study of study Level of
Outcome type* Quality of evidence design† Estimate certainty
Vaginal delivery II-2 Fair-Good: several large prospective Greatest General population: High
rate and retrospective studies; mostly 60.4%
consistent findings.
Tertiary/University:
75.9% (95% CI 69.5,
81.5)
Effect of induction/ Low
augmentation: —
Maternal death II-2 Fair: studies consistently found no Least NS Low
increased risk of maternal death
Hysterectomy Fair-Poor: many studies failed to Moderate NS Low
report indication for
hysterectomy
Transfusion Fair: 2 studies with consistent Moderate NS Low
findings of slightly increased risk
for transfusion in trial of labor
although not significant in one
Uterine rupture Fair-Poor: several large cohort Moderate Additional risk for trial Medium
studies using inconsistent of labor: 2.7/1,000
terminology; many with (95% CI, 0.73, 4.7/
consistent findings of increased 1,000)
risk of symptomatic uterine
rupture in trial of labor
Infection Poor: definitions were inconsistent Moderate 5.3–6.7% trial of labor; Low
among studies 6.4–9.73% repeat
cesarean
Incontinence/ No studies Moderate — Low
Pelvic floor
Perinatal death Poor: most studies found increased Least NS Low
risk of perinatal death for trial of
labor, but the magnitude of the
increase varied greatly
Neurologic Poor: few studies of poor quality. Least — Low
impairment
Respiratory No studies Moderate — Medium
impairment
CI ⫽ confidence interval.
* Study design categories—I: randomized, controlled trials; II-1: controlled trials without randomization; II-2: cohort or case-control; II-3: multiple
time series; III: opinions, descriptive epidemiology. (U.S. Preventive Services Task Force 关1996兴).
†
Suitability of study design categories—Greatest: For comparison studies: concurrent comparison groups and prospective measurement of
exposure and outcome; For rates: population-based or multicenter prospective cohort studies. Moderate: All retrospective designs or multiple
before or after measurements but no concurrent comparison group; Least: Single before and after measurements and no concurrent comparison
group or exposure and outcome measured in a single group at the same point in time.5

further evaluation of the safety of induction of labor, regarding the reliability of the data for these outcomes
particularly in women with prior cesarean, is crucial. because of inadequate or variable outcome definition,
Data regarding uterine rupture, showing less than 1% few studies, poor study quality, or suitability to the
increased risk of symptomatic uterine rupture, were outcome.
moderately reliable. Data from several large prospective What remains unanswered is what the relative in-
cohort studies generally agreed, although studies varied creased risk is for a patient choosing a trial of labor
in their reporting of rupture. There was insufficient compared with repeat cesarean, regardless of the ulti-
evidence regarding maternal pelvic floor protection or mate delivery route. This is the essence of intention to
disturbance, and neurologic or respiratory effect for the treat for randomized controlled trials or inception cohort
infant. The evidence for the remaining 6 outcomes indi- for cohort studies. Ultimately, some women will start
cates that neither route of delivery suffers from dramat- labor although they had intended to have a cesarean
ically high rates of complications. There are concerns without undergoing labor. Other women who intended

VOL. 103, NO. 3, MARCH 2004 Guise et al VBAC Safety: Systematic Review 427
VBAC may elect for cesarean once labor begins. This ments for reporting the uterine rupture–related maternal
level of detail, although difficult to measure accurately, or infant morbidity and mortality. Similar issues of def-
would provide the clinician with a deeper level of evi- initional inconsistencies exist for other important out-
dence to counsel women should they present in labor. comes such as infection, hemorrhage, and infant morbid-
Only 2 studies8,15 attempted to measure intention, with ity. Second, investigators need to be consistent in the
only 1 study15 clearly identifying groups of women definition of their population (conceptual cohort), ensur-
intending one route and delivering by another. ing that the elective cesarean group would have been
Given the large number of women affected and the eligible for a trial of labor and that intended route of
implications for length of hospital stay and costs, meth- delivery is recorded. Lack of this impairs the compara-
odological deficiencies in the literature evaluating the bility of outcomes for the groups. Lastly, studies should
relative safety of VBAC compared with repeat cesarean be careful to specify use of interventions such as induc-
are striking. Comparisons across studies were hampered tion or augmentation of labor and the indications for
by lack of standards for reporting severity of disease or their use so that we can understand the effects of this
condition, and inconsistencies in definitions of outcomes. increasingly common intervention. Further research is
Studies often did not pay close attention to comparability also needed on the long-term health outcomes such as
of groups; specifically, the elective repeat cesarean group pelvic floor dysfunction, incontinence, or the long-term
was often not ensured to be otherwise eligible for trial of repercussions of neonatal conditions such as neurologic
labor. Other factors such as intention for delivery, parity, and respiratory conditions.
and type and number of previous cesarean were often In conclusion, the ability of patients, clinicians, and
not considered. Investigators did not pay close attention policymakers to determine the safety of VBAC is ham-
to and account for the importance of cointerventions pered by the limitations in the literature. With safety
such as use of oxytocin and other medical agents for being a national priority, rectifying these limitations
augmentation or induction of labor. Variations in report- should be a national research priority.
ing of important clinical outcomes such as hysterectomy,
infection, maternal mortality, uterine rupture, and peri-
natal mortality made it difficult to determine true prob- REFERENCES
ability of outcomes, potential preventive measures, or 1. Jiang H, Elixhauser A, Nicholas J, Steiner C, Reyes C.
outcomes that were directly attributable to route of de- Care of women in US hospitals, 2000. HCUP fact book
livery or labor management. Lack of precision made it no. 3; AHRQ publication no. 02– 0044. Rockville (MD):
difficult to determine whether the rates truly represented Agency for Healthcare Research and Quality; 2002.
risk of clinically significant outcomes or significant mis- 2. Hamilton BE, Martin JA, Sutton PD; U.S. Department of
classification or confounding. Health and Human Services Centers for Disease Control
Based on the results of our review, we have several and Prevention. Births: preliminary data for 2002. Natl
recommendations for improving the ability to interpret Vital Stat Rep 2003;51:1–20.
the results of future studies comparing VBAC with 3. Guise J, McDonagh M, Hashima J, Kraemer D, Eden K,
elective repeat cesarean. First, development of standard- Berlin M, et al. Vaginal birth after cesarean (VBAC).
ized reporting measures for disease severity and out- Evidence report/technology assessment (Prepared by the
comes of delivery is necessary to provide accurate infor- Oregon Health & Science University Evidence-based Prac-
mation to patients and providers. This is particularly tice Center under Contract No. 290-97-0018). Rockville
(MD): Agency for Healthcare Research and Quality Pub-
true for the outcome of considerable interest, uterine
lication; 2003.
rupture. To that end, we held a uterine rupture terminol-
4. Cesarean childbirth: report of a consensus development
ogy conference with national experts in obstetrics and
conference sponsored by the NICHHD in conjunction
epidemiology on September 5, 2002. Participants gener-
with the National Center for Health Care Technology and
ally agreed that the diagnosis of uterine rupture should assisted by the Office for Medical Applications of
be based on anatomical findings. For example, the term Research. NIH Publication. No. 82–2067. Washington,
“incomplete uterine rupture of a cesarean scar” should DC: National Institutes of Health; 1981.
be used to refer to uterine wall separation that was not 5. Khan K, Riet G, Popay J, Nixon J, Kleijnen J. Study quality
completely through all layers of the uterine wall (eg, assessment. Undertaking systematic reviews of research
serosa intact), and “complete uterine rupture of a cesar- effectiveness: CRD’s guidance for those carrying out or
ean scar” should refer to the separation of the entire commissioning reviews. Stage II: phase 5. York (UK):
thickness of the uterine wall with or without expulsion of NHS Centre for Reviews and Dissemination, University
the fetal-placental unit. Further discussions are needed to of York; 2001.
establish standard outcomes definitions and require- 6. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson

428 Guise et al VBAC Safety: Systematic Review OBSTETRICS & GYNECOLOGY

 GD, Rennie D, et al. Meta-analysis of observational stud- 19. Martin JN, Harris BA, Huddleston JF, Morrison JC,
ies in epidemiology: a proposal for reporting: Meta-analy- Propst MG, Wiser WL, et al. Vaginal delivery following
sis Of Observational Studies in Epidemiology (MOOSE) previous cesarean birth. Am J Obstet Gynecol 1983;146:
group. JAMA 2000;283:2008 –12. 255– 63.
7. Whitehead A. Meta-analysis of controlled clinical trials. 20. Duff P, Southmayd K, Read JA. Outcome of trial of labor
West Sussex (England): John Wiley & Sons; 2002. in patients with a single previous low transverse cesarean
8. McMahon MJ, Luther ER, Bowes WA, Olshan AF. Com- section for dystocia. Obstet Gynecol 1988;71(3 Pt
parison of a trial of labor with an elective second cesarean 1):380 – 4.
section. N Engl J Med 1996;335:689 –95. 21. Smith GCS, Pell JP, Cameron AD, Dobbie R. Risk of
9. Flamm BL, Newman LA, Thomas SJ, Fallon D, Yoshida perinatal death associated with labor after previous cesar-
MM. Vaginal birth after cesarean delivery: results of a ean delivery in uncomplicated term pregnancies. JAMA
5-year multicenter collaborative study. Obstet Gynecol 2002;287:2684 –90.
1990;76(5 Pt 1):750 – 4. 22. Blanco JD, Collins M, Willis D, Prien S. Prostaglandin E2
10. Cowan RK, Kinch RA, Ellis B, Anderson R. Trial of labor gel induction of patients with a prior low transverse cesar-
following cesarean delivery. Obstet Gynecol 1994;83: ean section. Am J Perinatol 1992;9:80 –3.
933– 6. 23. Meier PR, Porreco RP. Trial of labor following cesarean
11. Stovall TG, Shaver DC, Solomon SK, Anderson GD. section: a two-year experience. Am J Obstet Gynecol 1982;
Trial of labor in previous cesarean section patients, exclud- 144:671– 8.
ing classical cesarean sections. Obstet Gynecol 1987;70: 24. Rayburn WF, Gittens LN, Lucas MJ, Gall SA, Martin ME.
713–7. Weekly administration of prostaglandin E2 gel compared
12. Raynor BD. The experience with vaginal birth after cesar- with expectant management in women with previous
ean delivery in a small rural community practice. Am J cesareans. Obstet Gynecol 1999;94:250 – 4.
Obstet Gynecol 1993;168(1 Pt 1):60 –2. 25. Lelaidier C, Baton C, Benifla JL, Fernandez H, Bourget P,
13. Flamm BL, Goings JR, Fuelberth NJ, Fischermann E, Frydman R. Mifepristone for labour induction after previ-
Jones C, Hersh E. Oxytocin during labor after previous ous caesarean section. Br J Obstet Gynaecol 1994;101:
cesarean section: results of a multicenter study. Obstet 501–3.
Gynecol 1987;70:709 –12. 26. Flamm BL, Anton D, Goings JR, Newman J. Prostaglan-
14. Lao TT, Leung BFH. Labor induction for planned vaginal din E2 for cervical ripening: a multicenter study of patients
delivery in patients with previous cesarean section. Acta with prior cesarean delivery. Am J Perinatol 1997;14:
Obstet Gynecol Scand 1987;66:413– 6. 157– 60.
15. Paul RH, Phelan JP, Yeh SY. Trial of labor in the patient 27. Meehan FP, Burke G. Trial of labour following prior
with a prior cesarean birth. Am J Obstet Gynecol 1985; section: a 5 year prospective study (1982–1987). Eur J
151:297–304. Obstet Gynecol Reprod Biol 1989;31:109 –17.
16. Flamm BL, Goings JR, Liu Y, Wolde-Tsadik G. Elective
repeat cesarean delivery versus trial of labor: a prospective Reprints are not available. Address correspondence to: Jeanne-
multicenter study. Obstet Gynecol 1994;83:927–32. Marie Guise, MD, MPH, Oregon Health & Science University,
17. Flamm BL, Lim OW, Jones C, Fallon D, Newman LA, Department of Obstetrics and Gynecology, UHN-50, 3181
Mantis JK. Vaginal birth after cesarean section: results of a SW Sam Jackson Park Road, Portland, OR 97239 –3098;
multicenter study. Am J Obstet Gynecol 1988;158: e-mail: [email protected].
1079 – 84.
18. Phelan JP, Clark SL, Diaz F, Paul RH. Vaginal birth after Received August 20, 2003. Received in revised form December 15,
cesarean. Am J Obstet Gynecol 1987;157:1510 –5. 2003. Accepted December 22, 2003.

VOL. 103, NO. 3, MARCH 2004 Guise et al VBAC Safety: Systematic Review 429