Article published online: 2022-12-01

Review

Trial of Labor after Three or More Previous Cesarean Sections:

Systematic Review and Meta-Analysis of Observational Studies

Authors
Arrigo Fruscalzo1,2 , Emma Rossetti3, Ambrogio P. Londero4, 5

Affiliations Supplementary material is available under https://

1 Department of Obstetrics and Gynecology, HFR Fribourg, doi.org/10.1055/a-1965-4125
Switzerland
2 Faculty of Medicine, University of Münster, Germany Abs trac t
3 Department of Obstetrics and Gynecology, Brixen
Aims To assess the success rate and prevalence of maternal or
General Hospital, Brixen, Italy

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

neonatal complications in women undergoing a trial of labor
4 Academic Unit of Obstetrics and Gynaecology; Depart-
after three or more ( ≥ 3) previous cesarean sections (CSs).
ment of Neuroscience, Rehabilitation, Ophthalmology,
Methods A systematic literature review and meta-analysis was
Genetics, Maternal and Infant Health, University of
conducted from inception to May 2022 in Medline, Scopus,
Genova, Italy
ENBASE, ClinicalTrials.gov, and the Cochrane Central Register
5 Ennergi Research (non-profit organization), 33050
of Controlled Trials and Reviews. Items detailing success rate
Lestizza, UD, Italy
and complications in women with a history of ≥ 3 previous CSs
were considered. Selected articles were evaluated for quality,
Key words
heterogeneity, and publication bias. A pooled prevalence or
trial of labor after cesarean section, uterine scar, uterine
odds ratio was calculated.
rupture, adverse neonatal outcome, adverse maternal
Findings  Twelve articles were included for a total of 540 wom-
outcome.
en with a history of ≥ 3 CSs, accounting for the 2 % (CI 95 %
1–4 %) of the whole cohort of trial of labor. Our findings show
received 23.05.2022
a 0.67 (CI 95 % 0.53–0.78) rate of successful vaginal delivery. A
accepted after revision 02.10.2022
higher success rate was observed in women having a history of
published online 01.12.2022
a prior vaginal delivery (0.90, CI 95 % 0.77–0.96) and when
Bibliography prostaglandins, peridural anesthesia or oxytocin were allowed
Z Geburtsh Neonatol 2023; 227: 96–105 (respectively 0.73, CI 95 % 0.62–0.83, 0,73, CI 95 % 0.57–0.85
DOI 10.1055/a-1965-4125 and 0.73, CI 95 % 0.64–0.81). Uterine rupture rate was 0.01 (CI
ISSN 0948-2393 95 % 0.00–0.01). No cases of fetal asphyxia or maternal or neo­
© 2022. Thieme. All rights reserved. natal death were registered.
Georg Thieme Verlag, Rüdigerstraße 14, Conclusions The success rate and low frequency of severe
70469 Stuttgart, Germany complications observed seem to support a trial of labor in se-
lected patients desiring a natural birth. However, a potential
Correspondence underestimation of serious maternal and neonatal complica-
Dr. Arrigo Fruscalzo tions should be considered in the decision-making process.
Clinic of Obstetrics and Gynaecology
HFR Fribourg
Ch. des Pensionnats 2-6
1708 Fribourg
Switzerland
arrigo.fruscalzo@h-fr.ch

Introduction ed coin: on the one hand, parents and their desire for a safe child-
The steady increase of cesarean section (CS) rate experienced in birth for both mother and newborn coupled with easy access to ad-
recent decades in developed and developing countries is an alarm- vanced health care systems in more developed countries; and the
ing issue that every obstetrician has confronted [1]. Several reasons other, the reduced preparedness to face complex situations during
can be put forth to explain this phenomenon, resembling a two-sid- labor and the fear of litigation of obstetric care providers [2–4].

96 Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved.
 Beside the risk of a drift toward a “trivialization of cesarean sec- event as a symptomatic clinical rupture confirmed at the time of ce-
tion,” the desire to give birth naturally can be more deeply seated sarean section. Incidental findings were not considered. Information
in women than expected [5]. These are the cases of women desir- from the full text articles reporting study time frame and geograph-
ing to give birth even in exceptional circumstances, like after three ic locations were noted to avoid any potential overlap of the popu-
or more previous CSs. Counseling in this situation can be particu- lations. Where there were multiple publications from the same group
larly challenging owing to the rarity of this request in clinical prac- of women or presenting partly overlapping data, the most complete
tice and the lack of evidence on this topic. Nonetheless, this topic and detailed results were used or both articles were considered if
should be professionally addressed, as our decisions have a medi- they described valuable different aspects of the same study. The fol-
cal and psychological impact on women later in life [6, 7]. lowing were the study exclusion criteria: studies considering only
To cover this lack of knowledge we performed a systematic re- trial of labor after one or two previous cesarean sections; case re-
view and meta-analysis of the literature. The primary aim was to ports, case series and studies including ≤ 3 cases; inability to retrieve
assess the success rate of trial of labor (TOL) in women with a pre- information about trial of labor; reviews; conference abstracts; let-
vious history of three or more cesarean sections. The secondary ters to the editor; editorials; studies involving nonhuman subjects;
aims were to assess the prevalence of maternal and neonatal com- or articles published in a language other than English or German. As
plications as well to compare success and complication rates among previously described, any disagreement relating to inclusion of stud-
different subgroups. Details about study aims are reported in Sup- ies, data extraction, or quality assessment between the reviewers

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

plemental Table 1. was resolved by re-examination of the original article until consen-
sus was obtained [10, 11].

Materials And Methods Data extraction

Using a data extraction form, two independent reviewers (AF and
Literature search APL) extracted predefined data from the eligible studies encom-
Using a standardized approach to aid our literature search strate- passing the following information: authors; year of publication;
gy, a search was independently performed by three authors (AF, study design; geographical location; type of previous uterine scar
APL, and ER) and a systematic review was undertaken in accord- (low transverse, longitudinal or classical, T shape, or unknown); use
ance with appropriate guidelines [8, 9]. Medline, Scopus, ENBASE, of oxytocin; medical labor induction with prostaglandins; peridur-
and Cochrane Central Register of Controlled Trials and Reviews al analgesia (PDA); total number of trial of labors; number of trial
were searched using a combination of specific search terms from of labors subdivided by history of one, two, and three or more pre-
inception to April 27, 2022. The following terms were considered: vious cesarean sections; number of events (success rate, uterine
“trial of labor OR vaginal birth OR vaginal delivery” AND “one or rupture, hysterectomy, need for blood transfusion, maternal death,
more OR two or more OR more than one OR more than two OR neonatal intensive care unit admission, perinatal asphyxia, perina-
three OR thrice OR third OR multiple” AND “caesarean OR cesar- tal injury, or perinatal death) subdivided by history of one, two, and
ean” (query details are reported in Supplemental Table 2). To iden- three or more previous cesarean sections.
tify eligible studies, meta-information, titles, and abstracts result-
ing from the search strategy were screened independently by three Quality assessment
reviewers (AF, APL, and ER). Full texts of the remaining items were The methodological quality of included studies was independent-
assessed to identify studies reporting the outcomes of interest. Fi- ly assessed by two authors (AF and APL) according to the Newcas-
nally, relevant references cited in these articles and in identified re- tle-Ottawa Scale (NOS), as previously described, which ranges from
view articles were also reviewed to find any other relevant articles. 0 to 9 points [11]. The NOS permits the analysis of selection, com-
parability, and outcomes (cohort studies) or exposure issues
Inclusion and exclusion criteria (case-control studies). As previously stated, the following defini-
Papers were selected for review if they satisfied the following inclu- tions were adopted: 9–8 points on the NOS = high-quality studies,
sion criteria: studies including women performing a trial of labor with 7–6 points = medium-quality, and < 6 points = low quality [11].
a history of previous three or more cesarean sections; randomized
controlled trials; prospective or retrospective cohort studies; nest- Data analysis
ed case-control studies; population-based case-control studies; ar- The software R (Version 4.2.0; R Foundation, Vienna, Austria) [12]
ticles written in English, French, Spanish, or German; and availability was used for statistical evaluation and the p-value < 0.05 was con-
of the full text. Authors of the screened studies were not contacted sidered significant. A summary statistic was computed that sum-
for full text versions or missing data. Eligible studies also had to pres- marizes success rate and events prevalence or odds ratio for the
ent specific data in women with a history of previous three or more comparison between groups. In addition, the Egger’s regression
cesarean sections (specific number of events and total number at test of funnel plot asymmetry and the rank correlation test by Begg
risk) about one of the following outcomes: trial of labor success rate and Mazumdar were used to assess publication bias [13–15]. As
(e. g., number of spontaneous or operative vagi­nal deliveries), uter- previously described, heterogeneity between studies was evaluat-
ine rupture, hysterectomy, need for blood transfusion, maternal ed by I-square index (heterogeneous with a value above 50 %), and
death, neonatal intensive care unit admission, perinatal asphyxia (ar- the Cochran Q (heterogeneous with a p-value below 0.100) and
terial cord blood pH < 7.0), perinatal injury, and perinatal death. A fixed- or the random-effects model were used as appropriate
consistent definition of uterine rupture was adopted, defining the [16, 17]. The primary outcome was presented as a pooled propor-

Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved. 97
 Review

tion with a 95 % confidence interval (CI), where arcsine transforma- ther five. Furthermore, classic or T-shaped scars were admitted to
tion was used in case of low-prevalence events, and the secondary TOL only in three studies, while this was unknown or not specified
outcomes were presented as pooled proportion or pooled odds in a further four.
ratio (OR) with 95 % CI. A priori subgroup analysis was performed
for the type of study, induction with prostaglandins, use of oxy- Quality assessment of the included studies
tocin, and PDA. In addition, a sensitivity analysis was performed to The quality of the included literature showed some variation most-
check the robustness of the pooled results by removing each item, ly due to the older studies and to the different type of considered
one by one. Meta-regression was also performed. The MOOSE (Me- studies (prospective or retrospective). In addition, considering the
ta-analysis Of Observational Studies in Epidemiology) guidelines Centre for Evidence-Based Medicine (CEBM) levels, the included
for accurately performing meta-analysis of observational studies studies could be classified as IIB-IV (mostly low quality) [30]. Ac-
[9] and PRISMA (Preferred Reporting Items for Systematic Reviews cording to the NOS score, four studies were graded 6–7 (medium
and Meta-Analyses) guidelines checklist [8] where considered to quality) and eight studies 8–9 (high quality) while the median NOS
plan and perform the present systematic review and meta-analy- value was 8 (IQR 7–8). Most of the included studies lost one or two
sis. The study was registered in PROSPERO, the international pro- points in the comparability section of the NOS because they do not
spective register of systematic reviews (date of registration: May correct for labor induction or other factors (e. g., time since the
13, 2022; registration number: CRD42022329790; title of the trial: previous CS, previous CS feto-pelvic disproportion, etc.).

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

Trial of labor after three or more previous cesarean sections). This
study is exempted from ethical approval since all human data used Risk of bias assessment
here is already published in an anonymized fashion. The wide time span and the heterogeneity of labor management
between the studies is the main limit of this analysis. In fact, oxy-
tocin use, prostaglandins, and PDA were clearly allowed only in
Results some studies. Also, previous classic or T-shaped uterine scars were
allowed only in some studies. All these factors could affect the prev-
Study selection alence of successful vaginal birth. Also, the retrospective study de-
In total 13,899 items were found and after duplicate removal the sign can generate confounding, selection, and information bias. To
remaining 7,531 items were further screened (▶ Fig. 1a). Then, allow an accurate interpretation of the results and gain better in-
after reviewing titles and abstracts, 7,351 items were removed be- sight into the actual knowledge about TOL in women with a histo-
cause they were irrelevant/not pertinent, and 8 items were re- ry of three or more previous cesarean sections, separate subgroup
moved because the full text was not available. Thus, 180 potential- analyses were performed.
ly relevant articles were retrieved for comprehensive review.
Among these 180 articles, 168 were excluded because they were Successful vaginal delivery and uterine rupture
not pertinent, required data were unavailable, or for other reasons prevalence
(e. g., non-English/German literature, case reports, etc.) (▶Fig. 5a Considering the primary outcome of all 540 women with a history
and Supplemental List 1). A list of the 168 excluded studies with of three or more previous CSs, the pooled successful vaginal deliv-
reason for exclusion and the list of the 12 included studies [18–29] ery rate was 0.67 (95 % CI 0.53, 0.78) (▶Fig. 2). Among them, 513
is provided in Supplemental Material (Supplemental List 1). had three previous CSs with an overall success rate of 0.62 (95 % CI
0.44, 0.77) (for two studies data about successful vaginal delivery
Characteristics of the studies considering only three previous CSs is missing), while 27 women had
All the included papers were published before 2016 comprising a four CSs (data present in four studies) with a success rate of 0.55
55-year span of patient recruitment, from 1960 to 2015. Most stud- (95 % CI 0.37, 0.72). Data about women undergoing a TOL after three
ies could be considered retrospective cohort studies: in nine cases or more CSs having a history of a prior vaginal delivery were found
they were clearly stated as retrospective studies [18, 19, 21–27] only in two studies, and the successful vaginal delivery rate in these
and in three cases the studies were stated to be prospective 42 women was 0.90 (CI 95 % 0.77, 0.96) [18, 19]. Overall, we found
[20, 28, 29]. Nine of the included studies were performed in the four cases of uterine rupture, for a pooled prevalence of 0.01 (95 %
USA. The total number of TOL in women with a history of previous CI 0.00, 0.01) (▶Fig. 3). In one of these cases, a hysterectomy with
three or more cesarean sections was 540 and the prevalence con- blood transfusion was required [26]. No case of maternal death and
sidering the whole cohort of TOL was 2 % (95 % CI 1 %, 4 %) but this no case of neonatal asphyxia or death were registered.
prevalence was lower in the most recent studies (1 %; 95 % CI 1 %, Considering the analysis of different labor management options,
1 %) than in the 1980s (3 %; 95 % CI 2 %, 5 %) and in the 1960s (10 %; the subgroups where prostaglandins, oxytocin, and PDA were
95 % CI 6 %, 15 %). A meta-regression was also performed, and a re- ­allowed presented the highest rate of successful vaginal delivery
duced prevalence of TOL in women with a history of previous three (▶ Figs. 2a, b, and Supplemental Figure 1A). The inclusion in the
or more cesarean sections was significantly negatively correlated study cohorts of classical or T-shaped scars or the possibility of
with the increase in years (coefficient –0.0583; 95 % CI –0.0789, making a trial labor only if in advanced labor at the time of admis-
–0.0377; p < 0.05). In ▶ Table 1 the main characteristics of the in- sion were associated with a lower prevalence of successful vaginal
cluded studies are reported. Oxytocin use was allowed in nine stud- delivery (▶Fig. 2c and Supplemental Figure 1B). Also, a lower rate
ies and not specified in one, whereas labor induction with prosta- of successful vaginal delivery was shown in prospective than in ret-
glandins was allowed only in five studies and not specified in a fur- rospective studies (0.56 vs. 0.74) (supplemental Figure 2A). The

98 Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved.
 a b Egger’s test p = 0.765
0.0
Records identified
through database
Identification

Standard Error
Additional records 0.5
searching
identified through
(PubMed n = 3 083;
other sources
Scopus n = 6 018; 1.0
(n = 4)
EMBASE n = 4 345;
Cochrane n = 619)
1.5
–2 –1 0 1 2 3 4
Transformed Proportion
Records after
duplicates removed c Egger’s test p = 0.526
(n = 7 531)

0.05
Screening

Standard Error
0.15
Records excluded*
Records screened (n = 7 351)

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

(n = 180) * of these 7 343 not pertinent 0.25
and 8 full text not available
– 0.4 – 0.2 0.0 0.2 0.4 0.6
Transformed Proportion
d Egger’s test p = 0.758
0.0
Full-text articles
Full text articles
Eligibility

excluded** (n = 168)
Standard Error
assessed for 0.5
** 140 not pertinent; 13 not
eligibility (n = 180)
possible to extract data about
trial of labor with a history
1.0
of three or more previous
cesarean sections; and 15
for other reasons (i.e., 1.5
non-English/-German/-French/-
Spanish literarure, duplicate 0.05 0.10 0.20 0.50 1.00 2.00 5.00 10.00
Studies included publication, or case reports) Odds Ratio
in qualitative e
synthesis (n = 12) Egger’s test p = 0.873
0.0
Included

Standard Error

0.5

1.0
Studies included in
quantitative synthesis 1.5
(n = 12)
0.05 0.10 0.20 0.50 1.00 2.00 5.00 10.00
Odds Ratio
f Egger’s test not assessed
g Egger’s test p = 0.602
0.0 0.0

0.5 0.5
Standard Error
Standard Error

1.0 1.0

1.5 1.5

0.1 0.2 0.5 1.0 2.0 5.0 10.0 20.0 50.0 0.05 0.10 0.20 0.50 1.00 2.00 5.00 20.00
Odds Ratio Odds Ratio

▶Fig. 1 Study flowchart and funnel plots. Panel a) Flow of studies through the selection process according to PRISMA guidelines. Panel b) Funnel
plot of successful vaginal delivery proportion meta-analysis in women with a history of previous three or more cesarean sections (Egger’s test p-val-
ue = 0.765). Panel c) Funnel plot of uterine rupture proportion meta-analysis in women with a history of previous three or more cesarean sections
(Egger’s test p-value 0.526, rank correlation test p < 0.05). Panel d) Funnel plot of odds ratio (OR) meta-analysis for successful vaginal delivery in
women with a history of previous three or more cesarean sections versus one previous cesarean section (Egger’s test p-value = 0.758). Panel e) Fun-
nel plot of odds ratio (OR) meta-analysis for successful vaginal delivery in women with a history of previous three or more cesarean sections versus
two previous cesarean sections (Egger’s test p-value = 0.873). Panel f) Funnel plot of odds ratio (OR) meta-analysis for uterine rupture in women with
a history of previous three or more cesarean sections versus one previous cesarean section (Egger’s test not assessed). Panel g) Funnel plot of odds
ratio (OR) meta-analysis for uterine rupture in women with a history of previous three or more cesarean sections versus two previous cesarean sec-
tions (Egger’s test p-value = 0.602).

Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved. 99
 Review

▶Table 1 Description of the included studies.

Study Cohort Centers Nation Study years Oxy- PG PDA Allowed classical TOL ( ≥ 3 TOL
type tocin or T-shaped scar previous CS) (total)

Vigorito 2016 [17] R 1 Italy 2011–2015 N N – – 10 1171

Cahill 2010 [18] R >2 USA 1996–2000 Y Y Y N 89 13706

Landon 2006 [19] P >2 USA 1999–2002 Y Y Y – 104 17898
Spaans 2003 [20] R 2 Netherlands 1988–1997 Y Y – – 4 59*
Emembolu 1998 [21] R – Nigeria 1998 N N N Y 39 139*
Miller 1994 [22] R 2 USA 1983–1992 Y Y – N 241 12707
Hansell 1990 [23] R 1 USA 1983–1987 Y – – N 6 35*
Novas 1989 [24] R 1 USA 1986–1987 Y – Y Y 9 180
Pruett 1988 [25] R 1 USA 1984–1986 Y Y – Y 4 55*

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

Stovall 1987 [26] R 2 USA 1986 Y – Y N 7 272
Martin 1983 [27] P 2 USA 1981–1982 Y – Y N 6 162
Riva 1961 [28] P 1 USA 1953–1960 – – – – 21 214

Acronyms: – = unknown value; N = no; P = prospective; PDA = peridural analgesia; PG = use of prostaglandins; R = retrospective; TOL = trial of labor;
USA = United States of America; Y = yes.; *It includes only TOL in women with a history of two or more previous cesarean sections.

increasing number of involved centers did not seem to affect suc- Publication bias
cessful vaginal delivery rate (supplemental Figure 2B). The potential publication bias was examined by the funnel plot
In ▶ Fig. 3 the subgroup analysis for the four cases of uterine method (effect size standard error plotted against transformed pro-
rupture is presented. All the cases of uterine rupture were regis- portion or OR) and by Egger’s or the rank correlation test (▶ Figs.
tered among the studies that allowed prostaglandin and oxytocin 1b–g). Nonetheless, results indicated that there was no significant
use (▶Figs. 3a, b). Furthermore, the prevalence of uterine rupture publication bias in the main outcome of this meta-analysis (p > 0.05
was higher in the studies that allowed the inclusion of classical or for successful vaginal delivery); in the prevalence of uterine rup-
T-shaped scars than in the studies that did not (0.05 vs. 0.01) (▶Fig. ture, a significant publication bias emerged in the rank correlation
3c). No cases of uterine rupture were registered among studies al- test (p < 0.05) but not in Egger’s test (p = 0.526). In any case this
lowing PDA use and none of the studies reporting data about uter- could suggest a possible shortage of studies with a high prevalence
ine rupture allowed TOL except if in advanced labor at the time of of events. Furthermore, no significant publication bias was ob-
admission (Supplemental Figures 3A and 3B). Supplemental Fig- served for the comparison between women with a history of pre-
ure 4A shows that all cases of uterine rupture were registered vious three or more CSs versus one or two previous CSs. However,
among the retrospective studies. No differences in uterine rupture it was possible to assess this difference only in a small number of
prevalence were observed when the studies were subgrouped by studies considered. Finally, all the other performed rank correla-
the number of involved centers. tion tests by Begg and Mazumdar were not significant.

Three or more versus one or two previous cesarean

sections Discussion
▶ Fig. 4 shows the pooled analysis of the differences between
women with a history of three or more and one or two previous ce- Key results
sarean sections among the same study cohort. The odds of suc- Our findings show a 0.67 success rate of vaginal delivery in women
cessful vaginal delivery were slightly but non-significantly lower in with a history of three or more previous CSs (CI 95 % 0.53, 0.78).
women with a history of three or more CSs than in women with a Success rate was influenced by the type of labor management,
history of one previous CS (OR 0.74; CI 95 % 0.44, 1.23), while al- demonstrating a higher success rate when oxytocin, prostaglan-
most no differences were observed compared to women with a his- dins, or PDA were allowed. The prevalence of uterine rupture was
tory of two previous CSs (OR 1.03; CI 95 % 0.83, 1.27). Also, there 0.01 (CI 95 % 0.00, 0.01), however this rate was possibly underes-
was a non-significantly higher risk of uterine rupture in women with timated owing to the low number of overall events reported. No
a history of three or more CSs than in women with a history of one cases of fetal asphyxia or maternal or neonatal death were regis-
(OR 2.43; CI 95 % 0.82, 7.19) or two (OR 0.95; CI 95 % 0.37, 2.45) tered. Furthermore, no significant differences in terms of success-
previous CSs, respectively (▶Figs. 4c, d). ful vaginal delivery or uterine rupture were found in the same co-
horts of patients between three or more CSs compared to one or
two previous CSs.

100 Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved.
 a a
Prostaglandins allowed Prostaglandins allowed
Study Success TOL Proportion CI 95 % Weight Study Rupture TOL Proportion CI 95 % Weight

Yes Yes
Cahill 2010 [18] 71 89 0.80 [0.70; 0.88] 13.2 % Cahill 2010 [18] 0 89 0.00 [0.00; 0.04] 22.4 %
Landon 2006 [19] 64 104 0.62 [0.51; 0.71] 13.8 % Landon 2006 [19] 0 4 0.00 [0.00; 0.60] 1.0 %
Spaans 2003 [20] 4 4 1.00 [0.40; 1.00] 3.1 % Spaans 2003 [20] 3 241 0.01 [0.00; 0.04] 60.7 %
Miller 1994 [22] 190 241 0.79 [0.73; 0.84] 14.1 % Miller 1994 [22] 1 4 0.25 [0.01; 0.81] 1.0 %
Pruett 1988 [25] 2 4 0.50 [0.07; 0.93] 5.5 % Fixed effect model 4 338 0.01 [0.00; 0.02] 85.1 %
Random effects model 331 442 0.73 [0.62; 0.83] 49.7 % Heterogeneity: I = 54 %, χ = 6.46 (p = 0.09)
Heterogeneity: I =72 %, χ = 14.29 (p < 0.01)
Unkown
No Hansell 1990 [23] 0 6 0.00 [0.00; 0.46] 1.5 %
Vigorito 2016 [17] 9 10 0.90 [0.55; 1.00] 5.2 % Novas 1989 [24] 0 9 0.00 [0.00; 0.34] 2.3 %
Emembolu 1998 [21] 11 39 0.28 [0.15; 0.45] 12.1 % Stovall 1987 [26] 0 7 0.00 [0.00; 0.41] 1.8 %
Random effects model 20 49 0.62 [0.07; 0.97] 17.3 % Martin1983 [27] 0 6 0.00 [0.00; 0.46] 1.5 %
Heterogeneity: I = 87 %, χ = 7.92 (p < 0.01) Riva 1961 [28] 0 21 0.00 [0.00; 0.16] 5.3 %
Fixed effect model 0 49 0.01 [0.00; 0.02] 12.3 %
Unkown Heterogeneity: I = 0 %, χ = 0 (p = 1.00)
Hansell 1990 [23] 4 6 0.67 [0.22; 0.96] 6.5 %
Novas 1989 [24] 8 9 0.89 [0.52; 1.00] 5.1 % No
Stovall 1987 [26] 7 7 1.00 [0.59; 1.00] 3.2 % Vigorito 2016 [17] 0 10 0.00 [0.00; 0.31] 2.5 %
Martin 1983 [27] 3 6 0.50 [0.12; 0.88] 7.0 % Fixed effect model 0 10 0.00 [0.00; 0.09] 2.5 %
Riva 1961 [28] 9 21 0.43 [0.22; 0.66] 11.1 % Heterogeneity: not applicable
Random effects model 31 49 0.65 [0.40; 0.84] 33.0 %
Heterogeneity: I = 48 %, χ = 7.64 (p = 0.11) Fixed effect model 4 397 0.01 [0.00; 0.01] 100.0 %
Heterogeneity: I = 0 %, χ = 7.94 (p = 0.54)
Random effects model 382 540 0.67 [0.53; 0.78] 100.0 % 0 0.2 0.4 0.6 0.8
Heterogeneity: I = 80 %, χ = 55.20 (p < 0.01) Proportion
0.2 0.4 0.6 0.8 1 b
Proportion Oxytocin allowed

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

b Study Rupture TOL Proportion CI 95 % Weight
Oxytocin allowed
Study Success TOL Proportion CI 95 % Weight Yes
Cahill 2010 [18] 0 89 0.00 [0.00; 0.04] 22.4 %
Yes Spaans 2003 [20] 0 4 0.00 [0.00; 0.60] 1.0 %
Cahill 2010 [18] 71 89 0.80 [0.70; 0.88] 13.2 % Miller 1994 [22] 3 241 0.01 [0.00; 0.04] 60.7 %
Landon 2006 [19] 64 104 0.62 [0.51; 0.71] 13.8 % Hansell 1990 [23] 0 6 0.00 [0.00; 0.46] 1.5 %
Spaans 2003 [20] 4 4 1.00 [0.40; 1.00] 3.1 % Novas 1989 [24] 0 9 0.00 [0.00; 0.34] 2.3 %
Miller 1994 [22] 190 241 0.79 [0.73; 0.84] 14.1 % Pruett 1988 [25] 1 4 0.25 [0.01; 0.81] 1.0 %
Hansell 1990 [23] 4 6 0.67 [0.22; 0.96] 6.5 % Stovall 1987 [26] 0 7 0.00 [0.00; 0.41] 1.8 %
Novas 1989 [24] 8 9 0.89 [0.52; 1.00] 5.1 % Martin 1983 [27] 0 6 0.00 [0.00; 0.46] 1.5 %
Pruett 1988 [25] 2 4 0.50 [0.07; 0.93] 5.5 % Fixed effect model 4 366 0.01 [0.00; 0.02] 92.2 %
Stovall 1987 [26] 7 7 1.00 [0.59; 1.00] 3.2 % Heterogeneity: I = 3 %, χ = 7.22 (p = 0.41)
Martin 1983 [27] 3 6 0.50 [0.12; 0.88] 7.0 %
Random effects model 353 470 0.73 [0.64; 0.81] 71.6 % Unkown
Heterogeneity: I = 56 %, χ = 18.35 (p = 0.02) Riva 1961 [28] 0 21 0.00 [0.00; 0.16] 5.3 %
Fixed effect model 0 21 0.00 [0.00; 0.05] 5.3 %
No Heterogeneity: not applicable
Vigorito 2016 [17] 9 10 0.90 [0.55; 1.00] 5.2 %
Emembolu 1998 [21] 11 39 0.28 [0.15; 0.45] 12.1 % No
Random effects model 20 49 0.62 [0.07; 0.97] 17.3 % Vigorito 2016 [17] 0 10 0.00 [0.00; 0.31] 2.5 %
Heterogeneity: I = 87 %, χ = 7.92 (p < 0.01) Fixed effect model 0 10 0.00 [0.00; 0.09] 2.5 %
Heterogeneity: not applicable
Unkown
Riva 1961 [28] 9 21 0.43 [0.22; 0.66] 11.1 % Fixed effect model 4 397 0.01 [0.00; 0.01] 100.0 %
Random effects model 9 21 0.43 [0.24; 0.64] 11.1 % Heterogeneity: I = 0 %, χ = 7.94 (p = 0.54)
Heterogeneity: not applicable 0 0.2 0.4 0.6 0.8
Proportion
Random effects model 382 540 0.67 [0.53; 0.78] 100.0 %
c
Heterogeneity: I = 80 %, χ = 55.20 (p < 0.01) Classic or T-shaped scars allowed
0.2 0.4 0.6 0.8 1
Study Rupture TOL Proportion CI 95 % Weight
Proportion
c Yes
Classic or T-shaped scars allowed Novas 1989 [24] 0 9 0.00 [0.00; 0.34] 2.3 %
Study Success TOL Proportion CI 95 % Weight Pruett 1988 [25] 1 4 0.25 [0.01; 0.81] 1.0 %
Random effects model 1 13 0.05 [0.00; 0.45] 3.3 %
Yes Heterogeneity: I = 67 %, χ = 3.04 (p = 0.08)
Emembolu 1998 [21] 11 39 0.28 [0.15; 0.45] 12.1 %
Novas 1989 [24] 8 9 0.89 [0.52; 1.00] 5.1 % No
Pruett 1988 [25] 2 4 0.50 [0.07; 0.93] 5.5 % Cahill 2010 [18] 0 89 0.00 [0.00; 0.04] 22.4 %
Random effects model 21 52 0.54 [0.17; 0.87] 22.8 % Miller 1994 [22] 3 241 0.01 [0.00; 0.04] 60.7 %
Heterogeneity: I = 74 %, χ = 7.61 (p = 0.02) Hansell 1990 [23] 0 6 0.00 [0.00; 0.46] 1.5 %
Stovall 1987 [26] 0 7 0.00 [0.00; 0.41] 1.8 %
No Martin 1983 [27] 0 6 0.00 [0.00; 0.46] 1.5 %
Cahill 2010 [18] 71 89 0.80 [0.70; 0.88] 13.2 % Random effects model 3 349 0.01 [0.00; 0.02] 87.9 %
Miller 1994 [22] 190 241 0.79 [0.73; 0.84] 14.1 % Heterogeneity: I = 0 %, χ = 3.73 (p = 0.44)
Hansell 1990 [23] 4 6 0.67 [0.22; 0.96] 6.5 %
Stovall 1987 [26] 7 7 1.00 [0.59; 1.00] 3.2 % Unkown
Martin 1983 [27] 3 6 0.50 [0.12; 0.88] 7.0 % Vigorito 2016 [17] 0 10 0.00 [0.00; 0.31] 2.5 %
Random effects model 275 349 0.78 [0.74; 0.83] 44.0 % Spaans 2003 [20] 0 4 0.00 [0.00; 0.60] 1.0 %
Heterogeneity: I = 1 %, χ = 4.04 (p = 0.40) Riva 1961 [28] 0 21 0.00 [0.00; 0.16] 5.3 %
Random effects model 0 35 0.00 [0.00; 0.03] 8.8 %
Unkown Heterogeneity: I = 0 %, χ = 0 (p = 1.00)
Vigorito 2016 [17] 9 10 0.90 [0.55; 1.00] 5.2 %
Landon 2006 [19] 64 104 0.62 [0.51; 0.71] 13.8 % Random effects model 4 397 0.01 [0.00; 0.01] 100.0 %
Spaans 2003 [20] 4 4 1.00 [0.40; 1.00] 3.1 % Heterogeneity: I = 0 %, χ = 7.94 (p = 0.54)
Riva 1961 [28] 9 21 0.43 [0.22; 0.66] 11.1 % 0 0.2 0.4 0.6 0.8
Random effects model 86 139 0.64 [0.43; 0.80] 33.2 % Proportion
Heterogeneity: I = 57 %, χ = 6.9 (p = 0.08)

Random effects model 382 540 0.67 [0.53; 0.78] 100.0 %

Heterogeneity: I = 80 %, χ = 55.20 (p < 0.01)
0.2 0.4 0.6 0.8 1 ▶Fig. 3 Forest plots of uterine rupture prevalence during TOL in
Proportion
women with a history of three or more previous CSs. Panel A) Analy-
sis subgrouped by prostaglandin use. Panel B) Analysis subgrouped
by oxytocin use. Panel C) subgroup analysis with subgroups defined
▶Fig. 2 Forest plots of successful vaginal birth after TOL in women
by the presence of classic or T-shaped scars in the cohorts.
with a history of three or more previous CSs. Panel a) Analysis sub-
grouped by prostaglandin use. Panel b) Analysis subgrouped by
oxytocin use. Panel c) subgroup analysis with subgroups defined by
the presence of classic or T-shaped scars in the cohorts.

ies of women undergoing a TOL after more than one CS were

Interpretation ­excluded, as no separate information on cases undergoing a TOL
In this meta-analysis we were able to include 12 studies, for a total after three or more CSs was given [31–36]. One further study was
of 540 women undergoing a TOL after three or more CSs. Six stud- excluded due to overlapping data [37], and five case reports were

Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved. 101
 Review

a
TOL 3–4 TOL 1
Study Success TOL Success TOL Odds Ratio OR CI 95 % Weight

Cahill 2010 [18] 71 89 9401 12535 1.31 [0.78; 2.21] 23.5 %

Landon 2006 [19] 64 104 12490 16915 0.57 [0.38; 0.84] 26.1 %
Miller 1994 [22] 190 241 9063 10880 0.75 [0.55; 1.02] 27.8 %
Novas 1989 [24] 8 9 102 144 3.29 [0.40; 27.16] 4.9 %
Stovall 1987 [26] 7 7 173 221 4.19 [0.24; 74.72] 2.9 %
Riva 1961 [28] 9 21 109 138 0.20 [0.08; 0.52] 14.9 %

Random effects model 349 471 31338 40833 0.74 [0.44; 1.23] 100.0 %
Heterogeneity: I = 71 %, χ = 17.08 (p < 0.01)
0.1 0.5 1 2 10
Reduced success in TOL 3–4 <– –> Increased success in TOL 3–4
b OR
TOL 3–4 TOL 2
Study Success TOL Success TOL Odds Ratio OR CI 95 % Weight

Cahill 2010 [18] 71 89 807 1082 1.34 [0.79; 2.30] 14.3 %

Landon 2006 [19] 64 104 584 871 0.79 [0.52; 1.20] 27.7 %
Spaans 2003 [20] 4 4 45 55 2.08 [0.10; 41.62] 0.4 %
Emembolu 1998 [21] 11 39 35 100 0.73 [0.32; 1.64] 8.2 %
Miller 1994 [22] 190 241 1186 1586 1.26 [0.90; 1.75] 38.3 %
Hansell 1990 [23] 4 6 23 29 0.52 [0.08; 3.56] 1.5 %
Novas 1989 [24] 8 9 21 27 2.29 [0.24; 22.08] 0.7 %
Pruett 1988 [25] 2 4 23 51 1.22 [0.16; 9.33] 1.0 %
Stovall 1987 [26] 7 7 36 44 3.49 [0.18; 67.30] 0.4 %
Riva 1961 [28] 9 21 41 55 0.26 [0.09; 0.74] 7.5 %

Fixed effect model 370 524 2801 3900 1.03 [0.83; 1.27] 100.0 %
Heterogeneity: I = 32 %, χ = 13.16 (p = 0.16)
0.1 0.5 1 2 10

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

Reduced success in TOL 3–4 <– –> Increased success in TOL 3–4
c OR
TOL 3–4 TOL 1
Study Rupture TOL Rupture TOL Odds Ratio OR CI 95 % Weight

Miller 1994 [22] 3 241 63 10880 2.16 [0.67; 6.94] 96.5 %

Stovall 1987 [26] 0 7 1 221 9.80 [0.37; 261.04] 3.5 %

Fixed effect model 3 248 64 11101 2.43 [0.82; 7.19] 100.0 %

Heterogeneity: I = 0 %, χ = 0.73 (p = 0.39)
0.01 0.1 1 10 100
Reduced risk in TOL 3–4 <– –> Increased risk in TOL 3–4
OR
d
TOL 3–4 TOL 2
Study Rupture TOL Rupture TOL Odds Ratio OR CI 95 % Weight

Spaans 2003 [20] 0 4 3 55 1.67 [0.07; 37.58] 5.8 %

Miller 1994 [22] 3 241 29 1586 0.68 [0.20; 2.24] 84.6 %
Novas 1989 [24] 0 9 1 27 0.93 [0.03; 24.84] 8.4 %
Pruett 1988 [25] 1 4 1 51 16.67 [0.82; 337.01] 1.2 %

Fixed effect model 4 258 34 1719 0.95 [0.37; 2.45] 100.0 %

Heterogeneity: I = 23 %, χ = 3.92 (p = 0.27)
0.01 0.1 1 10 100
Reduced risk in TOL 3–4 <– –> Increased risk in TOL 3–4
OR

▶Fig. 4 Forest plots showing the difference between women with a history of three or more previous CSs (TOL 3–4) versus one (TOL 1) or two (TOL
2) previous CSs. Panel a) Odds ratio (OR) of successful vaginal delivery difference between three or more previous CSs (TOL 3–4) versus one (TOL 1)
previous CS. Panel b) Odds ratio (OR) of successful vaginal delivery difference between three or more previous CSs (TOL 3–4) versus two (TOL 2)
previous CSs. Panel c) Odds ratio (OR) of uterine rupture occurrence between three or more previous CSs (TOL -3–4) versus one (TOL 1) previous CS.
Panel d) Odds ratio (OR) of uterine rupture occurrence between three or more previous CSs (TOL -3–4) versus two (TOL 2) previous CSs.

excluded according to the exclusion criteria of this meta-analysis quality to correct results for several confounding factors. Unfortu-
(one of the latter was in Bulgarian) [38–42]. nately, it lacks explicit information on neonatal complications, even
Among the included studies, one of the most recent was a large though it is assumed there were not, as it reports no cases of com-
multi-center survey conducted by Landon et al. during a three-year posite maternal complication, including uterine rupture [19]. A fur-
period from 1999 to 2002, including 104 women undergoing a TOL ther study was published by Emombolu et al., including a total of
after three or more CSs [20]. This study has the strength to be pro- 39 women [22]. It is a retrospective study performed in Africa, hav-
spective and to have corrected results for several confounding fac- ing the peculiarity of admitting to a TOL only women in advanced
tors for better comparability between cases and controls. It also labor in which, due to lack of time until delivery, the policy of a re-
has the advantage of reporting for comparison the outcomes for peated CS usually adopted in these cases was no longer practica-
the same study cohort of women undergoing one and two previ- ble. The fifth larger study from Riva et al. included 21 women [29].
ous CSs, respectively. The other large study was conducted by Mill- It was a prospective, single-center study, however it lacked correc-
er et al. [23]. It is a two-center study performed over a ten-year pe- tion for confounding factors. The other studies were minor ones,
riod between 1983 and 1992, including the larger number of including a further 45 women undergoing a TOL after three CSs
women included in this meta-analysis, 241 overall. Of note, it pro- and one woman after four CSs [18, 21, 24–28].
vides, as in the above cited study of Landon et al., a comparison of According to our data, the rate of success of TOL after three or
the outcomes with women undergoing one and two previous CSs more CSs was 0.67, with a higher rate of successful vaginal deliv-
[20, 23]. His main disadvantages are that it was a retrospective ery observed when no classical or T-shaped uterine scar was admit-
study and did not correct results for confounding factors. The third ted and when the use of PDA and PG or oxytocin was allowed. A
larger study was performed by Cahill between 1996 and 2000, in- higher success rate, 0.90 (CI 95 % 0.77, 0.96), was also observed
cluding 89 women. It is a multi-center retrospective study with the among women undergoing a TOL after three or more CSs having a

102 Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved.
history of a prior vaginal delivery [18, 19]. On the contrary, allow- Even a trial of labor after two cesarean sections has been de-
ing a TOL only for women already in advanced labor, as was the case scribed as an affordable option in selected cases [47]. A recent sys-
in the study of Emembolu et al. where a policy of repeated CSs was tematic review and meta-analysis comparing the outcome of a trial
applied whenever possible, was obviously associated with a low of labor after one or two cesarean sections supports this sugges-
success rate (0.28, 95 CI 0.15–0.45) [22]. Also of note, a lower suc- tion [48]. Even though statistically significantly different, similar
cessful vaginal delivery rate was found in prospective studies com- results were found in terms of rate of successful vaginal deliveries
pared to retrospective ones (0.56 vs. 0.74). The different manage- (71.1 vs. 76.5 %), uterine ruptures (1.36 vs. 1.59 %), hysterectomies
ment of the TOL between the different studies could explain these (0.56 vs. 0.19 %) and blood transfusion (1.99 vs. 1.21 %). No differ-
differences. However, in both groups women with different types ences were found in maternal and neonatal outcomes comparing
of uterine scars were included other than low uterine incision, and TOL after 2 CSs with repeated CSs [45]. The difference in uterine
the use of PG, oxytocin, or PDA was admitted. Thus, this difference rupture rate in this latter study compared to the above-cited pro-
could be due to a collection bias, typically found retrospective stud- spective study of Landon et al., namely 0.7 %, could be explained
ies, and results reported from prospective studies should be con- both by a selection bias and less stringent definition used for uter-
sidered as more robust. ine rupture (all ruptures, not only the symptomatic ones) of some
Besides the success rate, the incidence of maternal or neonatal studies included in the systematic review of Tahseen et al. [20, 48].
complications is a key issue for assessing the safety of TOL after No differences were found in maternal and neonatal outcomes

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

three or more CSs. Overall, we found a low incidence of severe com- comparing TOL after 2 CSs with repeated CSs [48].
plications: four cases of uterine ruptures, one requiring a hysterec-
tomy with blood transfusion; no case of maternal death; no case of Strengths and weaknesses
neonatal asphyxia or death. A subgroup analysis considering the Even with the above-mentioned amount of evidence, the old
management of labor in terms of type of previous uterine scar, use Cragin’s dictum dating back to 1916 “once a cesarean section, al-
of PG, oxytocin or PDA use was performed. Even though the rarity ways a cesarean section” seems to remain deeply anchored in ob-
of the events occurred limits the robustness of the results present- stetric culture as a dogma [49]. A recent survey conducted among
ed, this analysis suggests different strategies used for the TOL. All a group of French gynecologists asked them to estimate uterine
four cases occurred in studies where a classic or T-shaped uterine rupture rate after TOL. Interestingly, gynecologists heavily overes-
scar as well as the use of PG and oxytocin were allowed [23, 26]. timated it, indicating a hypothetical 2.8 and 14.2 % rupture rate,
However, except for the case described by Pruett et al. where the respectively, after one and two previous cesarean sections [50].
uterine rupture occurred on the base of an unknown uterine inci- Challenging prejudices with careful and qualitative information is
sion and was followed by a hysterectomy and blood transfusion worth the effort. With this point of view, this study has the quality
after a successful vaginal birth, we found no further information on to summarize evidence about the TOL in women with a history of
the other three cases reported by Miller et al. [23, 26]. three or more previous CSs and review the current knowledge in
Interesting are the results of the comparison of the results of a this field for a better understanding of this issue.
TOL after three or more versus one or two CSs performed on the Nevertheless, the results of this meta-analysis should be inter-
same cohorts of patients studied. This analysis allows a better in- preted with caution due to several potential limitations and bias.
terpretation of the results, having been obtained under compara- All the studies included were observational, with no randomized
ble management conditions of women undergoing a TOL after one, trial reported, and a very wide time span, ranging from the year
two, or three CSs. A slightly but non-significantly lower vaginal de- 1960 to the year 2015. Language bias may have been introduced
livery rate and higher uterine rupture rate was observed in women because articles published in languages other than English or Ger-
with a history of three or more CSs compared to women with a his- man were not included in the assessment. However, only one study,
tory of one previous CS, while almost no differences were observed a case report, in Bulgarian was found and excluded [42]. A full-text
when compared to women with a history of two CSs. bias might further influence the observations done in this study by
retaining only the full-text articles [51]. However, full text of all per-
Comparison with the literature tinent studies was found, except for a review paper in Hebrew [52].
Performing a trial of labor after one cesarean section has been sug- Most importantly, only three of ten studies were conducted pro-
gested as a feasible and relatively safe option by several studies and spectively, accounting for 131 of the 441 total TOL after three or
international institutions [43, 44]. A systematic review conducted more CSs considered [20, 28, 29]. This might have impacted the
by Guise et al. describe a vaginal delivery success rate ranging from accuracy of the results due to possible biases in collecting data but
60 % to 82 %, with no significant difference in maternal deaths or also due to possible issues related to confounding factors that can-
hysterectomy compared to planned cesarean sections [45]. A large not be resolved in this meta-analysis. Considering the risk of severe
multicentric prospective study evaluated the safety of TOL for maternal and neonatal complications, a probable publication bias
mother and newborn: symptomatic uterine rupture occurred in was found using the rank correlation test, implying an underesti-
0.7 % with a 0.05 % risk of derived neonatal ischemic encephalop- mation of this outcome due to the well-known issue that small
athy and a 1.7 % risk of blood transfusion compared to 1 % occur- studies showing a negative effect are unlikely to be published [53].
ring during an elective CS. Hysterectomy or maternal death did not Also, publication bias could be difficult to detect when fewer than
differ significantly between TOL and elective CS (respectively 0.2 10 studies are analyzed, as in the case of the subgroup analysis
vs. 0.3 and 0.02 vs. 0.04 %) [46]. comparing the risk of uterine rupture between three or more CSs

Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved. 103
 Review

and one or two previous CSs. Worthy of interest would be collect- [6] Clark EA, Silver RM. Long-term maternal morbidity associated with
repeat cesarean delivery. Am J Obstet Gynecol 2011; 205: S2–S10
ing further maternal and neonatal outcomes, like the neonatal
Apgar < 7 at the 5th minute, as well the indication for the previous [7] Jackson N, Paterson-Brown S. Physical sequelae of caesarean section.
Best Pract Res Clin Obstet Gynaecol 2001; 15: 49–61
CS, but lack of explicit information on these topics hindered the
[8] Moher D, Liberati A, Tetzlaff J et al. Preferred reporting items for
data analysis.
systematic reviews and meta-analyses: The PRISMA statement. PLoS
Med 2009; 6: e1000097–e1000097
Generalizability
[9] Stroup DF, Berlin JA, Morton SC et al. Meta-analysis of observational
Methodological weaknesses inherent to a meta-analysis, the qua­ studies in epidemiology: A proposal for reporting. Meta-analysis Of
lity of the studies included, the limited number of severe maternal Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;
and neonatal complications reported as well as the changes in ob- 283: 2008–2012
stetric practice during the last decades limit the generalization of [10] Bacchetti S, Pasqual EM, Bertozzi S et al. Curative versus palliative
these findings and preclude firm conclusions on this topic. surgical resection of liver metastases in patients with neuroendocrine
tumors: A meta-analysis of observational studies. Gland Surg 2014; 3:
Relevance of the findings 243–251
[11] Martella L, Bertozzi S, Londero AP et al. Surgery for liver metastases
Regardless of the limitations of this study, being prepared to tailor from gastric cancer: A meta-analysis of observational studies.
this special issue should be mandatory for caregivers. Despite the Medicine (Baltimore) 2015; 94: e1113

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

rarity of this event, there are still and there will always be women [12] R Core Team R: A Language and Environment for Statistical
desiring to give birth naturally, even after three or more CSs. This Computing. Vienna, Austria 2022
meta-analysis can therefore be helpful for obstetricians and mid- [13] Egger M, Davey Smith G, Schneider M et al. Bias in meta-analysis
wives in the counseling and management of this issue. detected by a simple, graphical test. BMJ 1997; 315: 629–634
[14] Begg CB, Mazumdar M. Operating characteristics of a rank correlation
test for publication bias. Biometrics 1994; 50: 1088–1101
Conclusions [15] Sterne JA, Egger M. Funnel plots for detecting bias in meta-analysis:
Performing a TOL after three or more CSs was associated with a Guidelines on choice of axis. J Clin Epidemiol 2001; 54: 1046–1055

substantially high rate of successful vaginal delivery and a low fre- [16] Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-
analysis. Stat Med 2002; 21: 1539–1558
quency of severe maternal or neonatal complications observed,
not significantly different compared to the same study cohorts un- [17] Tassi A, Parisi N, Londero AP. Misoprostol administration prior to
intrauterine contraceptive device insertion: A systematic review and
dergoing a TOL after one or two previous CSs. Results may support
meta-analysis of randomised controlled trials. Eur J Contracept Reprod
a policy of TOL in selected patients desiring a natural birth after Health Care 2020; 25: 76–86
three or more previous CSs. However, as current data is limited, no [18] Vigorito R, Montemagno R, Saccone G et al. Obstetric outcome
firm conclusions can be drawn. Careful counseling should be per- associated with trial of labor in women with three prior cesarean
formed, pointing out the possible underestimation of the poten- delivery and at least one prior vaginal birth in an area with a
tial severe complications reported in this meta-analysis. particularly high rate of cesarean delivery. J Matern Fetal Neonatal Med
2016; 29: 3741–3743
[19] Cahill AG, Tuuli M, Odibo AO et al. Vaginal birth after caesarean for
Conflict of Interest women with three or more prior caesareans: Assessing safety and
success. BJOG 2010; 117: 422–427
[20] Landon MB, Spong CY, Thom E et al. Risk of uterine rupture with a trial
The authors declare that they have no conflict of interest.
of labor in women with multiple and single prior cesarean delivery.
Obstet Gynecol 2006; 108: 12–20
[21] Spaans WA, van der Vliet LME, Röell-Schorer EAM et al. Trial of labour
References
after two or three previous caesarean sections. Eur J Obstet Gynecol
Reprod Biol 2003; 110: 16–19
[1] Betrán AP, Ye J, Moller AB et al. The increasing trend in caesarean [22] Emembolu JO. Vaginal delivery after two or more previous caesarean
section rates: global, regional and national estimates: 1990–2014. sections: Is trial of labour contraindicated? J Obstet Gynaecol 1998;
PLoS One 2016; 11: e0148343 18: 20–24
[2] Zwecker P, Azoulay L, Abenhaim HA. Effect of fear of litigation on [23] Miller DA, Diaz FG, Paul RH. Vaginal birth after cesarean: A 10-year
obstetric care: A nationwide analysis on obstetric practice. Am J experience. Obstet Gynecol 1994; 84: 255–258
Perinatol 2011; 28: 277–284
[24] Hansell RS, McMurray KB, Huey GR. Vaginal birth after two or more
[3] Gibbons L, Belizan JM, Lauer JA et al. Inequities in the use of cesarean cesarean sections: A five-year experience. Birth 1990; 17: 146–150.
section deliveries in the world. Am J Obstet Gynecol 2012; 206: 331. discussion 150–151
e1–19
[25] Novas J, Myers SA, Gleicher N. Obstetric outcome of patients with
[4] Fruscalzo A, Salmeri MG, Cendron A et al. Introducing routine trial of more than one previous cesarean section. Am J Obstet Gynecol 1989;
labour after caesarean section in a second level hospital setting. J 160: 364–367
Matern Fetal Neonatal Med 2012; 25: 1442–1446
[26] Pruett KM, Kirshon B, Cotton DB et al. Is vaginal birth after two or
[5] Ashford JB, LeCroy CW, Lortie KL. Human Behavior in the Social more cesarean sections safe? Obstet Gynecol 1988; 72: 163–165
Environment: A Multidimensional Perspective. 4th ed. Belmont, CA:
[27] Stovall TG, Shaver DC, Solomon SK et al. Trial of labor in previous
Brooks/Cole, Cengage Learning; 2010
cesarean section patients, excluding classical cesarean sections.
Obstet Gynecol 1987; 70: 713–717

104 Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved.
[28] Martin JN, Harris BA, Huddleston JF et al. Vaginal delivery following [42] Katsulov A, Koleva Z, Iankov M et al. Successful vaginal delivery after
previous cesarean birth. Am J Obstet Gynecol 1983; 146: 255–263 three previous cesarean sections – a case report and comment. Akush
[29] Riva HL, Teich JC. Vaginal delivery after cesarean section. Am J Obstet Ginekol (Sofiia) 2005; 44: 11
Gynecol 1961; 81: 501–510 [43] ACOG ACOG Practice bulletin no. 115: Vaginal birth after previous
[30] Burns PB, Rohrich RJ, Chung KC. The levels of evidence and their role in cesarean delivery. Obstet Gynecol 2010; 116: 450–463
evidence-based medicine. Plast Reconstr Surg 2011; 128: 305–310 [44] RCOG. Birth after previous caesarean birth. Green-top Guideline No.
[31] Asakura H, Myers SA. More than one previous cesarean delivery: A 45. October 2015. Available from: https://www.rcog.org.uk/media/
5-year experience with 435 patients. Obstet Gynecol 1995; 85: kpkjwd5h/gtg_45.pdf
924–929 [45] Guise JM, Berlin M, McDonagh M et al. Safety of vaginal birth after
[32] Granovsky-Grisaru S, Shaya M, Diamant YZ. The management of labor cesarean: A systematic review. Obstet Gynecol 2004; 103: 420–429
in women with more than one uterine scar: Is a repeat cesarean [46] Landon MB, Hauth JC, Leveno KJ et al. Maternal and perinatal
section really the only “safe” option? J Perinat Med 1994; 22: 13–17 outcomes associated with a trial of labor after prior cesarean delivery.
[33] Flamm BL, Newman LA, Thomas SJ et al. Vaginal birth after cesarean N Engl J Med 2004; 351: 2581–2589
delivery: Results of a 5-year multicenter collaborative study. Obstet [47] Roberts LJ. Elective section after two sections – where’s the evidence?
Gynecol 1990; 76: 750–754 Br J Obstet Gynaecol 1991; 98: 1199–1202
[34] Farmakides G, Duvivier R, Schulman H et al. Vaginal birth after two or [48] Tahseen S, Griffiths M. Vaginal birth after two caesarean sections
more previous cesarean sections. Am J Obstet Gynecol 1987; 156: (VBAC-2)-a systematic review with meta-analysis of success rate and

Downloaded by: ULB Universitäts- und Landesbibliothek Bonn. Copyrighted material.

565–566 adverse outcomes of VBAC-2 versus VBAC-1 and repeat (third)
[35] Porreco RP, Meier PR. Trial of labor in patients with multiple previous caesarean sections. BJOG 2010; 117: 5–19
cesarean sections. J Reprod Med 1983; 28: 770–772 [49] Cragin E. Conservatism in obstetrics. NY Med J 1916; 54: I
[36] Saldana LR, Schulman H, Reuss L. Management of pregnancy after [50] Doret M, Touzet S, Bourdy S et al. Vaginal birth after two previous
cesarean section. Am J Obstet Gynecol 1979; 135: 555–561 c-sections: Obstetricians-gynaecologists opinions and practice
[37] Paul RH, Phelan JP, Yeh SY. Trial of labor in the patient with a prior patterns. J Matern Fetal Neonatal Med 2010; 23: 1487–1492
cesarean birth. Am J Obstet Gynecol 1985; 151: 297–304 [51] Emmerick ICM, Oliveira MA, Luiza VL et al. Access to medicines in Latin
[38] Bowyer L, Chapman M. Successful vaginal birth after three previous America and the Caribbean (LAC): A scoping study. BMJ Open 2013; 3:
Caesarean sections with no prior labour. Aust N Z J Obstet Gynaecol e002224
2003; 43: 471–472 [52] Granovsky-Grisaru S, Diamant YZ. Vaginal delivery following two or
[39] Sharma S, Thorpe-Beeston JG. Trial of vaginal delivery following three more caesarean sections. Harefuah 1993; 124: 221–222
previous caesarean sections. BJOG 2002; 109: 350–351 [53] Sterne JA, Gavaghan D, Egger M. Publication and related bias in
[40] Lawson GW. Vaginal delivery after 3 previous caesarean sections. Aust meta-analysis: Power of statistical tests and prevalence in the
N Z J Obstet Gynaecol 1987; 27: 115–116 literature. J Clin Epidemiol 2000; 53: 1119–1129

[41] Ibraheim M, Mohamed S, Ikomi A. Unmonitored trial of labour and

home delivery following three previous caesarean sections. J Obstet
Gynaecol 2006; 26: 165–166

Fruscalzo A et al. Trial of Labor after … Z Geburtsh Neonatol 2023; 227: 96–105 | © 2022. Thieme. All rights reserved. 105