The safety of trial of labor after cesarean section (TOLAC) versus elective repeat cesarean section (ERCS): a systematic review and meta-analysis

Abstract

Objective

Women’s choice of birth following a cesarean delivery either includes a trial of elective repeat cesarean section (ERCS) or a trial of labor after cesarean (TOLAC). No comprehensive overview or systematic summary is currently available.

Methods

EMBASE, PubMed, and the Cochrane Library databases were searched from inception to 1 February 2020. Studies reporting the safety of TOLAC and ERCS in pregnant women with prior cesarean delivery were included. Statistical analysis was performed using RevMan 5.3 and Stata 15.0. Odds ratios (ORs) and 95% confidence intervals (CIs) were adopted as the effective measures.

Results

A total of 13 studies covering 676,532 cases were included in this meta-analysis. The results demonstrated that the rates of uterine rupture (OR = 3.35, 95%CI [1.57, 7.15], I² = 81%), neonatal asphyxia (OR = 2.32, 95%CI [1.76, 3.08], I² = 0%) and perinatal death (OR = 1.71, 95%CI [1.29, 2.25], I² = 0%) were higher in the TOLAC group compared with the ERCS group. The rates of peripartum hysterectomy (OR = 0.70, 95%CI [0.44, 1.11], I² = 62%), blood transfusion (OR = 1.24, 95%CI [0.72, 2.12], I² = 95%), and puerperal infection (OR = 1.11, 95%CI [0.77, 1.60], I² = 95%) showed no significant differences between the two groups.

Conclusion

TOLAC is associated with a higher risk of uterine rupture, neonatal asphyxia, and perinatal death compared with ERCS. Nevertheless, it should be noted that the risks of all complications were small in both groups. This information is important for healthcare providers and women choosing the delivery type.

Keywords:

• Safety
• trial of labor after cesarean section (TOLAC)
• elective repeat cesarean section (ERCS;)
• uterine rupture
• meta-analysis

Introduction

For women with a history of cesarean section, the choice of birth in their next pregnancy includes ERCS or TOLAC. Both options have their benefits and risks. Compared with planned ERCS, TOLAC is associated with higher risks of hemorrhage, endometritis, uterine rupture, need for blood transfusion, perinatal death, and hypoxic-ischemic encephalopathy. [1] Women receiving ERAC are at risk of surgical complications. According to the recommendations of current guidelines, women who had prior cesarean section should be informed regarding the risks associated with ERCS and vaginal birth after cesarean (VBAC).[2]

Selecting the best approach requires ample evidence regarding the complications after performing TOLAC compared to ERCS. Though adverse events such as uterine rupture following TOLAC or ERCS have been explored, studies comprehensively assessing the complications associated with ERCS and TOLAC are limited. [3,4] A previous study demonstrated that TOLAC was more likely to cause uterine rupture. [5] However, the UKOSS data collected from four studies performed between 2005 and 2012 showed no clear differences in the risk of uterine rupture associated with TOLAC and ERCS. [6] Uterine rupture rarely occurs following TOLAC, but it is serious, and women with uterine rupture may have an increased risk of other severe complications.

Much attention has been drawn to the safety of TOLAC and ERCS. Several studies compared the safety of TOLAC and ERCS, but few studies summarized these findings. Therefore, we conducted a systematic meta-analysis to evaluate the safety of ERCS and TOLAC in pregnant women with a prior history of a cesarean section.

Materials and methods

Search strategy

This meta-analysis abided by the PRISMA statement guideline. Literature was searched through the EMBASE, PubMed, and Cochrane databases for full-text articles published from inception to 1 February 2020, conducted independently by two investigators (Lili Qiu and Jingjing Zhu). The following keywords or Medical Subject Heading (MeSH) terms were checked: ERCS, VBAC, elective repeat cesarean section, vaginal birth after cesarean, the trial of labor after cesarean section, and trial labor after previous cesarean delivery and TOLAC.

Study selection

Endnote X8.0.1 was used to assess the eligibility of all citations for this meta-analysis. The two authors (Lili Qiu and Jingjing Zhu) independently reviewed the titles, abstracts, and full texts of all articles in a standardized, blinded manner. Any disagreement on the eligibility of a study between the investigators was resolved by mutual discussion.

According to the PICOS principles, the included studies had to meet the following inclusion criteria: Participants (P): pregnant women with a prior cesarean delivery; Intervention (I): successful TOLAC; Comparison (C): elective repeat cesarean section (ERCS); Outcomes (O): safety associated with TOLAC and ERCS; and Study design (S): RCTs, case-control studies and cohort studies. Studies appeared as case reports, expert opinions, editorials, reviews, letters, or non-human studies or were excluded. Two researchers (Lili Qiu and Jingjing Zhu) independently screened the publications by titles, abstracts, and full texts. For duplicate datasets, only the largest sample size study was selected. A third researcher participated in coinciding with any disagreement.

Data extraction and quality assessment

The above-mentioned two researchers conducted data extraction independently, and a third researcher joined to resolve any disagreement. The data were collected and summarized into simple standard forms based on the following contents: (1) the surname of the first author and year of publication; (2) sample size (N); (3) country of origin; (4) study design (RCTs, case-control study, prospective study, or retrospective study); and (5) Newcastle-Ottawa Scale (NOS) or Jadad score. Non-randomized study qualities were evaluated by NOS [7], which comprised eight items in three categories, including study selection; comparability; and data ascertainment pattern (objectively or subjectively). The highest score was 9, and 6 points or above indicated high quality. The Jadad score was used to evaluate the quality of RCTs, and a score of <3 was considered as low quality, otherwise as high quality.

Data synthesis and analysis

All statistical analyses were carried out using RevMan 5.3 and the Stata^® Statistical Software (version 15.0, StataCorp LP, College Station, USA). Heterogeneity among studies was evaluated using Q and I² statistics, wherein an I² of >50% indicated moderate to high heterogeneity. The heterogeneity source was traced by subgroup and meta-regression analyses. The selection of fixed- or random-effects models was based on heterogeneity analysis [8,9]. Fixed-effects model was adopted if I² was <50%. In other cases, a random-effects model was applied. Additionally, ORs and 95% Cis were used to assess RCTs, case-control, and cohort studies.

Results

Search results

The database search yielded 516 publications, and 236 documents were selected after removing duplicates. Studies published from inception until 1 February 2020, were searched manually. After reviewing the titles and abstracts, 121 documents were excluded, and 115 full texts were read carefully. Finally, 13 studies with 676,532 patients were included in the quantitative analysis [3,6,10–-20]. A detailed flow diagram of the study selection process is shown in (Figure S1).

Characteristics and quality assessment of studies

Totally, 13 studies published from 1989 to 2019 were included in a meta-analysis. One study was an RCT, one was a case-control study, and 11 were cohort studies (three were retrospective and eight were prospective studies). The sample size ranged from 137 to 494,118 cases. The NOS quality scores of 11 studies were more than or equal to 6, and the Jadad scores were more than 3, indicating that these studies had a high methodological quality level. Table 1 shows the characteristics of these 13 studies. In addition, the details of the quality scores were presented in Tables S1 and S2 separately.

Table 1. The baseline characteristics of studies.

Study, year	N	Country	Type of studies	Score*
Phelan, 1989 [11]	1088	America	Prospective	4
Iglesias, 1991 [12]	137	Canada	Prospective	6
Abitbol, 1993 [13]	312	America	retrospective	6
Flamm, 1994 [14]	7229	America	Prospective	7
McMahon, 1996 [15]	6138	America, Canada	Prospective	4
Hook, 1997 [16]	989	America	Prospective	5
Pegeth, 1999 [17]	29,046	Switzerland	Prospective	7
Landon, 2004 [18]	33,699	America	Prospective	7
Crowther, 2012 [3]	2345	America, Australia	RCT*	4
Madi, 2013 [19]	730	Portugal	case-control	6
Miller, 2015 [20]	10,262	America	Prospective	6
Nair, 2015 [6]	494,118	America	retrospectizzve	7
Pont, 2019 [10]	90,439	Australia	retrospective	6

The safety of TOLAC vs. ERCS

There were 10 studies (N = 212,440) that described uterine rupture. The results showed a significant difference between TOLAC and ERCS (OR = 3.35, 95%CI [1.57, 7.15], I² = 81%, Figure 1). Subgroup analyses showed moderate uterine rupture rate differences (Z = 3.12, p = .002). Based on the meta-regression analysis, the number of patients (B = 0.66, Z = 3.39, p = .01) positively affected the uterine rupture rate. Additionally, the publication year (B = 0.004, Z = 0.08, p = .937) and the score (B = 0.27, Z = 0.69, p = .51) exhibited moderate effects on the incidence of uterine rupture.

Figure 1. Meta-analysis on the rate of uterine rupture.

A total of nine studies [N = 199,291] covered hysterectomy in this meta-analysis. The results demonstrated no significant difference between TOLAC and ERCS (OR = 0.70, 95%CI [0.44, 1.11], I² = 62%, Figure S2). Subgroup analyses did not identify any significant difference. Meta-regression analyses revealed that the number of patients (B = −0.54, Z =2.53, p = .045) negatively affected the hysterectomy rate in both groups. The publication year (B = 0.05, Z = 2.4, p = .053) and the scores (B = −0.11, Z = −0.44, p = .672) showed no effects on the hysterectomy rate.

A total of eight studies [N = 302,627] included blood transfusion in this meta-analysis. This complication did not differ significantly between TOLAC and ERCS (OR = 1.24, 95%CI [0.72, 2.12], I² = 95%) (Figure S3). Subgroup analyses based on the type of studies revealed a moderate difference in blood transfusion rate (Z = 2.69, p = .007). According to meta-regression analysis, the number of patients (B = 0.29, Z = 0.79, p = .46), publication year (B = 0.04, Z = 1.31, p = .24), and the scores (B = −0.12, Z = −0.44, p = .68) were not associated with the rate of blood transfusion.

Seven studies (N = 192,683) reported puerperal infection and no significant difference was found between TOLAC and ERCS (OR = 1.11, 95%CI [0.77, 1.60], I² = 95%, Figure S4). Subgroup analyses based on the type of studies demonstrated moderate difference in the rate of puerperal infection (Z = 2.78, p = .005). Meta-regression analysis was performed, and the results showed that the number of patients (B = 0.29, Z = 0.79, p = .46), publication year (B = 0.007, Z = 0.29, p = .782), and scores (B = −0.055, Z =−0.32, p = .758) did not exert any effect on the rate of puerperal infection.

A total of six studies (15,746 patients) covered neonatal asphyxia, and a significant difference was detected between TOLAC and ERCS (OR = 2.32, 95%CI [1.76, 3.08], I² = 0%, Figure S5).

Nine studies with 83879 patients discussed perinatal death, and a significant difference was found between TOLAC and ERCS (OR = 1.71, 95%CI [1.29, 2.25], I² = 0%, Figure 2).

Figure 2. Meta-analysis on the rate of perinatal death.

Publication bias

Funnel plot (Figure S6) and Egger’s tests suggested no obvious publication bias across the studies regarding uterine rupture (t = 2.28, 95% CI: −0.01 to 1.77, p = .052, Figure S7), hysterectomy (t = −0.90, 95% CI: −1.23 to 0.55, p = .4, Figure S8), blood transfusion (t = −1.38, 95% CI: −10.4 to 2.89, p = .216, Figure S9), and puerperal infection (t = −0.69, 95% CI: −1.41 to 0.81, p = .518, Figure S10). According to funnel plots and Egger’s tests, publication bias existed for neonatal asphyxia (t = 5.06, 95% CI: 0.03 to 1.02, p = 0.007, Figure S11), and perinatal death estimates (t = 3.77, 95% CI: 0.2 to 0.94, p = .009, Figure S12).

Discussion

This review identified 13 studies examining the safety of ERCS and TOLAC in pregnant women with prior cesarean delivery. The analysis revealed that TOLAC was linked with a higher incidence of uterine rupture, neonatal asphyxia, and perinatal death compared with ERCS.

Several studies [6,10,19] have assessed the safety of ERCS and TOLAC, but these trials assessed the rates of uterine rupture, hysterectomy, blood transfusion, puerperal infection, neonatal asphyxia, and perinatal death. McMahon et al. [15] showed that pregnant women undergoing a second cesarean section who delivered with a TOLAC were almost twice likely to develop major maternal complications than those who experienced an ERCS. Hook et al. [16] demonstrated that infants born by ERCS had an increased risk of developing respiratory conditions compared with those delivered by TOLAC. However, Nari et al. [6] showed that the incidence rate of all examined complications in both groups remained low, and the risks of severe sepsis, peripartum hemorrhage, peripartum hysterectomy, and failed tracheal intubation showed no significant difference between ERCS and TOLAC in absolute or relative terms. Pont et al. [10] demonstrated a higher transfusion risk following TOLAC compared to ERCS. Though the side effect of transfusion remains low, it is critical for women to choose TOLAC for delivery, as facilities must have ready access to blood products. We summarized these studies and other trials. The results demonstrated that the rates of uterine rupture, neonatal asphyxia, and perinatal death significantly differed between the TOLAC and ERCS groups, and the rates of peripartum hysterectomy, blood transfusion and puerperal were similar between the two groups.

Perinatal complications and death associated with uterine rupture among women having prior cesarean should be primarily considered when counseling patients on TOLAC. [21] Hypoxic brain injury is considered an underreported poor prognosis associated with uterine rupture. Still, the definition of perinatal asphyxia is conflicting or poorly reported in previous studies that examined TOLAC. Of importance, Apgar score and cord-blood gas level were reported in only a few studies. Although vaginal delivery is generally considered to cause lower maternal morbidity and mortality compared with cesarean section, other data have suggested an elevated trend of fetal death among those receiving TOLAC. [22] Among people who attempt vaginal delivery before cesarean section, those with a ruptured uterus are at the greatest risk of serious complications. Although there is an increase in maternal mortality after cesarean section compared to that after vaginal delivery, pregnant women should be counseled on the frequency of death and other confounding variables that may complicate the comparison of mortality, such as maternal disease, emergency operation, or a non-emergency procedure. In this meta-analysis, maternal deaths were not significantly more common with ERCS than with TOLAC. Still, such deaths in patients were rare, so the statistical power to detect a significant eventual difference was low.[23] The information we provided in this meta-analysis was important for healthcare providers and women choosing the type of birth delivery.

Our results demonstrated that TOLAC correlated with a higher incidence rate of uterine rupture, neonatal asphyxia, and perinatal death compared with ERCS. Still, the risks of these complications were small in both groups. Using a decision-model analysis based on the risk-benefit ratio of ERCS or TOLAC, Paré et al. [24] have suggested performing ERCS or TOLAC for women preparing for a second cesarean section should depend on future pregnancies planned. ERCS could be the choice for women who expected one more pregnancy, while TOLAC should be selected for those desiring two or more pregnancies, minimizing the death risk caused by multiple cesarean sections.

However, there are still some limitations in our analysis. Firstly, only one RCT was included in our meta-analysis, so it was necessary to enclose more well-designed studies with larger sample sizes to assess the safety of ERCS and TOLAC comprehensively. Secondly, the findings of such studies should be generalized with caution due to the inherent methodological limitations of observational studies. Thirdly, high heterogeneity was found across some outcomes. Finally, only a limited number of each research type study were included, affecting the integrity and authenticity of the collected data.

A brief assessment of non-RCT studies

• Possible biases for the Phelan et al. study are that they used different entry criteria each year and did not report detailed patient characteristics such as age and BMI. But they did report detailed birth outcomes, which is a strength of the study.

• The Iglesias et al. study was conducted in a community hospital with a smaller sample size and provided slightly different information in terms of demographic background. However, the authors could have adjusted their results with potential confounders to provide more valid results.

• An apparent weakness of Abitbol’s study is that clinical workers did not assess the suitability of all participants for TOLAC; they grouped patients according to their willingness to try TOLAC. This study could have covered more information about patients’ concerns and thoughts about TOLAC, i.e. a qualitative analysis, which would have added more information to the field.

• The large sample size in the study by McMahon and colleagues adds credibility to the study, but the control group included patients who were not suitable for TOLAC, which would undoubtedly have biased the final results.

• Hook et al. did not report on the grouping methods for TOLAC and ERCS; their study investigated and recorded detailed data relating to the labor process with a satisfactory sample size.

• As for the Rageth study, in addition to the huge sample size, they also collected data on other possible risk factors such as allergies, smoking status, and cardiovascular disease. However, we do not know the study’s grouping method and intervention process, so it is difficult to examine the bias involved.

• One of the strengths of the Landon study is that they collected data from multiple centers with large sample sizes, and they also did a refined data analysis that adjusted all results for potential confounders. The demographic data was complete, covering information from health status to socio-economic status.

• Madi et al. matched all participants according to basic clinical characteristics to reduce the effect of bias; they also recorded detailed information on the delivery process. It would have been clearer if they had specified the factors they took into account when matching.

• In Miller’s study, they applied well-designed statistical methods to build a model to predict risk factors for obstetric outcomes. Unfortunately, the limited clinical data did not provide additional new evidence on delivery choices.

• In their study, Nair and colleagues carefully excluded women with medical conditions that would have confounded the findings and did sensitivity analyses to test the robustness of the results. However, they may have mistakenly included some women who planned to have an ERCS but ended up having a vaginal birth in the TOLAC group.

• In the study by Pont. et al. inferred the patients’ expected mode of delivery from their medical records. This approach carries some risk and may lead to biased outcomes. In addition, they aimed to examine the difference in transfusion rates between TOLAC and ERCS but lacked information on the drugs and conditions that led to more frequent transfusions. The strength of their study was the exclusion of patients with diseases that would have confounded the outcome and the design of sophisticated statistical models specifically designed to adjust the results with variables known to influence the outcome.

Conclusion

In conclusion, our data indicated that TOLAC might be linked with more frequently occurred uterine rupture, neonatal asphyxia and perinatal death compared with ERCS. However, it should be noticed that the risks associated with all complications were small in both the groups. The information obtained from this study is of great significance for healthcare providers and women who are choosing the delivery mode.

Ethics approval and consent to participate

Not applicable.

Author’s contributions

(I) Conception and design:

(II) Administrative support:

(III) Provision of study materials or patients:

(IV) Collection and assembly of data:

(V) Data analysis and interpretation:

(VI) Manuscript writing: All authors

(VII) Final approval of manuscript: All authors

Reporting checklist

The authors have completed the PRISMA reporting checklist.

Acknowledgements

None.

Disclosure statement

All authors have completed the ICMJE uniform disclosure form. The authors have no conflicts of interest to declare.

Data availability statement

All data generated or analyzed during this study are included in this published article.

Additional information

Funding

None.