Authors: Melania M R Amorim, MD,PhD, Maíra Libertad Takemoto, RN, MSc, Marcos Nakamura, MD, MSc, Ricardo Herbert Jones, MD, Roxana Knobel,MD,PhD

Urban Midwifery Group, Rede de Humanização do Nascimento (ReHuNa), Brazil

Dear Editor:


The academic community has long awaited the publication of what was being described as an Australian randomized controlled trial (RCT) comparing planned elec-tive cesarean section with trial of labor (TOLAC) in women with previous cesarean section. Therefore, we read with great interest the article by Crowther et al. (1) How¬ever, what we see, not without some disappointment, is that this is an observational cohort study with a small “nested” RCT that could include only 22 patients.

On the one hand, this low number reinforces the belief that there are circumstances where it is impossible to randomize women to delivery route; on the other, it leads to a more criti¬cal analysis of the results, since observational studies are by nature more suscepti¬ble to biases that should be considered in their interpretation. Despite the emphasis in the title on being “an RCT nested in a cohort,” we think it is more appropriate to consid¬er this study as purely observational and to discuss its findings accordingly, critically analyzing its design and its results in light of the quality criteria appli¬cable to this type of study.

One of the most important points to consider in the analysis of a cohort study is the comparison between the groups with regard to risk factors (or predictors) that may influence the outcomes and mask the real effects of the exposure variable (planned elective cesarean section (CS), or vaginal birth after cesarean (VBAC).

In RCT, this concern is resolved with a qualified randomization process, which seeks to ensure homogeneous and comparable groups. In non-randomized studies, it is for the investigators to provide data that allow readers to assess whether these factors were balanced between groups in terms of baseline characteristics and care for the potential imbalances between these factors in the analysis. What we observe in this publication is that it does not present the distribution within the groups of a series of prognostic factors relevant to the issue at hand (e.g., parity, previous or current obstetric disorders) that were certainly collected in the study. The absence of such information affects the assessment of comparability between groups and leaves room to question whether the observed differences in outcomes must be effectively related to the mode of delivery. Additionally, being actually a cohort (not an RCT), one would expect the authors to include p-value for the tests comparing the baseline characteristics of the groups.


Another key strategy in observational studies to ensure valid conclusions about the relationship between exposure and outcome is to control, in the analysis, the possible confounding variables. The analysis was adjusted only for the following variables: body mass index (BMI), socioeconomic status, and the reason for the previous cesarean, but neither the article nor even in the protocol published in full is there any justification for the choice of these variables at the expense of many others, equally or more relevant.

The authors highlighted that an intention-to-treat (ITT) analysis was performed, which is currently considered the gold-standard approach for pragmatic RCT. However, in observational studies, the ITT analysis leaves room for further questions about confounding and should have been supplemented by successful TOLAC subgroup analysis, since high quality evidence indicates that successful vaginal births are associated with lower maternal morbidity rates and similar neonatal mortality and morbidity, as stated above.

The clinically sound neonatal outcome (fetal death or death before discharge) was not different between study groups, as previously reported by studies investigating neonatal morbidity in VBA1C, VBA2C or elective cesarean section,(2) although the perinatal mortality was higher in women undergoing TOLAC as compared to elective cesarean (1.3/1,000 vs. 0.5/1,000). Nevertheless, given the small magnitude of this difference, the authors concluded that there is strong evidence in favor of the safety and feasibility of VBAC.(3) In the current paper, none of the individual outcomes presented significant between-group differences. A critical appraisal of the absolute numbers shows that only cord blood pH<7 (planned CS=1 event vs. VBAC=6 events) and base deficit ≥ 12 (planned CS=1 event vs. VBAC=9 events) were potentially different, variables that should not even be included in this composite analysis. The absolute risk of fetal death was markedly low, as shown in previous studies(3), which raises doubts about the real motivation for the selection of a “neonatal morbidity composite outcome,” given that almost all relevant and irrelevant endpoints had to be pooled to reach statistical significance.

Also regarding the neonatal results, meaningful confounders were not controlled in the multivariate analysis, as gestational age at birth (which was significantly different between groups). Furthermore, in the planned VBAC group, the paper reported 12.4% of labor induction and 93.8% of anesthesia/analgesia, interventions which can adversely impact the outcomes, particularly among women with previous cesarean. Detailed information about these procedures (induction method, characteristics of analgesia/anesthesia) was not provided.

In relation to maternal composite outcome, there were no deaths in any of the groups analyzed, different what is shown in other studies, such as that conducted by the Oregon group, which found a significantly increased maternal mortality in elective cesarean delivery compared to women who had a trial of labor (13.4/100, 000 x 3.8/100, 000). Moreover, the rate of uterine rupture was also similar between groups (planned CS = 1 vs. VBAC = 3), but remained below (0.2%) the rate found in previous observational studies (which, for example, reported rates of uterine rupture for TOLAC of 0.047% vs. 0.003% for elective caesarean section (3), or 1.36% in women who had evidence of labor after 2 cesarean sections (2)), which can be considered a positive point. However, the success rate of VBAC (56.8%) was low compared to other studies, which described vaginal delivery in 71,1% after two C-sections (2), reaching 87% after only one CS (3).

In the literature currently available, maternal morbidity, both short and long term, is greater after elective caesarean section than that observed for VBAC (3) and is comparable in women submitted to a third or fourth caesarean section. (2) In the Crowther et al. article, the only maternal outcome that showed significant difference between planned elective CS vs. VBAC was postpartum hemorrhage (PPH) greater than 1500 ml in the planned VBAC group, but information to critically assess whether the groups were comparable in terms of presence of risk factors for PPH are lacking in the article. There is also no available information (both in the published article and the protocol) with respect to administration of prophylaxis and treatment for PPH in both groups, for example, if oxytocin was administered according to protocols well established in the literature.

Also in relation to the incidence of PPH in the group with planned VBAC, it is curious that this difference was observed regardless of the very low rate of uterine rupture in both groups, which reinforces the assumption that this increased risk of patients in the group who supposedly "tried" a VBAC may have been due to the excess of cesareans and not to complications of vaginal delivery. Unfortunately, multivariate analysis was not performed to control this and other confounding variables; such analysis would be extremely relevant and informative to the debate.

It is noteworthy that 27% of women who had a planned VBAC instead experienced an elective cesarean section. Analyzing all the cesarean sections, we can observe 98% of elective CS in the group of women who planned a cesarean section and a total of 58% of cesarean deliveries in the group of candidates for a VBAC. The text of the article states that the main indications for CS in the VBAC group were previous cesarean section (which really does not make sense - these women should at least have been excluded from the study), failure to progress and fetal distress, but the percentages are not shown. Of the women who had planned a VBAC, 1/3 underwent elective cesarean section. In addition, the lack of details on the management of labor in women who attempted a VBAC (both in the article and in the protocol), calls into question the true commitment and ability of the involved obstetricians to conduct labor in patients with a previous CS. Thus, it is evident that Australian obstetricians have low tolerance for waiting for normal delivery in this situation. Possibly with a larger number of normal deliveries we could observe more cases of dehiscence of the scar-- a risk outweighed by the benefits of successful vaginal delivery.

We are concerned about the consequences of the current publication, since, despite its obvious bias, the study is being advertized as showing that TOLAC is dangerous for women. We should also consider that the results of this present study are insufficient to confirm the safety of the procedure, yet the impacts of repeated cesarean in later pregnancies also have to be considered. It is known that disorders such as placenta praevia and placenta accreta are more frequent in women with previous caesarean section, and that this risk increases with the number of cesarean sections. (3.4) It also seems to be a consensus that the more CS a woman has, the greater the risks for future pregnancies. (4) Therefore, to reduce these risks and therefore future morbidity and maternal and neonatal mortality, we believe it is urgent, at the same time, to reduce the number of both primary cesarean sections in low-risk women (5) and repeat caesarean sections that are medically indicated only because of previous cesarean sections.


1. Crowther CA, Dodd JM, Hiller JE, Haslam RR, Robinson JS. Planned Vaginal Birth or Elective Repeat Caesarean: Patient Preference Restricted Cohort with Nested Randomised Trial. PLoS Medicine. 2012 Mar;9(3):e1001192.
2. Tahseen S, Griffiths M. Vaginal birth after two caesarean sections (VBAC-2)-a systematic review with meta-analysis of success rate and adverse outcomes of VBAC-2 versus VBAC-1 and repeat (third) caesarean sections. BJOG : an international journal of obstetrics and gynaecology. 2010 Jan;117(1):5–19.
3. Jeanne-Marie Guise, MD, MPH, Karen Eden, PhD, Cathy Emeis, PhD, CNM, Mary Anna Denman, MD, Nicole Marshall, MD, Rongwei (Rochelle) Fu, PhD, Rosalind Janik, BA, Peggy Nygren, MA, Miranda Walker, MA, and Marian McDonagh P. Vaginal Birth After Cesarean: New Insights. Rockville: Agency for Healthcare Research and Quality (US); 2010.
4. Marshall NE, Fu R, Guise J-M. Impact of multiple cesarean deliveries on maternal morbidity: a systematic review. American journal of obstetrics and gynecology. 2011 Sep;205(3):262.e1–8.
5. Delbaere I, Cammu H, Martens E, Tency I, Martens G, Temmerman M. Limiting the caesarean section rate in low risk pregnancies is key to lowering the trend of increased abdominal deliveries: an observational study. BMC pregnancy and childbirth. 2012 Jan;12:3.