This week (August 2014), a Dutch research team published the results of a large retrospective cohort study concerning the results of all breech births in the Netherlands. They concluded that an increase in the caesarean section rate following publication of the Term Breech Trial (TBT) has resulted in a significant reduction in perinatal mortality related to breech presentation, and therefore a policy of universal caesarean section for breech would improve outcomes for breech babies even further.
In the year prior to October 2000 (the team does not present data from earlier than 1999; why not?), the perinatal death rate was 1.3/1000 for all breech deliveries (VBB and CS inclusive). Elective sections increased from 24% prior to publication of the TBT to 60% afterwards, and from December 2000-2007, the perinatal mortality rate was 0.7/1000 for all breech deliveries, with an overall VBB rate of 22%. Moreover, the team showed that all of the perinatal deaths occurred during what they called planned breech deliveries (although their understanding of ‘planned breech delivery’ is very different from mine, as I discuss below), so the actual perinatal mortality rate for breech babies born vaginally remained steady at 1.6/1000 during both periods.
A rate of 1.6/1000 is actually quite low compared to the mortality rate of approximately 1/100 reported in the Term Breech Trial. However, the authors propose that the results of this study should replace the information currently given to women in Dutch national guidelines, because according to their calculations, “A policy of elective caesarean section for all term breech deliveries could lower the overall term neonatal mortality in term deliveries by 6.8%, from 172 to 162 per year.”
Unfortunately, it’s not so simple.
Dreaming the impossible dream.
The first problem with this prediction is that it’s not possible to pursue a policy of elective caesarean section for all term breech deliveries, even if you ‘convinced’ the 40% of women who choose to plan a VBB in the Netherlands to plan a CS. The researchers themselves noted that approximately 1:5 of the perinatal deaths observed occurred when breech presentation was not diagnosed until birth. Without instituting expensive changes to breech screening on a national basis, these outcomes will not necessarily be improved, certainly not without performing many more risky caesarean sections in advanced labour. (When these births were excluded from the analysis, the perinatal mortality rate for VBB’s which were actually planned was 1.3/1000 overall. The Netherlands has a high rate of home birth, so some of these unplanned VBB probably occurred at home with surprised, rather than prepared, midwives.)
Also, the researchers note that they have not performed an intention-to-treat analysis of their data. Multiple studies have noted that approximately 10% of women who plan caesarean sections go into labour unexpectedly before their scheduled operation, and 9.7% of the woman randomised to CS in the TBT gave birth vaginally. The researchers say that those for whom this was the case ‘could not be included in the caesarean section group’ for their study. Why? Were they included in the ‘planned’ vaginal breech birth group? The authors note this category was a ‘composite of vaginal delivery and emergency cesarean.’ Regardless of whether these women actually planned a VBB?
In an intention-to-treat analysis, the outcomes for babies who turn head-down spontaneously would also be included, as the decision to plan a VBB influences whether or not they will. In the Term Breech Trial, twice as many turned when a VBB was planned than when a CS was planned, so that 3.8% of all babies who planned a VBB were born in a cephalic position. Failing to do an intention-to-treat analysis disregards the complexity of breech decision-making and the full range of consequences.
Putting the figures into perspective.
The researchers note that 40% of women in the Netherlands choose to plan a vaginal breech birth, in collaboration with their doctors and midwives, and approximately 55% of them achieve this. This results in an overall perinatal mortality rate for breech presenting babies in the Netherlands of 0.7/1000, which happens to be the same perinatal mortality rate for low-risk women giving birth in hospital settings in the Netherlands (de Jong et al 2009; the mortality rate was slightly lower for women who planned a home birth at 0.6/1000). From where I am sitting, this looks like a good example of collaborative decision-making reducing risk while preserving choice.
Earlier this year, another team also led by Vlemmix (2013) published an abstract in the AJOG using further data from the Dutch nationwide perinatal registry from 2000-2007. This demonstrated that in addition to greatly increased maternal morbidity, neonatal mortality in pregnancies which followed an elective CS for breech presentation was 2.5/1000, compared to 1.3/1000 following pregnancies where a VBB was planned (which will be an average of the 2.5/1000 risk following a CS, and substantially lower risk following successful planned VBB’s). How then does this affect the prediction that elective caesarean section for all breech presentations would further significantly reduce the overall national perinatal mortality rate? Only if these women do not have any more children. Seems worth a mention to me, when you are recommending that all breech babies with a neonatal mortality risk of 1.3/1000 be delivered by CS.
Consider statistics given in the RCOG guideline on Birth After Previous Caesarean Birth. They summarise the data relating to term VBACS: “Planned VBAC is associated with a 10/10,000 risk of antepartum stillbirth beyond 39 weeks of gestation and a 4/10,000 risk of delivery related perinatal death (if conducted in a large centre).”
So a woman choosing to await spontaneous labour past 39 weeks and attempt a VBAC has a 1.4/1000 risk of losing her baby, but this is not only perceived as a reasonable decision, it is positively encouraged in most hospitals in the UK. Entire midwifery-led care pathways are set up to support women making this choice, and primiparous women to whom elective section for breech is recommended are proactively counselled that they can feel positively about attempting a VBAC the next time around.
But a woman choosing to birth her breech baby vaginally, knowing that she has a 1.3/1000 chance of losing her baby, is suspect. Does anyone else get the feeling we are robbing Peter to pay Paul?
The researchers identified no sub-classification of women for whom breech birth was more risky, or less risky compared to CS. They did observe that babies with a birth weight of over 3500 g (often excluded from VBB) actually only had a perinatal mortality rate of 0.8/1000. Contrary to what they have concluded, encouraging any woman who plans to have more children and wishes to attempt a VBB (at a relative PMR of 1.3/1000) as opposed to advocating universal elective CS for breech (and exposing subsequent children to double the mortality rate, at 2.5/1000), seems a measured approach.
The team’s representation of morbidity statistics also distorts the picture, as long bone fractures (included in their composite statistic) are common in cephalic births as well. They heal well, and are not generally considered ‘serious morbidity.’ Nonetheless, the morbidity rate of 22/1000 reported in this research was also significantly less than the rate of approximately 1/20 reported in the Term Breech Trial.
Designing research which meets the needs of women and their partners
The researchers have called for the results of their study to replace the information in their national patient information leaflets, which are based on the data from the Term Breech Trial, including the information that outcomes at 2 years of age did not differ between planned VBB and planned CS. One of the authors named in this Vlemmix study has also conducted research into what information matters most to women and their partners (Kok et al 2008).
Kok’s study (2008) demonstrated that women were mostly concerned with the safety of their baby and fear for a handicapped child, and that the 2-year outcome was what mattered most to them. Why then has this team conducted yet another study telling us what we already know – that short-term morbidity and mortality is significantly greater when VBB is planned in most cases – rather than robust research addressing what matters most to the families making these decisions? Whyte’s team (2004) emphasised that the 2-year results surprised them, as the group of children who went on to have handicaps at 2 years of age did not overlap at all with the group of children who experienced severe morbidity in the neonatal period. They were all born apparently healthy, and in the Vlemmix study would have been included in the ‘proof’ that universal CS is a safer policy. Given the numerous studies demonstrating a higher risk of cerebral palsy and other adverse outcomes for breech-presenting babies regardless of mode of delivery (O’Collaghan and MacLennan 2013), the question of whether the babies we save from death in the first 7 days go on to lead full and healthy lives is a fundamental question. And it is the concern which matters most to women.
On the other hand, according to Kok et al (2008) the concern which matters most to women’s partners is the outcomes for women. This study reported only 2 maternal deaths in the 1999-2007 period, despite acknowledging another study already published by the Dutch Maternal Mortality Committee reporting 4 deaths following elective section for term singleton breech in the Netherlands in 2000-2002 alone.
Why are the women disappearing?
If this study does not address the central concerns of women and their partners, why should it take precedence in the information given to aid informed decision-making? If the authors have not performed an intention-to-treat analysis, how can they possibly claim to know that their treatment will have the predicted result? If the study has demonstrated a risk similar to that of planning a VBAC, why are both choices not considered equally reasonable?
These authors are all very much involved with research concerning external cephalic version (ECV) and also feel the results of this research demonstrate the need to use ECV more liberally in order to reduce the incidence of breech presentation at term. While I am a fan of their ECV research and a proponent of ECV as a readily available option for women with breech presenting babies, again we cannot ignore the fact that ECV has not yet been demonstrated to improve neonatal outcomes, possibly for the same reasons the 2 year outcomes for children are not affected. However, ECV does significantly reduce the CS rate for women, particularly in centres where the option of VBB is not well supported. Again, I would like to see this highly experienced and highly influential team turn their attention to answering questions we do not already know the answer to, including whether or not manually turning a breech baby from a breech to a head-down position improves the short- and long-term outcomes for these babies, beyond increasing their chance of being born vaginally.
De Jonge, A., van der Goes, B.Y., Ravelli, A.C.J., Amelink-Verburg, M.P., et al. (2009) Perinatal mortality and morbidity in a nationwide cohort of 529,688 low-risk planned home and hospital births. BJOG : an international journal of obstetrics and gynaecology. [Online] 116 (9), 1177–1184. Available from: doi:10.1111/j.1471-0528.2009.02175.x [Accessed: 12 August 2014].
Vlemmix, F., Kazemier, B., Rosman, A., Schaaf, J., et al. (2013) 764: Effect of increased caesarean section rate due to term breech presentation on maternal and fetal outcome in subsequent pregnancies. American Journal of Obstetrics and Gynecology. [Online] 208 (1, Supplement), S321. Available from: doi:http://dx.doi.org/10.1016/j.ajog.2012.10.102.
Vlemmix, F., Bergenhenegouwen, L., Schaaf, J.M., Ensing, S., et al. (2014) Term breech deliveries in the Netherlands: did the increased cesarean rate affect neonatal outcome? A population-based cohort study. Acta Obstetricia et Gynecologica Scandinavica. [Online] 93 (9), 888–896. Available from: doi:10.1111/aogs.12449 [Accessed: 12 August 2014].