The Nebulous Nature of Risk: An Analysis of the Canadian Breech Study
There is risk and value to pretty much everything we do. It’s the balance between these that motivates decision making in every individual. And, obviously, there is no universality to the conclusion. It should not be expected that two people, given the same information, will always reach the same determination. Life experience and bias will always enter into the calculation. But we must always remember that in a free society the choice of what is risk and what is benefit belongs to an informed individual.
Selected vaginal breech delivery at term has remained an option for the 3-4% of women at term who face this dilemma. Because of some disputed evidence presented in a paper 15 years ago suggesting cesarean section is the safer method of delivery the vaginal option has rapidly declined and training of future practitioners has all but disappeared thus ensuring that the surgical route is likely to become the only option.
It can be very difficult for the experienced, let alone the lay person, to analyze the data and opinion pieces published since the Term Breech Trial first came out in 2000. That paper changed everything in regards to breech thinking and nearly eliminated vaginal breech delivery in hospitals around the world. Since that time there have been a number of papers and commentary that discuss the immediate risk to the newborn, many refuting and some supporting the earlier findings. What to believe and do became a choice of confusion and bias. Those that support the thought that “elective” cesarean section is best generally base their position by selectively highlighting the data showing increased immediate risk to the newborn. But risk does not exist in a vacuum! Pregnancy and birth exist in four dimensions and cannot, as such, be restricted by two or three dimensional thinking. None of these papers espousing the risks of vaginal breech delivery consider the short and long term risk of a scheduled cesarean section to the baby or the mother.
In the April 2015 edition of the Green Journal is a new original research article authored by Dr. Janet Lyons from the University of British Columbia titled, “Delivery of Breech Presentation at Term Gestation in Canada, 2003-2011”. The objective of this paper is: “To examine the neonatal mortality and morbidity rates by mode of delivery among women with breech presentation at term gestation.” The paper’s conclusion is: “Among term, non-anomalous singletons in breech presentation at term, composite neonatal mortality and morbidity rates were significantly higher after vaginal delivery and cesarean delivery in labor compared with cesarean delivery without labor.”
In order to accurately interpret any scientific paper we cannot simply look at headlines or conclusions. We must critically analyze the methods, results and endpoints used to make such conclusions. This is a difficult task for even the most well-informed reader but it must be done or we are left with research by press release. This is what drew me to this article in the first place. My colleague Emilee Benner sent me a Medscape article by writer, Jenni Laidman, dated April 17, 2015 and titled, Cesarean Without Labor Safest for Breech Delivery. While this headline may describe the findings in Dr. Lyon’s paper it tells us nothing of the legitimacy of the conclusions in that paper. It reminds me of the fallacy of words like “safer” and “riskier”. We must have more information to make reasonable decisions. What is the risk of doing something vs. not doing something? What do we surrender when always deferring to safety? How do we define safety? The safety argument is often misused as a means of coercion. As we have said, not everyone should be expected to define risk and safety the same way. Sometimes the risk of a choice may be greater but still small. Ultimately, we need to know specifically what those risks are in order to make an informed choice. For example, if the baseline risk of something is 1/10,000 and another method has a relative risk (RR) of 4, meaning it is 4 times more likely to happen, well, that sounds awful. Yet the risk is still only 1/2500 or 0.04%. For many, that is still a small risk. In statistics it matters what the denominator is.
I am an advocate of properly selected term breech vaginal birth and have been assisting women in this choice for 30 years. I am aware of the benefits of a vaginal birth both short and long term to mothers and babies. I have written about many of these topics in previous blogs. My experience with breech delivery contradicts the initial impression one would get from reading these headlines and abstracts. If the results of any study defy experience and common sense it must be questioned. Therefore, I have carefully read the research paper by Dr. Lyons, et al and what follows are some of my thoughts and conclusions. Hopefully, you will find them helpful in pursuit of reason and truth.
This paper is considered a Level II-2 study (Evidence obtained from well-designed cohort or case-controlled analytic studies, preferably from more than one center or research group). This is pretty good science and I thought the authors did a decent job of studying what they set out to study (My issue is that these parameters do not define all that is important in birth. In fairness, no study can do that. It is up to us to place it all in perspective. And that, I guess, is the key). It is a retrospective review of all hospital deliveries in Canada (excluding Quebec, which has its own separate system) between 2003 and 2011 taken from the Discharge Abstract Database of the Canadian Institute of Health Information. Patients included had to be term, non-anomalous, singleton and breech. Mode of delivery was divided into 3 groups: vaginal delivery, cesarean delivery in labor and cesarean delivery without labor (elective or pre-labor as Dr. Michel Odent prefers to call them). Here the author freely admits there could be significant crossover in the labored cesarean group including planned cesarean patients who entered labor early or undiagnosed breeches found while in labor.
The primary outcomes that were looked at were: Death, Assisted Ventilation, Convulsions and “Specific” birth injury potentially attributed to mode of delivery (which includes intracranial hemorrhage, other CNS injury, skeletal injury, nerve injury or injury to external genitalia). I found this grouping of injuries which were all lumped together to come up with a “composite” risk to be problematic. I could not rectify why they might include injured scrotum and potentially serious intracranial hemorrhaging in the same grouping for statistical analysis. This also combines transient morbidity such as assisted ventilation or a bone fracture with injuries that are much more severe and long lasting. Also, there is no explanation of how neonates who might have more than one morbidity were classified. Of importance is there were no neonatal deaths in the vaginal delivery group (8 year total of 1,593 births). To analyze their determined endpoints the authors used logistic regression to control for maternal age, parity and obesity. If you are like me you have no idea what this means. So I looked it up. Logistic regression is an accepted means of analysis that estimates the probability of an event occurring. But it has limitations which I defer to the internet to help explain:
Identifying Independent Variables
Logistic regression attempts to predict outcomes based on a set of independent variables, but if researchers include the wrong independent variables, the model will have little to no predictive value. For example, if college admissions decisions depend more on letters of recommendation than test scores, and researchers don't include a measure for letters of recommendation in their data set, then the logit model will not provide useful or accurate predictions. This means that logistic regression is not a useful tool unless researchers have already identified all the relevant independent variables.
Limited Outcome Variables
Logistic regression works well for predicting categorical outcomes like admission or rejection at a particular college. It can also predict multinomial outcomes, like admission, rejection or wait list. However, logistic regression cannot predict continuous outcomes. For example, logistic regression could not be used to determine how high an influenza patient's fever will rise, because the scale of measurement -- temperature -- is continuous. Researchers could attempt to convert the measurement of temperature into discrete categories like "high fever" or "low fever," but doing so would sacrifice the precision of the data set. This is a significant disadvantage for researchers working with continuous scales.
Independent Observations Required
Logistic regression requires that each data point be independent of all other data points. If observations are related to one another, then the model will tend to overweight the significance of those observations. This is a major disadvantage, because a lot of scientific and social-scientific research relies on research techniques involving multiple observations of the same individuals. For example, drug trials often use matched pair designs that compare two similar individuals, one taking a drug and the other taking a placebo. Logistic regression is not an appropriate technique for studies using this design.
Overfitting the Model
Logistic regression attempts to predict outcomes based on a set of independent variables, but logit models are vulnerable to overconfidence. That is, the models can appear to have more predictive power than they actually do as a result of sampling bias. In the college admissions example, a random sample of applicants might lead a logit model to predict that all students with a GPA of at least 3.7 and a SAT score in the 90th percentile will always be admitted. In reality, however, the college might reject some small percentage of these applicants. A logistic regression would therefore be "overfit," meaning that it overstates the accuracy of its predictions.
Granted, this is probably over most of our heads but it’s important to realize that science is not an exact science.
Over the 8 years of the study period there were 52,671 total breech deliveries which was 2.6% of all Canadian births (except Quebec). Interestingly, 45.4% were nulliparous but there is no mention of parity and mode of delivery. I guess, because logistic regression is supposed to correct for that. Still, a breakout by parity and morbidity would have been useful to determine if risks were greater for nulliparous women over multiparas. The total number of vaginal breech births in the first year of the study was 158 which rose slightly to 228 by the last year reviewed. The total number of cesareans in labor rose from 509 in year one to 573 in year 8. Pre-labor cesarean sections made up 88.6% of breech births in year one and 86.7% in year eight. The slight rise in vaginal and drop in elective cesarean delivery between year one and year eight is likely attributable to the 2006 statement of support for vaginal breech delivery made by the Society of Obstetrics and Gynecology of Canada (SOGC). Clearly, the statement did not make much of a difference and these small vaginal breech numbers (about 200 per year in all of Canada) cannot help much in the retraining of breech skills for future practitioners. One has to also wonder about the skill level of those that were assisting the term vaginal breech deliveries that made up the data for this paper. There is no mention of the qualifications of the practitioner nor is there any way to determine from the methodology if a vaginal breech delivery was planned or unplanned. The author admits to the assumption that the vaginal births were planned and that the pre-labor cesarean sections were planned. These are some of the same major flaws of the Term Breech Trial (TBT) and are inherent when using a retrospective review of discharge records. Thus, this paper is open to some of the same critiques used to discredit the TBT. See article by M. Glezerman, MD
I do not know what a researcher might find but I think a fascinating study proposal would be to do the exact same research as was done here in the Lyons paper but look at say the years 1983-1991. It is very possible Canada has this information. This was a time when breech vaginal delivery was routine, strict selection criteria had been established and training in breech was likely universal. I suspect the vaginal breech numbers were much higher and so were the skills of physicians at every hospital. I would hypothesize that we might find quite a different morbidity table. And if my hypothesis is correct then one might conclude that it isn’t the mode of delivery but the skill of the practitioner that directs the safety of term vaginal breech delivery.
The current Lyons study found that risks of adverse outcomes were greater for all women having a vaginal breech delivery or a cesarean deliver after labor has started when compared to cesarean section without labor. There is no differentiation from women who just started labor compared to those who were sectioned after a failed attempt at a vaginal breech birth or for fetal distress, for that matter. The overall neonatal death rate (none in the vaginal delivery group) in all breech deliveries was similar to vertex deliveries. Table 1 looks at neonatal mortality and morbidity of all term, non-anomalous breech babies compared to non-breech babies and found no significant differences which may come as a surprise to those who believe breech delivery is more dangerous. They did find that morbidity was higher for women in all breech groups beyond 40 weeks compared to the 37-40 week group. The great majority of non-labor cesareans were done before 40 weeks (86.2%) which may skew the safety numbers slightly. The higher composite morbidity rates in the over 40 week group translate into a “number needed to harm” of 41 for the vaginal delivery group and 50 for the cesarean with labor group. (The number needed to harm (NNH) is an epidemiological measure that indicates how many patients on average need to be exposed to a risk-factor over a specific period to cause harm in an average of one patient who would not otherwise have been harmed.) Another way to look at this number is that according to this study, using their parameters of morbidity, the safety of a vaginal breech birth at term beyond 40 weeks without injury is 97.4% and of cesarean with labor, 98%. That seems easier to understand. It also does not seem that frightening to me.
This is the quandary with statistics and using relative risk rather than actual risk. Some clients might think a risk of 1 in 41 is very high but if presented as 97.4% safe it seems better. Another way of saying this is although the RR of composite morbidity may be 5.39 times higher with a vaginal breech delivery beyond 40 weeks than a scheduled pre-labor cesarean section the risk is still small. But that does not make a good headline and so the deception and mistrust of science goes on. Also, over the study period there were 1,995,379 births in Canada with 52,671 being in the breech presentation (2.6%). Of the 52,671 women with term singleton breech presentations (not all of whom would have met standard criteria for a vaginal birth) only 1,593 (3%) succeeded in having a vaginal birth. Not only is that a pretty dismal record but how can Canada ever expect to retrain its future doctors and midwives when there is only an average of 200 successful breech vaginal deliveries per year in the entire country!
What this study doesn’t say is probably more important that what it does. The world definitely needs to re-teach breech! More cesarean sections cannot be a long term solution. All of organized medicine and the WHO agree that the rate is way too high. It would seem to me that poor outcomes as defined in this study can be directly correlated to limited resources and lack of skill. This study is not able to differentiate outcomes with skilled practitioners from those unprepared to assist a term vaginal breech delivery. If you review data from specific breech centers and my own clinical experience the results would be much different. Research like this is designed to look at a finite set of parameters. Application to the real world thus is limited. There is no mention here of the risks to mothers and babies from a pre-labor, scheduled cesarean section. Pre-labor cesarean section has been linked to obesity, endocrine dysfunction and altered microbiome. The welfare of the mother in this and future births is not considered. Neither short term risks nor the myriad of long term risks of surgical birth which are becoming more evident every year are considered. I do not fault the authors here as it was not part of their study but it is worrisome when news editors put out misleading headlines such as the Medscape article mentioned earlier where cesarean is proclaimed safest.
There are many limitations to any scientific study. It is impossible to control for all variables. The authors here admit to this and other weaknesses and I believe did not have any conformational bias when they set out to look at the numbers. Dr. Lyons concludes that composite risks were “significantly higher after a vaginal breech delivery and after a cesarean delivery in labor as compared with cesarean delivery before labor onset. The risks associated with vaginal and cesarean delivery should be carefully considered by women contemplating a singleton breech delivery at term gestation and by their physicians.” I agree that all risks and benefits of reasonable birth choices should be carefully considered by women and their families but her last few words are very concerning to me. Although Dr. Lyons does not and would not say vaginal birth is contraindicated her message will almost certainly be construed that way. The physician leaders in the current dominant medical community continue to defend surgical birth and do nothing about the disappearing art of breech delivery. This article notwithstanding, selected term vaginal breech delivery remains a very reasonable option in skilled hands and the expansion of the number of skilled hands should be our goal.
Stuart Fischbein, MD FACOG
J Lyons, MD et al link to abstract