Wednesday, September 06, 2006

A new study raises more questions about homebirth

I was alerted to the following study by a homebirth advocate who was "shocked" that it showed a higher neonatal mortality rate for C-sections than for vaginal deliveries. The study itself is interesting, but it's implications for the safety of homebirth are even more interesting. The "shocking" neonatal mortality rate for C-sections in the study is actually lower than the rate of neonatal mortality at homebirth. So what does that tell us about the neonatal mortality rate at homebirth?

Infant and Neonatal Mortality for Primary Cesarean and Vaginal Births to Women with "No Indicated Risk," United States, 1998–2001 Birth Cohorts
Marian F. MacDorman, PhD, Eugene Declercq, PhD, Fay Menacker, DrPH, CPNP, and Michael H. Malloy, MD, MS, Birth
Volume 33 Page 175 - September 2006

The authors of the study (2 of the 4 are actually editors of the journal) seem to think that the study's results are surprising because the neonatal death rate in the C-section group (1.77/1000) is higher than in the vaginal delivery group (0.62/1000). They do not say it explicitly, but they seem to imply that the C-sections somehow lead to the increased mortality. That's not at all what the data shows. Furthermore, there are some serious problems with the way that the study itself was conducted.

As the title of the study indicates, the authors set out to investigate neonatal mortality in a low risk group. They determined risk by looking at birth certificate data.
The 1998–2001 birth cohort national linked birth/infant death data sets were analyzed to examine infant and neonatal mortality for women with no indicated risk. These data sets link the birth record to the infant death record for each infant who dies in the United States. The purpose of the linkage is to use the many additional variables available from the birth certificate for infant mortality analysis. Information on all of the approximately 4 million live births in the United States each year is also included...

For the purposes of this analysis, women with no indicated risk were defined as those with singleton, term (37–41 weeks' gestation), infants in vertex presentation who were not reported to have any [of 30 specific] medical risk factors, and for whom no complications of labor or delivery were reported on the birth certificate... The study was further restricted to women who had never had a previous cesarean,..
The authors then looked at the data that they derived, and calculated a neonatal mortality rate of 0.62/1000 for vaginal deliveries and 1.77/1000 for C-sections. They express concern that the neonatal mortality rate was almost 4 times higher for C-sections than for vaginal deliveries. However, that is precisely what you would expect. Most reasons for C-section fall under the heading of protecting the baby. You would not expect very many if any deaths in the vaginal delivery group. Almost every baby in the C-section group is there because an indication for C-section arose during the course of labor. Clearly, the C-section group is almost by definition a higher risk group.

Let's take a step back and look at the actual neonatal deaths rates. They are exceedingly low and are a testament to the stunning success of modern obstetrics. I have not seen lower neonatal death rates in any other studies of the US. So the rates themselves are cause for satisfaction. They could go lower still, and we should always be working to lower them to zero, but they appear to be the lowest they have ever been.

As I said above, there are serious problems with the paper which render the conclusions suspect. The first problem is a substantial methodologic problem and the authors actually acknowledge it within the paper. Here's what they say:
"underreporting of individual medical risk factors and complications of labor and delivery on birth certificates has been documented. Our measure of "no indicated risk" ... included only those births where none of the more than 30 items on risks or complications was reported... [I]t is possible, based on either poor reporting or because the risks involved items not recorded on the birth certificate, that the cesarean group was still an inherently higher risk group and those risks accounted for both the decision to perform a cesarean and the neonatal death."
So, the authors themselves admit that they put people into the no risk group because they did not find one of 30 different risks noted on the birth certificate, and that the recording of these risks is known to be unreliable. Therefore, it is almost certain that the C-section group was of higher risk than the authors have surmised.

The second problem is that the vast majority of the deaths are due to congenital anomalies. Indeed a full 55% of the deaths in the C-section group are due to "congenital malformations, deformations and chromosomal anomalies." When congenital anomalies are removed, the neonatal death rates for vaginal delivery drop to 0.33/1000 and for C-section 0.76/1000. Obviously, none of the babies with congenital anomalies were low risk.

Third, the death rate from asphyxia is much higher in the C-section group. The death rate for asphyxia is 0.02/1000 in the vaginal delivery group, and 0.15 in the C-section group. The rate of dying of birth asphyxia was 7.5 times higher in C-section group. That is conclusive evidence that the C-section group contained higher risk pregnancies. These were the babies that developed substantial problems during labor. This suggests that the C-sections should have been done sooner, and it certainly does not indicate that the C-sections should not have been done or that the C-sections caused the neonatal deaths.

Fourth, the 2nd most common cause of neonatal death (after congenital anomalies) is "other". Since the authors don't tell us what these babies died of, we have no way of knowing whether they are related to having a C-section or completely unrelated.

The study was designed to give us information about whether the high C-section rate increases the neonatal death rate. However, this study tells us precisely NOTHING about the risk of neonatal death DUE to a C-section. Indeed, the authors do not identify even one case in which a C-section was implicated as being responsible in any way for the death of any baby.

So, the study does not do what it was designed to do, and the conclusions are not justified by the evidence presented. However, I still find the study to be very interesting for what it tells us about hospital birth compared to homebirth. If you have read any of my critiques of the Johnson and Daviss study (BMJ, 2005; linked in the sidebar), you will know that one of the most substantial failings of the study is the use of improper controls. The homebirth group consists of almost all planned homebirths in the year 2000. The appropriate control group would have been all low risk white women with singleton, vertex babies who delivered in the hospital in the year 2000. Here is that control group! Actually the group in this study is of slightly higher risk than the homebirth group since the hospital group contains women of all races, but we will set that aside for the moment.

If you use the neonatal mortality rates from this group as the control group in the J&D study (instead of control groups from previous studies done much earlier), homebirth by midwives has a much higher rate of neonatal death than hospital birth. If you remove congenital anomalies (as the J&D did) the death rate in the homebirth group was 2/1000 and in the hospital group was 0.31/1000. So, far from being comparable to the hospital death rate, the neonatal mortality rate in the J&D study of homebirth was almost 7x (650%) higher than for hospital birth of low risk women of any race that occured at in the same time frame as the study!

So not only does this study fail to tell us anything about neonatal deaths due to C-sections, it is yet another confirmation that homebirth is not nearly as safe as hospital birth.

0 Old Comments: