Two new papers raise serious questions about banning elective deliveries before 39 weeks


I’m a mainstream obstetrician. Most of my views about obstetric practice are in line with that of other obstetricians, pediatricians and neonatologists. In one area in particular, though, I am bucking the conventional wisdom, and that is in my oppposition to the hard stop policy banning elective deliveries before 39 weeks gestation.

I realize that the March of Dimes believes strongly that this is an effective and necessary policy (and a lot easier to accomplish than trying to prevent the prematurity that actually kills babies), and I know that the American Congress of OB-GYNs (ACOG) and the American Academy of Pediatrics (AAP) have enthusiasticly climbed on the bandwagon. And almost every time I write about this issue I get emails of other obstetricians (including regular readers of this blog) disagreeing with me.

But I view this issue as a subset of a serious problem plaguing American medical practice: the implementation of “preventive” care guidelines (particularly those that promise to save money) in the absence of evidence to support those guidelines. As I’ve written in the past, I’m afraid of preventive medicine. From routine estrogen replacement therapy to routine use of prostate cancer screening, preventive measures have been implemented without appropriate, large scale, long term studies to determine unanticipated side effects. I’m afraid that the 39 weeks ban will also turn out to have serious side effects, and in this case, we can’t even claim that they were unanticipated.

Why should we anticipate the serious consequences of banning elective deliveries before 39 weeks? Because we know (as demonstrated by the chart below) that each additional week of pregnancy beyond 36 weeks raises the stillbirth rate.


I’m not the only obstetrician to point this out. Indeed two new papers address this issue specifically, one in a theoretical argument, the other using the results of a study.

The first paper is Theoretical and Empirical Justification for Current Rates of Iatrogenic Delivery at Late Preterm Gestation by Joseph and Dalton, published in the most recent issue of Pediatric and Perinatal Epidemiology.

The authors point out that the 39 week ban is based on 3 erroneous beliefs:

1. The erroneous belief that the difference in death rates between babies born at 34 weeks and babies born at 39 weeks is merely due to gestational age.

Numerous studies have quantified the excess morbidity and mortality among late preterm infants compared with term infants. Although such quantification accurately reflects differences between the two groups due to differences in pregnancy duration and pregnancy complications, it is disingenuous to suggest that a pregnancy with evident fetal compromise at 34 weeks gestation could be safely delivered at term.

2. The erroneous belief that “too many” babies are born before 39 weeks.

Expectant management given fetal compromise at late preterm gestation is associated with a potential risk of fetal demise, neonatal death or serious neonatal morbidity (due to progression of the fetal compromise). On the other hand, iatrogenic late preterm birth given fetal compromise is associated with a potential risk of neonatal death or serious neonatal morbidity (especially respiratory morbidity due to lung immaturity). Studies show that recent increases in iatrogenic preterm birth have been associated with declines in perinatal mortality. To our knowledge, no population-based study has demonstrated an increase in rates of neonatal mortality or respiratory morbidity due to the recent increases in iatrogenic late preterm birth.

It is also noteworthy that about one-third of iatrogenic late preterm birth is carried out for maternal indications.Yet no study to date has examined the effects of increases in iatrogenic late preterm birth on maternal health status…

3. The erroneous belief that multiple studies show that unindicated premature deliveries are rampant.

… These studies have been criticised because of their weak retrospective design; two studies were based on retrospective abstraction of medical charts and the third was based on a national database with information from birth certificates (known to overestimate non-indicated labour induction). The lack of detail regarding the clinical context makes judgement regarding the appropriateness of iatrogenic late preterm birth in these studies uncertain. The absence of an indication in the medical chart could imply an elective delivery or could represent a problem with the documentation of a legitimate indication.

So the purported theoretical basis for banning deliveries before 39 weeks is very weak.

What happens to the stillbirth rates if such bans are implemented? That’s the question addressed by the second paper, The risk of fetal death: current concepts of best gestational age for delivery by Mandujano et al., published in this month’s issue of the American Journal of Obstetrics and Gynecology.

According to the authors:

Linked birth and infant death data for the US from the National Center for Health Statistics analyzed nonanomalous singleton pregnancies between 2003 and 2005. Pregnancies were classified as high risk or low risk based on preexisting maternal complications. Out- comes of 8,785,132 live births and 12,777 FDs between 34 and 42 completed weeks’ gestation were examined…

What did they find?

Between 34 and 40 weeks’ gestation, the FD [fetal death] risk of those remaining undelivered for all pregnancies declined and then increased at term. For high risk pregnancies, the FD risk of those remaining undelivered is substantially higher than for low risk pregnancies. The number of FDs that can be avoided by delivery exceeds the neonatal death rate between 37 and 38 weeks’ gestation in low risk pregnancies and at 36 weeks’ gestation in high risk pregnancies.

The inevitable conclusion is:

These findings suggest that delivery at 39 weeks’ gestation in both high and low risk pregnancies would result in an increased number of perinatal deaths. Decisions regarding the “optimal time for delivery” should include the risk of remaining undelivered.

The authors note:

Much of the na- tional conversation and literature on this subject have surrounded the neonatal morbidities associated with a delivery before 39 weeks’ gestation. Although these analyses have demonstrated that delayed delivery reduces neonatal morbidities and the subsequent neonatal mortality from prematurity, they failed to include stillbirth in their analysis. We hope that by directly comparing fetal and neonatal mortality, we highlight what must be considered when determining optimal GA for delivery: both the risk of delivery and the risk of non-delivery. Recommendations that consider only one element should be considered incomplete.

That’s the very point I’ve been striving to make in multiple posts I have written on this issue. The underlying assumption of banning elective delivery before 39 weeks is that morbidity can be reduced without increasing mortality from stillbirths. As these two papers show, that is an assumption that is entirely unjustified. It is probably impossible to reduce morbidity without increasing mortality from stillbirths. As between the two, a short NICU admission is far preferable to a preventable perinatal death.