Oh Dear, here we go again…

This is a joint statement from ACOG and the SMFM. There is some good in here, but you’d think they could at least get the facts right.

“Delivery before 23 weeks of gestation typically leads to neonatal death irrespective of newborn resuscitation (5-6% survival)”.

This is just nonsense. What does this mean, that if you don’t resuscitate you will get the same survival as if you do? Survival is 0% without resuscitation and between 21 and 40% if you do. What is the point of mixing figures from patients with active treatment, and those without? So that you can say to an expectant mother, “in a mixed group of infants, some of whom received active care and others did not, (but we aren’t going to tell you how many) 5 to 6% survived”?

The references they give for that percentage show what can happen if you provide active treatment. Survival in Stoll BJ, 2010 at 22 weeks was 21%, in Rysavy et al it was 23%, among actively treated infants. Other figures show similar things in even more detail. The EXPRESS group in Sweden have been leaders in this, their article in JAMA in 2009 shows that in 51 infants of 22 weeks (there were 2 babies of slightly less than 22 weeks included) there were 19 who were admitted to the NICU for active care, 5 survived, or 26%. Including the infants who were born alive but not resuscitated decreases the survival to 9%, but telling a mother that the risk of survival is 9% makes no sense at all.

You can only give informed consent if you have not been misinformed! The chance of survival is either 0 or 26% from those figures.

I was recently “accused” in a public meeting of wanting to resuscitate 100% of live born babies. Which is untrue. But I do want 100% of mothers to be able to make decisions based on accurate information, based on sensitive personalized counselling. To make decisions that are best for them and their family, which they cannot make unless they really know what the options are. Which means good data.

Lets return to the statement, and another inaccurate piece of information:

“Among rare survivors significant morbidity is universal (98-100%)”.

Although morbidity is not defined here, I presume this is referring to adverse long-term outcomes, as they are the only things discussed in this statement. If so it is a gross exaggeration, the 2 references given show no such thing. Rysavy et al showed 61.1% ‘significant morbidity among survivors’ and Stoll et al contains no data on long term outcomes. It does show that all infants born at 22 weeks are sick, which should be no surprise, and surely should not be confused with data about long term neurological and developmental prognosis.

In fact Rysavy et al’s paper is grossly misquoted in Figure 2. The percentages shown are completely wrong.

Also it is ridiculous to mix together, in Figure 2 the following :

Greg Moore’s systematic review of outcomes at 4 to 8 years which defined moderate and severe impairment as an IQ less than 70 and Cerebral Palsy with a GMFCS of 2 or worse.

Rysavy et al’s data which examined babies at 18 to 22 months and defined moderate and severe impairment as a Bayley3 less than 85 and CP with a GMFCS of 2 or worse

Two publications with data from different tests at 2 different ages from the original EpiCure dataset. Marlow 2006, for example, defined severe disability as a cognitive score less than 70 and non-ambulatory CP, at 6 years of age.

Ishii from Japan, who at 3 years defined profound impairment as cognitive score less than 50 or “profound CP”

What possible use is a figure that mixes all those together, and then completely misses the mark with one study (Rysavy) by showing outcomes among all babies delivered, including those who were not offered intensive care.

This is another statement that should be junked. At least get your literature review right first. Then you can start on the value-laden judgements about when to intervene.

And if the whole point is to make an individual decision, based on the families values and interests (which is the message being given by tables 1 and 2), then why on earth do you need the completely inconsistent table 3? Table 3 which then goes back to the standard old-fashioned approach making decisions according to completed weeks of gestation. So for example antenatal corticosteroids are “Not recommended” up to 22 weeks and 6 days, but you can consider them at 23 weeks and 0 days, up to 23 weeks and 6 days, when one day later they become recommended.

How “Not recommended” was graded as a recommendation 1A which means, according to their system “strong recommendation, high-quality evidence”, is completely beyond me. What evidence is there of harm? Which is what you need according to their own framework to make that statement. Their own brief review shows no harm, although there is no proof of major benefit at 22 weeks, and suddenly at 23 weeks and 0 days there is a reduction in death, severe IVH, NEC and PVL.

It should be “strong recommendation, entrenched prejudice”.

The best you can really say is that whether or not steroids help at 22 weeks is uncertain, and there is currently no evidence of harm. An honest evaluation of the literature should lead to a recommendation of “we don’t know”. Which would mean discuss this with the mother.

Which is really what a statement regarding counselling should be all about. How to understand the evidence, the limitations of the evidence, how to explore the values, hopes and desires of a woman in distress, and her partner if there is one. How to come to a decision that is within the bounds of what is reasonable, and to take into account the interests of the baby.

I started off this post by saying that there is some good here. I think that many of the issues addressed in those first 2 tables are spot on. There is a recognition of the poor accuracy of gestational age assessment, a brief note regarding the limitations of survival figures, an acknowledgement that there is no good evidence that antenatal risk calculators affect decision-making or improve satisfaction. There are some general recommendations about how where and by whom counselling should be done, and about taking into account other factors than gestational age which alter outcomes. They then seem to ignore most of their own advice in table 3.

My main issue is that to make a blanket recommendation for all women at 23 weeks and 4 days, or at 22 weeks and 3.24 days, is inconsistent with good medical practice, and is inconsistent with the initial paragraphs and the first 2 tables. After recognizing the limitations in the evidence, the limitations in gestational age assessment, the uncertainty of other risks, and the variations in individual values and preferences, then steroids may or many not be a reasonable intervention for a particular mother (for example).  To say that would be much more useful, I think than, at 24 weeks and 0 days do X (or consider X, or don’t do X). It might not make things cut and dried and easy for the Obstetrician, or MFM specialist, but it ain’t supposed to be easy. Oh and next time talk to a neonatologist, someone who understands a bit about perinatal epidemiology, and someone who understands about long term follow-up.