Planned C-Sections and Infant Mortality

The New York Times reports today on a study of over 5.7 million births in the US (between 1999 and 2001) that the neonatal mortality rate for Caesarean deliveries was 1.5 times that for vaginal births (at 1.77 deaths per 1000 live births). There is speculation only as to why the difference exists but it is statistically significant and certainly food for thought. One can only wonder how this interacts with evidence that suggests that birth timing is a choice made for reasons other than the purely medical (click here).

It is interesting to couple this report with one in The Age last weekend that about a third of all women who have inducements end up having emergency caesarians. However, it is not clear that this effect is statistically significant.

Its research showed that 38 per cent of first-time mothers who were 37 or more weeks pregnant with a single baby required emergency operations after medical intervention in their labour. Those who had emergency surgery outnumbered those who gave birth naturally (35 per cent) or with assistance using instruments (28 per cent).

Inducements are the other form of planned birth timing. One can only imagine that emergency C-sections carry even greater risks than planned ones.

6 thoughts on “Planned C-Sections and Infant Mortality”

  1. I’d suggest design artifact as a better explanation of this finding. The study excludes emergency caesers from its analysis. This dramatically deflates the mortality rate for planned vaginal deliveries, by excluding most such deliveries where something goes wrong (typically resulting in a change of plan to an emergency caesers.) Planned caesers, on the other hand, always result in caesers, whether everything is hunky-dory or not. To assess the effect of birth planning on mortality, the study should look at all deliveries, not just deliveries that go according to plan. The NYT headline “Voluntary C-Sections Result in More Baby Deaths” is a case study in bad reporting of data. It omits the crucial point: more baby deaths than what?

    On the other hand, the NYT does better than the article snypopsis (or this blog), by detailing the raw numbers. Out of the 5.7 million live births in the study, only 300,000 were in the ‘voluntary caeser’ group where the higher death rate was observed. The total number of infant deaths is 12000. Unfortunately, no-one does the math. That’s 9500 or so deaths from vaginal births and (wait for it!) 80 deaths from voluntary caesers. Maybe that’s statistically significant. But it doesn’t come close to passing the smell test.

    Add both the above points and I would be surprised if the study is measuring anything of significance at all. Not that that would stop it being published and publicised. The movement that deems anything ‘natural’ to be good (or godly or womanly) won’t hesitate to design studies to produce the result it wants and to allow claims to be made that wouldn’t get a look-in elsewhere.

    Like

  2. Oops. I totally got the maths wrong (and in a way that gave false support the point I was making. Read into that what you will.) 1.77 deaths / 1000 live births for planned caesers means 552 deaths out of 311927 planned caesers, not 80. Not so smelly. And I should learn to use a calculator.

    The number of deaths from vaginal births is unclear, as it seems that the 5.7 million live births in the study include both planned and emergency births (even though the latter, with the lion’s share of infant deaths, were dropped from the headlines comparison.) Neither the NYT nor the synopsis say how many vaginal births there were. Assuming the usual reported rate of 2/3 vaginal births per live births, there would be 2356 deaths amongst the 3.8M vaginal births, leaving 8989 deaths amongst the remaining 1.5M emergency ceasers (166 deaths per 1000 births.)

    The crucial question remains: how many of the 552 deaths from planned casers would have gone on to join the 8989 deaths from the emergency caesers if they had been planned as vaginal births instead?

    (And another thing, how do stillbirths compare for planned vaginals and planned caesers?)

    Like

  3. OK, my last post surely. I’ve used University of Melbourne’s limitless resources to get hold of the actual paper. This revealed that the NYT misdescribed the numbers. There were 5.76M vaginal births (i.e. rather than ‘live births’ as reported.) And the 11897 deaths (which were only from the vaginal deaths, not the total) were infant mortality, not neo-natal (which is what the study is about.)

    The correct raw numbers for neonatal deaths are 551 out of 311927 planned caesers and 3586 out of the 5.76M planned vaginals. The study found that there were 358 ‘excess deaths’ in the planned caeser group over three years. The question is what would have happened to this excess if the planned caesers had all been vaginals? The study suggests that they would not have shifted to the vaginal births if they had been planned vaginal births. But would they have disappeared or shifted to the emergency caesers (and hence left the study altogether?)

    Another interesting thing in the paper: women with previous caesers were excluded from the study. This has two consequences: (1) the study says nothing at all about VBAC. (2) I’m guessing that women who have vaginal births the first time are unlikely to plan caesers for subsequent births. So, the planned caeser group would have more first-time mothers (surely a higher risk group for neonatal deaths) than the vaginal group.

    Sorry about the multiple comments. It’d be nice if we could edit our posts!

    Like

  4. Back again! I just noticed the paper does distinguish between first-time and later-time mothers (primiparous and multiparous.) I was right that excluding prior caesers meant that vaginals were weighted towards later-time mothers (2 to 1) while planned caesers were weighted towards first timers (just over half.) But I was wrong about my suggestion that first-time mothers were high risk. Actually, for vaginals, there was no difference. For planned caesers, the proportion of deaths rose for later-time mothers. Do caeserians cause health problems that are especially more deadly for later-born kids? Or is this another selection artifact?

    Also, the cause of death stuff is interesting reading. There were disproportionately more congential defects (the main neonatal cause of death) in planned caesers than in vaginals. Obviously, electing to have a caeser couldn’t cause congential defects. The writers noted that if you exclude the congenitals, the finding of disproportionate deaths still holds. They certainly try hard to identify similar selection artifcacts (but are hamstrung by lousy risk data.) Surely there comes a point when you have to start questioning your whole design.

    Like

  5. We just critically analysed this article in journal club, and as Jeremy said, found that the methods used were less than ideal, to answer the aim. Excluding women who decided to labour and then had difficulties seems not to be valid, as if these women had decided to have a c-section they would have been included. Also, more information could have been given about congenital defects, as sometimes babies with congenital defects have to be born by c-section, and depending on the defect may also have a limited chance of survival.

    Although, I think that the main point here is that if journalists could read and critically analyse articles, it would be more helpful, rather than just parroting from press releases. All research is not equal, and a working knowledge of epidemiology could only improve reporting and the information that the public receives.

    Like

  6. I agree that the methods are less ideal, but you also have to take into consideration the type of delivery the mother-to-be wants regarding C-section (natural, epidural, etc.). It’s true that there should be really analyzable and conducive to bringing more insight on all factors related to this (including all known research from different sources), but that would take an enormous amount of time. I think that in the end, one would have to go with the research at hand and make a educated guess as to which stance one wants to take.

    Like

Comments are closed.