Back in November 2007, tigtog and I discussed at length the paper on the baby bonus introduction by Andrew Leigh and Joshua Gans, “Born (Again) on the First of July: Another Experiment in Birth Timing”. The paper has now been revised for international publication, this time with mortality data.
Their thesis back in 2007 was there was an introduction effect “delaying” births around the time of the baby bonus introduction, and their data does show a clear change in pattern.
The introduction effect has the potential to create potential issues with hospital resource management, an issue Gans and Leigh should have stuck to. However, they boldly branched out into saying that the “delays” were leading to “high birthweight” babies, and that this could only lead to adverse health outcomes. No alternative hypothesis was considered, and their assumptions weren’t elucidated or examined.
Gans and Leigh
hypothesisedbaldly stated in their 2007 paper:
Although our study cannot speak directly to the health effects of delaying births, it seems likely that decisions to delay births for non-medical reasons can only have adverse health consequences for babies and parents. While babies born pre-term and/or underweight are less likely to be healthy, the same is also true of babies born too late and/or overweight.
Health consequences for “parents”? For a start, what are the likely health implications for fathers and non-bio parents? OK, so they’re using “parents” as a proxy for “mothers”. Where do they present data on consequences for mothers? Nowhere. The hands are flapping as fast as they can.
On to the babies. The more sensible and obstetrically-experienced among us rapidly realised that what was more likely to be happening was that “routine” inductions and C sections, typically done around 38-39 weeks, were probably being rescheduled to a time closer to when the birthdate would have naturally lain had the interventions not been performed.
Most of these procedures are done for no urgent medical reason (the weekend effects shows that very clearly) – twins or breech presentation, “social” reasons, repeat C sections, presumed or fabricated macrosomia, and so on. For most of these situations, not only is there no medical reason to intervene so very early and/or no compelling medical reason to do a C section at all. In fact, in the absence of very urgent indications like impending eclampsia, there is often no reason to interfere to change the birthdate at all. Oh, apart from so that obstetricians can start their inductions first thing, run a money-spinning C section list in the morning, then catch or excise the induced babies in the afternoon, instead of attending births when they happen spontaneously and inconveniently.
I think the concocted obstetric workforce shortage could be playing a role here. Perhaps if there were wider recognition of the fact that obstetricians don’t need to be involved at all in most births, obstetricians would be less likely to get away with their “Oh, if we don’t schedule it now, I might not be available at all, because I’m so terribly busy” act?
On the birthweight issue: if you start holding off a little on unnecessarily early deliveries, it follows that the overall average birthweight will rise slightly, and more babies will slip into the >4000 g range (>8.8 pounds). That weight range is now considered a disaster waiting to happen by those who like to panic and slice, but many many completely normal health full-term babies fall into it, and most of them can be birthed normally without complications. In fact, one study found that the biggest determinant of birth outcome was not whether the infant was actually high birthweight, but whether the obstetrician predicted it was going to be. More on why projected “fetal macrosomia” is an overplayed hand in the birth-intervention game is here.
There ensued cross-blog discussions (at Hoyden (and here), at Leigh’s place, and at Gans’s place) on the underlying mechanisms and likely health implications, in which Leigh steadfastedly refused to examine his faulty assumptions – the assumptions that obstetric interventions were only done for pressing and clear medical reasons, and that any “delays” in the current pattern of births and deliveries was likely to lead to “overweight” babies and adverse health consequences. The press latched on to the idea of unhealthily fat babies – there’s a current moral panic buzzword for you! – and ran with it, repeatedly. (The Age, the West, news.com.au, ABC, ABC)
Now there is what looks to be a revised and expanded version of the First of July paper available [via the Social Science Statistics Blog at Harvard]:
This newest paper doesn’t seem to be obviously dated at that link, but from what I can deduce, it is the version in press at the Journal of Public Economics. [Note that it is not the original version of the Leigh and Gans July 2006 first-draft discussion paper , though they have the same title.]
In the 2007 version that we were debating, the hand-waving section on birthweight contained only one piece of medical background, a bit of fairly tangential Swedish medical data, a study finding a correlation – NOT a causative relationship – between very high birthweight and low Apgar scores, and post-term and low Apgars. [Apgar scores are a crude measure of neonatal well-being shortly after birth, and correlate somewhat with outcomes.] It contained no mortality or morbidity data from Australia, and more importantly, no other medical data, and no consideration of the background or implications of the one piece of data they did include.
Note that in the Swedish study, the Apgar slump with post-term birth did not occur until 42 weeks and after. When interventions for non-urgent reasons are typically scheduled at a presumed 38-39 weeks, the births can be as early as 37 weeks (or even earlier), as medical dating cannot be more accurate than that – and births at 38 weeks and before are also associated with a sharp rise in infant mortality in the Swedish data and elsewhere. The optimal birth timing in the Swedish data was at 39, 40, and 41 weeks. When routine C sections and inductions are held off for a week or even two, they are very unlikely to be “delayed” beyond 41 weeks, so the babies born with “shifted” birthdates around the baby bonus time were, in fact, very likely to have been born in the safest timing zone, as far as the Swedish study is concerned – very possibly even more likely than at other times.
On to the current paper and the punchline, if you’ve lasted with me this far. Gans and Leigh have confirmed that the shift in birthdates was driven by changes in induced and scheduled C section births:
“The largest drop in births in June occurred in vaginal induced births, and the largest rise in July occurred in Caesarean sections.”
They now have perinatal mortality data for the periods immediately before and after the baby bonus cut-off, the period during which births were so “dangerously delayed”. The result? Despite there being a 3% increase (*gasp!*) in “high birth weight” neonates, mortality was unchanged. This is a giant fizzog.
So women weren’t rampantly and recklessly harming their babies for money after all?
No-one appears surprised.
 Joshua S. Gans and Andrew Leigh. 2007. “Born (Again) on the First of July: Another Experiment in Birth Timing” The Selected Works of Joshua S Gans
Available at: http://works.bepress.com/joshuagans/15
 Joshua S. Gans and Andrew Leigh, “Born on the First of July: An (Un)natural Experiment in Birth Timing*”
In press, Journal of Public Economics
 Gans, J.S. and A. Leigh (2006a), “Born on the First of July: An (Un)natural Experiment in
Birth Timing,” ANU-CEPR Discussion Paper, No.529, Canberra. ” Available from http://econrsss.anu.edu.au/pdf/DP529.pdf