It's an interesting paper, but I don't (personally, based on my limited understanding of this type of study) think it's conclusive. There were significantly fewer mothers in the home birthing and birth center groups. They were also quite a bit older and more likely to be 1-2 weeks past their due date, which could correlate with more negative outcomes. The authors seemed to draw conclusions about CNM and physician-attended home births, but I did not see the data for those births separated out (although I may have missed it). Several of the authors work at teaching hospitals with high-risk populations and equally high (35-38%) c-section rates, which may have colored their views on the safety/naturalness of birth in general. Also, according to ACOG, up to 25% of U.S. home births are unplanned. The authors said they attempted to exclude unplanned home births, but that leaves room for subconscious or conscious bias to affect the data. I'd like to know whether they compared data from states where home birth is illegal with states where it is legal, and if so, if there were differences in birth outcomes. I did find it interesting that hospital births attended by midwives had the best outcomes. (I wonder how births attended by both midwives and doctors are recorded on birth certificates?)
It's worth noting that there are other possible negative birth outcomes beyond the (very serious) ones studied here...
I don't think I'm ready (or ever would be ready) to say one type of birth is always better than another for all women. I'm glad research is being done with the goal of improving birth outcomes for women, but I don't think the "either you care about your baby or your own 'experience'" tenor of some studies does much to improve the level of discourse.
Thank you so much for this thoughtful reply!
I think there weren't any physician-attended homebirths in this study, but I'll double check. One thing I didn't like about the study, though, is that it didn't/couldn't separate "lay midwives" (with no training) from "midwives with actual degrees" (CNMs). But it did note that CNMs in hospitals did just as well as physicians, as you said, suggesting it's the being-at-home part that causes the difference.
I absolutely agree that there are limits to this study--9/10 women in their study chose hospital birth, which is probably unavoidable since that's most likely a representation of the population in the US. Also, any homebirths that were referred to the hospital (the paper said
45% of the homebirths were referred to the hospital!) ended up counting as hospital births. Both of those things could skew the data, although they skew it in different directions.
I think the age difference between the two groups is also unavoidable, because an 18 year old mother is less likely to feel free to choose whatever birth she wants. Older mothers are just more likely to choose a home birth. You're right that it could be part of the issue. But does that then suggest that older mothers should be especially cautious about homebirth?
I completely agree that choosing an APGAR score of 0 (which means no signs of life in the baby) does make the data more dramatic. I don't think this is a limitation of the study, though, because when you are studying
13 million data points (as they were) you have to put some limits on it or you'll be drowned in information. I think the authors are trying to avoid shades of gray, so by choosing the absolute worst outcome, they're more likely to actually see data that is statistically significant.
I'm not sure I agree about possible bias. I think the 13 million data points helps erase any bias that the authors may have had. Plus, I think only 3 of the authors were actually MDs, the rest were PhDs and so would not have any interaction with actual patients regularly. But I'll have to think about that more.
One other thing to note, there is a 10x greater risk of APGAR score of 0, but i think the risk went from something like .001% (so 1 in 100,000) in a hospital to .01% (so 1 in 10,000) at home. Which is
really tiny risk, but still a 10x increase.