My midwife teaches a childbirth education class that all of her clients are required to attend. We had our first class this week, and I am greatly looking forward to the next four. It is different from other classes in that it is geared toward homebirth and is a heavily research-based attempt to increase our knowledge of birth, homebirth, and the problems within today’s medical model of maternity care. In short, it is right up my alley. My midwife knows that I already know a good deal about these things (perhaps more than anyone else in the class), but I am no less excited to participate.
We spoke in our first class of the general positives and negatives of homebirth, and one class will be devoted entirely to ALL THAT CAN GO WRONG. In thinking of ALL THAT CAN GO WRONG I have been reminded of a study that came out last year, referred to as the Wax meta-analysis, that has been plastered all over the news as saying that babies are three times more likely to die during a homebirth than a hospital one. The study, however, is greatly flawed. Unplanned homebirths were included, statistics were poorly analyzed, incompetent software was used, etc. (Did I ever mention how I got into a disagreement about this with an OB during a panel discussion at the birth fair I volunteered at?) Here is a good critique of the study, which in fact came to the conclusion not that babies are three times more likely to die at home, but that “less medical intervention during planned home birth is associated with a tripling of the neonatal mortality rate.” Not quite the same thing really.
So why am I telling you about this? Because my midwife made it clear in our first class that she has dealt with the loss of babies. Nine out of 1700 have died, and I wanted to compare her rate to the national average. Using 1700 “births,” although it could be 1695 or 1737, I don’t know specifically, her death rate is 5.29/1000. I don’t know from what point in the pregnancy she counts it as a death–from 20 weeks? 28? during labor? for up to a week or month after?–so I based my comparison on the CDC’s rate of infant deaths of less than 7 days and fetal deaths with gestation of 28 weeks or greater, per 1000 live births and fetal deaths, as this seemed a reasonable comparison. The number is from 2005 because that is the most recent one the CDC seemed to have. And what is the national average for the United States of America? Drum roll please…6.64/1000!
I don’t think too many folks, i.e. my mom, are as concerned this time around, but it does make me feel better to know I’m in the hands of a caregiver (although I suspect the same is true for most homebirth midwives) whose perinatal death rate is lower than the national average, and by a whole baby. Not only then is there one more living baby under her care, but there are all the other advantages we listed in class along with it: no drive to or from the hospital; freedom to change positions, eat and drink, and have as many people there as you’d like, including the baby’s siblings; being in familiar surroundings with your own bed, bathroom and food (I’d add dog, too, because Scout was such a laid back labor companion); no IV (unless medically indicated) or continuous fetal monitoring (unless medically indicated); water, and I don’t mean to drink; no separation from baby after birth; no procedures done to your baby without your consent (like putting goop in their eyes “for the gonorrhea and chlamydia you don’t have,” as my midwife put it); no bright lights or time limits or “purple pushing” or warnings to not make so much noise; no induction or augmentation with Pitocin; lower risk of c-section with its longer recovery time and all the complications that can come with it (including, what is it, a four times greater risk of maternal death?); and the touchy-feely stuff like intimacy and sacredness. It was a long list.
So there you have it. It may seem as though I’m advocating for homebirth, but I am not. As always, I am an advocate of informed decision-making regarding maternity care.
P.S. Mama Raw has a FB page now!
20 responses to “Birth, death and statistical analysis”
Carin
June 21st, 2011 at 13:56
Why do you think the Michal et al paper is a good critique of the Wax et al study?
AMK
June 21st, 2011 at 16:08
Because it was linked to by an OB I follow on FB and because I could understand most of it.
Besides that, other critiques I’ve read only looked at single issues, say the computational tool used, while this one addresses multiple aspects of the paper. It also had, instead of MD’s, mostly PhD’s as authors, who are generally actually trained to do scientific research.
Carin
June 21st, 2011 at 21:52
Ok, so you think all the issues they raised are valid?
AMK
June 22nd, 2011 at 08:42
Speaking as someone who is not trained in research, I think the points they bring up are valid enough to make me question the Wax study and think that reading each of those individual studies might be a more reliable way of gathering information than just reading this one.
Carin
June 22nd, 2011 at 10:50
Ok. I have an issue with several of the points they brought up, but thought perhaps I was missing something. My brain is fried and while I’ve read other papers that have used meta-analysis before and know generally the strengths and weaknesses of using that approach, I’d have to put in quite a bit of work to actually assess the quality of the Wax paper -and- the quality of the critique and I don’t feel like doing that right now.
Anyway, I will say that while meta analysis has drawbacks, it can be very valuable in determining overall trends and is less influenced by local differences that can drive the outcomes of different studies. So I like that they tried it. I think it was a good idea.
AMK
June 22nd, 2011 at 11:39
I think the meta-analysis was a good idea, too. I’m not sure if others have been done on this subject or not–I haven’t come across any, but I haven’t been looking either. Based on your initial impression, what points did you have an issue with? Sometimes my midwife needs help assessing the reliability/validity of research related to childbirth, so I’d like to be better at it so I could volunteer to help, especially since she bases her suggestions to me on it…
Carin
June 22nd, 2011 at 17:29
Uff. I knew we’d get here. Again, I want to qualify this by saying that I’m very close to brain dead as far as analyzing papers the slightest bit outside my field goes. I can barely make it through a pre-teen fantasy novel right now let alone a technical paper. So without committing myself too much – the “paradoxical results” don’t look like a paradox to me. I understand what Michal et al are saying, just not sure why they assume Wax et al should report the neonatal mortality as a percentage of the whole population. It makes more sense (to me) to report it the way Wax did. Second, the numerical errors – meta analysis changes parameter values by weighting them based on the effect size (I think that’s the right term) of the study. This is a good thing – studies with larger samples sizes have a stronger impact on the overall result. So you shouldn’t expect to see the exact same numbers as reported in the studies themselves. So those are a couple of examples.
As far as getting good at reading the literature… all I can say is that if you’re not in the field and don’t have a statistical background, it’s really, really hard. Two personal examples: (1) I have 6 semesters of statistics, which is more than many phd students. I am often not comfortable assessing the validity of other people’s statistical analyses – especially if it’s not in my field. When the stats are relevant to me for whatever reason, I consult a statistician. (2) My lab studies stress hormones in birds. This is something I know about. I often read papers about testosterone in birds. Testosterone is also a steroid hormone, it’s made in a similar way in the body, is regulated in a similar way in the body, the methodology for measuring it is very similar. I’m not comfortable judging the merits of research that looks at testosterone in birds. I rely on input from another student in my lab who did her MS on testosterone in birds and who knows the literature better than I do. In both cases, that doesn’t mean that I know nothing or that my reaction to these papers is invalid -or- that I necessarily agree with the opinions of these people. It just means that there’s a shit-ton of stuff to know before you can make a good assessment and I don’t always know what I don’t know on a topic. What am I missing? What am I not thinking of? You know? I hope that’s not too discouraging and I sure don’t think it’s impossible for lay people to read primary literature – I definitely think that’s the best source for information … But I do think there’s a reason that people spend an average of 5 1/2 years before they’re qualified to get a phd.
/end ramble.
AMK
June 22nd, 2011 at 18:32
Hey, I took a semester of statistics, so I totally resent being called a lay person. (Ha ha). This reminded me that there was something recently I thought of sending you to read just for this reason, but I can’t remember what it was.
Just one question for now. I’m either misunderstanding or overlooking the part about reporting neonatal mortality as a percentage of the whole population. Where is that?
Carin
June 23rd, 2011 at 09:28
I wasn’t calling you a lay person, just generalizing.
If your class was similar to Jon Graham’s 2 semester intro course here, which -is- taken by a lot of undergrads, I’d think you’d do a pretty good job assessing the stats part of papers. Those 2 semesters are probably worth 4 of mine – he’s an incredible teacher. Anyway, I think high school students can and should be required to read and discuss scientific papers so it’s not like I think journals are like tomes of mystery that only the initiated can read.
Re: neonatal mortality… It’s the in the paradoxical results part. I think what they’re saying is that because neonatal mortality (birth to 28 days vs. ) is a subset perinatal mortality (20 weeks term to 28 days, Wax should report neonatal mortality as a percentage of that whole data set (the total population of births examined (which isn’t really a population, it’s a sample, but whatever), which would make it a smaller percentage than what’s reported. That’s the only way I can figure out the statement “According to the investigators’ definitions, these results (“neonatal death rates are actually far higher than the corresponding perinatal death rates”) are impossible”.
Jeez though, I feel like we could go through this paper line by line because they do go on to say that the paradox is a result of dramatic differences in outcomes of the studies included in the meta analysis. So maybe I’m misinterpreting. This is seriously why we read “big” papers as a group in my lab and why I join multiple journal clubs every semester. I often feel like a complete moron when I read papers, then I get together in my group and everybody is like, WTH were they trying to say???
AMK
June 23rd, 2011 at 13:20
I’m just kidding about the layperson thing…I really don’t remember much from my statistics class even though I did really well in it.
While you’re “brain dead,” I have pregnancy brain, so I may be way off on this and unable to explain what I mean but… I thought they were reporting both perinatal and neonatal mortality as a percentage of one whole data set. (I have to look at Wax to confirm). In which case the results could be accurate because there were dramatic differences in the outcomes of different studies. But even though they’re accurate they are not reasonable because, by definition, there couldn’t be more neonatal than perinatal deaths in the sample as a whole. So then you’re left with little information of practical value. And on this particular topic I think that’s pretty much par for the course because you could never ethically create an ideal study with women being randomly assigned to hospital or home births, etc. Which is why I’ve put less stock in any study and more in my midwives’ personal statistics and asking myself if I can live with a transfer rate of X% or a death rate of Y%.
AMK
June 23rd, 2011 at 16:50
I just realized I may have been making it more complicated that I needed to. I still haven’t looked at the Wax paper again, but even if the perinatal stats were drawn from one population and neonatal from another and the two were never combined into one larger one, everything else I said would still hold true (I think). Very different outcomes in the different studies could mean that the neonatal mortality rate appeared to be higher than the perinatal…
Carin
June 24th, 2011 at 10:01
I think we’re on the same page as far as understanding what Michal et al were saying, but I don’t necessarily agree with… I would have to go back to the Wax paper as well, which is more work than I’m willing to put in.
… ugh, you know what? I’m going to just have to look through both. Thanks a lot.
Carin
June 24th, 2011 at 10:02
This will take a while.
Carin
June 24th, 2011 at 11:48
Actually, I don’t think it will take that long. It actually looks pretty straightforward. But I took a second to actually process the numbers. And holy shit. No wonder people who are into home birth don’t like this study. 1:5 chance of mortality is pretty damn shitty. I’m going to have to go check out the other studies they reference.
AMK
June 24th, 2011 at 11:54
Yeah, I can’t believe that would be even close to the truth…
AMK
June 24th, 2011 at 11:50
LOL! I guess that means I have to look at them, too, huh? If we flush out anything of substance, though, I have a few ideas about what to do with that information…yippee!
Carin
June 24th, 2011 at 14:08
Can you get them? I can (probably) email you any you can’t get. The short answer is that they used different data sets to look at each rate. So I get better what Michal et al are saying now. They’re using results from different studies and generalizing to the whole population, but we still wouldn’t expect such large differences between perinatal and neonatal mortality… I wouldn’t call that a paradox per se, but wouldn’t mind a more thoughtful explanation from Wax. They attribute it to differences in risk in women that plan home vs. hospital birth. … But I thought they controlled for that. I’m seriously crying over here. I hate science. It’s too hard.
AMK
June 24th, 2011 at 15:06
Just got done with Wax and haven’t looked to see if I can get the others yet.
OK, yes they looked at different data sets. Got it. But still how could they fail to discuss the neonatal death rate being higher than the perinatal one? Ugh. Annoying.
One question, I don’t see anywhere that mentions a 1/5 chance of mortality, as you mentioned…
I also thought they were only looking at low-risk pregnancies, but then there’s this: “It [higher neonatal mortality in home births] is especially striking as women planning home births were of similar and often lower obstetric risk than those planning hospital births. The planned home delivery group commonly exhibited fewer obstetric risk factors such as excessive body mass index, nulliparity, prior cesarean, and previous pregnancy complications.” But then women with these “risk factors” can still be considered lower risk and therefore eligible for a home birth depending on whose definition you’re using (i.e. prior cesarean many times=high risk to OBs but not home birth midwives). But this is still vague…were breech babies and multiples included in the hospital groups…who knows?
Carin
June 24th, 2011 at 15:42
Re: 1/5, I thought they were reporting prevalence not percentages (see, I’m fried!). 0.2% isn’t quite as bad as 20% is it?
For the rest of it, have to go to the original papers! I think these guys did some clarifying as well.
AMK
June 24th, 2011 at 16:28
OK, yeah, I thought you might have misread that. No, .2% isn’t so bad when you really think about it. But when the blazing headline is that your baby is 2-3 times more likely to die, it’s annoying to me because I think most people don’t think about how small the risk actually is (whether that risk is .09% or .2% or whatever).
1 Trackbacks / Pingbacks
Why Women Shouldn’t Fear Homebirth « SmiffyBaby // Certified Organic Births May 21st, 2011 at 00:49
[...] Birth, death and statistical analysis (mama-raw.com) [...]