I am disappointed and very, very frustrated.
Scientific journals continue to publish more and more junk. Do the editors read what they publish? Do they think about whether it is true? Or do they simply print it, send out a press release and wait for attention?
The latest piece of junk published in a reputable scientific journal is Publicly funded homebirth in Australia: a review of maternal and neonatal outcomes over 6 years, just published in the Medical Journal of Australia.
I reviewed the findings when they were first presented at a medical conference in April (Australian midwives boast about terrible homebirth death rate):
During the 5 years of the study, there were 1807 women who intended, at the start of labor, to give birth at home. 83% had a homebirth, 52% in water (I have no idea why they mention this except to check women’s performances against the midwifery ideal.) The transfer rate was 17%. The C-section rate was 5.4% and the neonatal death rate was 2.2/1000. That’s more than 5X the rate of 0.4/1000 found in a 2009 report on birth in South Australia.In addition, 2 babies suffered hypoxic ischemic encephalopathy (brain damage due to lack of oxygen).
And that probably undercounts the deaths and complications because reporting was voluntary and only 9 of 13 program directors responded. Nonetheless, the authors conclude:
This study provides the first national evaluation of a significant proportion of women choosing publicly funded homebirth in Australia; however, the sample size does not have sufficient power to draw a conclusion about safety. More research is warranted into the safety of alternative places of birth within Australia.
Actually, the study is not underpowered to detect an extremely high death rate.
What is statistical power?
In plain English, statistical power is the likelihood that a study will detect an effect when there is an effect there to be detected. If statistical power is high, the probability of making a Type II error, or concluding there is no effect when, in fact, there is one, goes down.
Statistical power is affected chiefly by the size of the effect and the size of the sample used to detect it. Bigger effects are easier to detect than smaller effects, while large samples offer greater test sensitivity than small samples.
In most studies we find very small differences between the two groups under investigation. Therefore, we need a lot of individuals in each group in order to be sure that the difference we have found is real, and not the result of chance.
In contrast, if we find a very large difference, we don’t need a lot of individuals in each group in order to be sure that the result is real. A 400% increase in the death rate is an extremely large difference.
The authors never bothered to conduct a statistical analysis of any kind, which means that they literally have no idea whether any of their claims are valid. They simply announced that they could make no determination of safety, but nonetheless boasted about excellent outcomes. You can’t have it both ways. Either the study has too few individuals to draw ANY conclusions, in which case the entire paper is meaningless, or the study contains enough individuals to provide a meaningful result.
Caroline Homer, one of the authors of the study, and Hannah Dahlen, a spokesperson for the Australian College of Midwives, take to the lay press to boast about the results of the study (Study of low risk women reveals good news on the home birth front):
Hannah Dahlen, Professor of Midwifery at University of Western Sydney, said the findings we “very reassuring” and showed a very low perinatal mortality rate, comparable with birth centres.
That is an utter falsehood.
The study shows a VERY HIGH neonatal mortality rate, 400% higher than comparable risk hospital birth.
Which raises the question: Is Dahlen deliberately trying to trick readers, since a neonatal mortality rate of 2.2/1000 is 5X higher than comparable risk hospital birth? Or are she and the authors of the study so ignorant of childbirth safety statistics that they don’t realize that the homebirth death rate 400% higher than comparable risk hospital birth?
And what about the MJA?
Why did they publish such a misleading paper? Why didn’t they insist on a discussion of the very high death rate? Why did they allow the authors to declare that the study is underpowered to determine safety when they authors did no statistical calculations of any kind? If the study is underpowered, why did they bother to publish it?
The publication of this study is disappointing and very, very frustrating. The very best we can say about this paper is that it is utterly misleading.
As I said above, I don’t know if Hannah Dahlen and Australian midwives are trying to trick the Australian public into believing that homebirth is safe when it clearly is not, or whether they are so ignorant of basic science, statistics, and mortality data that they don’t realize that have shown that homebirth is dangerous.
It doesn’t really matter. Boasting about a hideous death rate is both bizarre and unacceptable.
Wow!! I accidently stumbled across this blog and I’m amazed at your viewpoints Dr Amy! It seems you feel threatened by midwives, or something has happened in your career to make you very angry and subscribe to a culture of fear. I work in Australia and am very proud that we celebrate midwifery and obstetric care and respect women, their babies and families the way we do. I have no idea if that ‘one born every minute’ show in the US is representative of birth in the US, but it was completely shocking the way medical intervention was so routine in uncomplicated births and women who refused such interventions were disrespected. From what I can tell birth in your country seems to have been made into a medical problem/procedure/legal issue. Hopefully in Australia we can keep the US culture of fear away so we can continuing to promote childbirth as a normal physiological process.
We are running a much better system with far better perinatal outcomes than the wonderful USA. Please don’t ever move here to work.
I am definitely bookmarking this page and sharing it with my
friends.
I am glad to catch idea from your article. It has
information I have been searching for a long time. Thanks so much.
This doesn’t surprise me in the slightest. I’m an Aussie and during a home visit after my last birth, the midwife asked why I hadn’t tried for a homebirth. I was stunned and replied “Did you actually read my file?” as I’d required an IV just in case things went pear-shaped (low B12 and low iron and was told that I’d have needed a transfusion if I’d lost more than 500mL), needed a little boost of pitocin for that one, and my babies tend to be Persistent Occiput Posterior Position deliveries. There is no way that I would willingly birth outside of a hospital because I feel much safer knowing that help is readily available if we needed it.
I am going to put my flame suit on here, but I have to disagree with a few things! This is unusual, as I am a daily visitor to this website and work in the field (medicine and childbirth that is, not home birth!)
OK-so I have only briefly scanned the article, but I think it has been reported in a responsible fashion. The abstract which most people will read clearly states that they cannot comment on safety and with there numbers I don’t consider it valid to perform a statistical comparison either. To do this it needs to be done “a-priori”, ie designed from the outset to be a case controlled study. This was purely an observational study, its not truely research, more an audit of their outcomes. It is not valid in these circumstances to perform statistical comparisons.
You can be critical and say they should have potentially done a matched case control study from the outset, but this is a much bigger undertaking and realistically is not feasible unless you have massive funding-for each homebirth you would need 1-4 controls.
The MJA is the Journal of the Australian Medical Association, a very staunch anti-home birth group of doctors. You can tell reading the article that it has been well reviewed and toned down and kept relatively neutral.
So, I actually have no problem with the way they have reported the data from this study. How you then assess and use that data is open to debate. For instance, 3% of data was not included from the 9 centres providing data, just 1-2 deaths in this 3% could seriously impact on their safety. What of the 3 centres that did not report? These aspects are beyond the control of the investigators and they have handled this well. Whether they should then promote this article in the lay press as showing how safe homebirth can be is another question.
So, as a skeptical ob fanatic, it is unusual for me to disagree, but in this case I do!
I’m a ob fan too and proud aussie too but I can’t quite see how you can disagree. ( although of course you have every right to!) The authors don’t appear to know that power is irrelevant in their case and they also don’t seem to be aware that their death and morbidity rate is anything but safe given they had a very low risk population .
I see it a bit differently – I can’t see how a paper that didn’t look at causes of death, or comment on the neonatal death rate in comparison to low-risk hospital birth, made it to publication.
The AMA has its conservative aspects, but I’m not sure the MJA could be considered conservative these days.
Perhaps someone in the health / sciences can submit a Letter to the journal, pointing out these issues with the analysis? Is that a possibility?
I read about that study yesterday and was wondering what you would make of it.
I’d vote so entrenched in a belief system that they are unable to acknowledge anything that challenges that belief system.
You could say the same of Dr. Amy….
Which belief system?
How so? I don’t see Dr. Amy as being entrenched in a belief system. She is against natural childbirth and breastfeeding being *shoved* upon women, but she is not against them per se. Here are some examples:
1. She speaks out against those who would tell a mother lies about pain relief in labor in order to shame her into having a med-free birth. At the same time, because of her own personal circumstances, she chose to give birth without pain meds twice.
2. She speaks out against those who make breastmilk out to be some sort of liquid gold and who manipulate women by calling formula “poison”. And at the same time, she is supportive of breastfeeding moms. She breastfed all 4 of her kids.
3. She speaks out against CPMs (lay midwives who are tricking the public into believing they have medical training), but is supportive of midwifery programs that do produce real midwives (CNMs) who can provide safe care.
4. She wants women to know the real risks of homebirth. But even though it has been shown again and again to be more risky, Dr. Amy has repeatedly stated that it should be legal to birth at home and that she would never outlaw it even if she could.
All this sounds very flexible to me.
The Plastic and Reconstructive Surgery journal recently published a trash article speculating on why fewer plastic surgeons were doing hand surgery. The section editor wrote a 10 page “discussion” pointing out all the ways in which the article wasn’t worth the paper it was printed on.
I have no problem with publishing crap, as long as there is some constructive analysis, if nothing else to prove the editors a actually read it.
Why did they print it?
The author kept submitting it over and over again, so it was published against the section editor’s wishes. Which is why he wrote the 10-page response.
You can do that? In my field, once you are rejected, that’s it.
I agree it is troubling to produce a paper with a high point estimate for perinatal death, and claim to have demonstrated safety. However, it is not the case that they had adequate power to detect a “400% increase in the death rate.” As some of the commenters above have suggested, it is difficult to have a precise estimate of a proportion when the event is very rare. The way that proportions are statistically distributed, for a given sample size you have much more power to detect a difference between 20% to 80%, than you have to detect a difference between .2 per 1000 to .8 per 1000.
We can run a confidence interval on a proportion, and that gives you a good idea of the precision of their estimate, and whether it is way far off of a population estimate of 0.4 per 1000. Online confidence interval calculators vary slightly in the calculation method they use, which at these very low rates does make a slight difference, but using the authors’ rate of 3 deaths in 1804 births (excluding 3 anomalies), which is 1.7 per 1000, the confidence interval is approximately 0.3 per 1000 to 5 per 1000.
You can’t exclude the anomalies, because they aren’t excluded from the hospital group. I don’t know the perinatal mortality rate for low risk births in Australia, but I do know that the neonatal death rate is 0.4/1000. The study reported 4 deaths for a death rate of 2.2/1000. What happens when you use those numbers?
I’m curious as to whether the hone birth group included those who were high risk but had a home birth anyway. If they were, wouldn’t comparible risk be difficult to measure? Or were those moms excluded? I’m talking about the breech, vbac, twins, etc who decide to home birth against medical advice. Thoughts?
I doubt it. These programmes are run in conjunction with hospitals and are funded by the government. The midwives are employed within the health system. Twins, Vbac, breech would be excluded for homebirth and be sent to the hospital. It is meant to be as good as homebirth gets.
You can still do a homebirth in Australia by going it alone or by finding a private midwife to take you on. It’s actually what happened with a mum where I live, she was risked out of the local homebirth programme run in conjunction with the maternity hospital because she was due with twins. She paid out of pocket for a private midwife (Lisa Barrett) and delivered her twins at home. The second twin died.
The paper is only looking at the hospital based government programmes, not the private midwife homebirths that are outside of the health system.
According to the actual paper, the HB population did include first-timers, older mothers and breech presentations. There was clearly some risking out, but not as tightly as might be advisable in view of other evidence. One wonders why the authors didn’t comment on this aspect. We know from the UK Birthplace study that primips were a significant risk group.
4 deaths in
1805 is a rate of 2.2 per 1000,
95% Confidence Interval is 0.6 per 1000 to 5.9 per 1000
It’s easy to do this calculation, try http://graphpad.com/quickcalcs/ConfInterval1.cfm
So, AT BEST, with a 95% CI the death rate is 0.6.
Which is a relative risk of 1.5.
And at worst 5.9, which is a relative risk of 14.75. Holy buckets! So basically for best case scenario 1 out of every 3 babies that die at home would have been saved if their mother had chosen to deliver in hospital. And worst case scenario 14 out of 15 babies that die at home would have been saved if their mothers had chosen hospital. Those are BAD numbers no matter how you look at them. And this population was totally low risk. Even best case scenario how could you live with the guilt?
Well, but, but, 2.2 minus 0.4 is only 1.8, and that’s a very very small difference. Less than 2 percent, because 1.8 is less than 2. That’s good right? Death + small number = good outcome!
In the US, the lifetime risk for developing melanoma is 2.03%. That risk doubles if you’ve ever had a sunburn that blisters. They are still both “low” numbers. But when we are talking mortality, especially PREVENTABLE mortality, each percentage is tragic.
Based in the above facts, I would never minimize the risks of sunburn to a patient or the media. It is misleading and irresponsible.
The article in no way shows a “good outcome.”
The two brain-damaged babies are a greater tragedy than the deaths, especially if they require medical interventions for the rest of their lives. We also don’t know how many stillbirths there were.
That’s a bit presumptuous and depends a lot on your value system – a disabled life might still be preferable to non-existence and it might be argued that having a Homebirth increases the risk of death but decreases the likelihood of disability (trades death for morbidity).
I think you may have misread things. The homebirth cohort had both a higher death rate and a higher injury rate (the two HIE babies were homebirth). It’s not a tradeoff- the risk of both death and serious injury are higher at homebirth.
I’m not sure what you’re trying to say, so it’s possible I’ve misread you. If so, feel free to provide correction on what you meant.
What do we know about the criteria for suitability for homebirth? It’s not just enough to say the women “planned to have a homebirth”. So can a cat.
My guess is that it would be fairly strict. These are the hospital provided, government funded (via medicare) homebirth programmes. They don’t take on stuntbirths.
But they did take on primips (first-timers), older mothers and breech presentations.
While I agree with you that they are manipulating the results and that they could actually be worse since the reporting was voluntary, it’s important to note that a mortality rate of 2.2/1000 for a population of 1807 is 4 individuals (a rate of 0.4/1000 for the same population – the rate for hospital births – is 0.7 individuals). So yes, there is a 400% higher rate, but a real difference of 3 deaths.
The difference in the rates is very high, but that does not make it statistically significant when the events themselves are so rare. Yes, they are different, and yes, the mortality rate for homebirths is higher, but the population size is so small that the statistic is highly sensitive to noise in the data. One fewer death in the homebirth group would bring the mortality rate down to 1.6. One more would bring it up to 2.8. Same thing happens if you took a hospital birth population of the same size. One freak death there and one lucky save (or just randomness) in the homebirth group and suddenly the mortality rates are much more similar.
One extra dead baby is one too many and the midwifes are out of way out of line with their boasting of homebirth “safety”, but your response to the statistics in the paper is misleading.
how do you know that the difference is due to noise in the data unless you run a statistical test? That is the function of test statistics, isn’t it? To give you a mathematical idea of how likely it is that the difference is due to chance alone? It would not be so hard to get the rate for comparable low risk women who planned hospital birth and run a chi-square. I think you can even do that in excel…no need for fancy statistical software with an expensive licence.
The authors should have done this. The editors and reviewers should have demanded it.
That doesn’t change the fact that the death rate is extraordinarily high. There really shouldn’t be any deaths at homebirth if candidates are being selected correctly and if transfers are occurring appropriately and in a timely fashion.
If a few babies get killed by a faulty carseat, we don’t blow it off by saying that it was only a few; why should we blow off homebirth deaths because there are only a few?
Did you actually run a power calculation? Binomial data usually get quite big standard errors, which are likely to overshadow something with a mean of 0.002. So I would be very surprised if they have enough power. (It’s the absolute difference in means that matters in a power calculation, not the relative difference)
I figure that the uncertainty in a sample of 4 counts is ±2, which means that the expected value of 1 is well within 2 standard deviations.
I doubt the difference in this study is actually statistically significant, but then again, it is still in better agreement with previous studies that have shown the 3x increase in homebirth death.
But the expected value isn’t one. It’s 0.001.
“There really shouldn’t be any deaths at homebirth if candidates are being selected correctly and if transfers are occurring appropriately and in a timely fashion.”
I’m confused by this. I thought that home birth was inherently risky, even when candidates are properly screened and transported, simply because some complications occur without warning and can’t be treated adequately at home, or are more difficult to detect outside of a hospital. Shouldn’t deaths be expected even in a well screened home birth group? What am I missing?
I live in Australia. As others have already pointed out, the homebirths that were the subject of this study were the homebirths that carried a low enough risk to be accepted into government-funded homebirth programmes. That’s what makes the 400% higer rate, and the real difference of 3 deaths, significant. 4 of these 1807 low risk births resulted in neonatal death. If an Australian hospital had figures like that, the authorities would be swarming like bees.
What this study really shows is that even in optimal circumstances, homebirth carries an unacceptably high risk of neonatal death. The zealots can put whatever ridiculous spin on these figures they want to, but the statistics are there in black and white: 2.2 in 1000, or 1 in 500. I don’t know any rational woman who would go ahead with a homebirth in the knowledge that at least 1 in 500 babies, even in optimal circumstances, die. And I say “at least” because this study didn’t include women who were high risk. Australian government-funded homebirth programmes don’t cater for high risk women. They don’t allow VBACs, they don’t take women with gestational diabetes or pre-existing medical problems, they transfer women with breech babies to hospital care before labour begins and they don’t take women who are carrying more than one baby. In fact, they are so careful that the majority of Australian women can be risked out for one reason or another.
I’m all for women being able to make informed decisions. But for God’s sake, let’s not pretend that 1 in 500 equals ‘safe’. Let’s be honest about the fact that 1 in 500 homebirthing women will spend the rest of their lives having to live with being the “one”. All statistics can be downplayed or glossed over or ignored until you are the “one”. As one homebirth mother, who was a veteran homebirther and prominent in the homebirth community, said at an inquest into the death of her baby last year, “If you are the ‘one’, it’s forever. You can’t go back and do it over”. That woman said that if she could go back and do it over, she’d have gone to hospital for an elective caesarian. Given the alternative, which she was now doomed to live with for the rest of her life, the hospital didn’t seem such a bad place.
FSM that was awful! The paper you reviewed Friday had a major flaw and could be criticized on certain grounds, but it was a decent paper and it doesn’t seem outrageous that it was published. This one…is MJA a real journal or is it one of the pseudo-reviewed journals that keep appearing recently?
FSM == Flying Spaghetti Monster?
Yeah.
Unfortunately it’s a real journal, and generally quite well regarded.
It appears that the standard of evidence form midwives is different from the standard of evidence demanded of other professions. Is statistics not in the midwifery model? Are there ‘other ways of knowing’ in midwifery? Is maths too hard?
The standard of just about everything is different for midwives, because they are special. There is a midwife in British Columbia soliciting money on the Internet to pay for her lawyer because she was denied hospital privileges. I would assume that there are processes in place to deal with privileging and if she was denied, there likely are very good reasons for that. Funny enough, the “college” of midwives of BC doesn’t have an opinion on the ethics of that behaviour. Midwives are not held to the same standards as other regulated health care professions , IMO, because they are not professionals. They offer a service, but they do not seem to get the importance of professionalism. They seem more drawn to the qualities of self-righteousness, entitlement and some falsehood of protecting their clients from the medical system, by pretending to be all about informed choice and autonomy. I call bullshit.
Some of the strongest facilitators and advocates for informed choice and autonomy I have ever encountered have been OBGYNs.
But this is in a medical journal not a midwifery journal. I am very surprised that this got through review!
I drove over 3 1/2 hours to Chicago and back this weekend without an accident. I am very reassured that accidents are quite rare anymore.
Perfect!