A tale of two homebirth studies

To much fanfare, the Midwives Alliance of North America recently released the results of its five-year MANA Stats project, at last publishing the data collected from almost 17,000 midwife-attended home births in the US during the period 2004-2009. In the UK at the same time, a similar study was being conducted: the Birthplace study, which compared homebirths with obstetric unit and birth centre births in the period from April 2008 – April 2010. These results were published in 2011.

Both studies aimed to provide data on the risk of planned home birth. Only one study – the British study – actually gives any information that’s useful in decision making. The reasons why are mostly due to differences in the healthcare systems in the two countries. The reason you should care is that medical providers have a responsibility to deliver evidence-based care; if you can’t monitor and test the safety of the care you’re providing, you’re failing in your duty to your patients.

Firstly, who performed these studies? The organisation collecting the data in the American study is a professional midwifery association and their stated goals include “to increase access to midwives in all settings” and “unifying and strengthening the profession” – so they are hardly impartial. Conversely, the research group in the British study was from the National Perinatal Epidemiology Unit, an academic unit independent of the NHS with a wide range of research interests.

How well do the two studies identify the population and sample they’re studying? Well, in the MANA study, participation in the study was entirely voluntary with only 20-30% of the (estimated number of) active Certified Professional Midwives participating, and even fewer of the Certified Nurse Midwives (CNMs). There is no information given about how the midwives who did not participate were followed up and encouraged to submit stats, or the characteristics of the participating midwives beyond the certification they held. It is entirely possible that only those midwives who felt their statistics would portray them favourably, or who were not breaking the law with their practice would decide to take part.

In another paper the validation of the MANA Stats database is further described. As this database was electronic, one simple safeguard is to not allow logging of patient data too long after it has happened – to limit recall problems and prevent selective reporting of outcomes. The MANA Stats database did not allow a record to be created after the actual date of delivery, but contained no further safeguards (and even this safeguard was breached in about 8% of cases). Additionally, no data about the number of incomplete records is given (apart from the assurance that it is ‘a small number’) – leading to the possibility that if things had gone wrong a record was simply abandoned. There’s plenty of anecdotal evidence that the records were not filled in in the way MANA describes. Similarly, the ‘data review’ was entirely done by people employed by MANA, who phoned up study midwives if they thought they detected an inconsistency or ‘adverse outcome’ in their data – hardly the most objective method.

How did the Birthplace study recruit its participants? They approached all the midwife-led units in the UK, all the NHS trusts providing homebirth services, and a random sample of 36 hospital obstetric units. Of these units, three quarters achieved a response rate of 85% or more. Forms were distributed by a coordinating midwife at each organisation and were designed to be started during labour and completed by 5 days after birth. Data was collected every month, giving little opportunity to forget, fill them out too late or remove less than favourable outcomes. If an adverse event was recorded, a further (standardised) form was sent to the hospital to collect more data.

What about the quality of the data? In the UK, care of pregnant women is by a team of midwives and doctors, and as the NHS is a national organisation the sort of tests that are done and information that is recorded is standardised, so if a woman ends up being cared for by someone who doesn’t know her they have all the information they need. In America, CPMs are not trained to national standards and generally work alone or in small groups, so they will not have consistent record-keeping procedures or standards of care. In addition, in certain states there are laws which govern a midwife’s scope of care and she may be breaking the law if, for instance, she fails to transfer a woman to hospital after a high blood pressure reading. As such, conflicts of interest might lead to records being incomplete or falsified. One example is that we have no idea how each midwife in the study defined a ‘lethal congenital anomaly’ – some of these babies might conceivably have had a severe but survivable condition. It is also implausible from data in other places that 50% of women in this dataset did not sustain any perineal damage at all, and the British study noted the duration of pregnancy in weeks, compared with ‘clinical gestational age assessments following birth’ in the MANA Stats, which might be all you have to work on if the women hasn’t had an early dating scan as standard, or you perhaps don’t want to admit to attending a 35-week or 43-week labour.

I’m not particularly going to comment on the outcomes – this has been extensively discussed elsewhere. The Birthplace study provides good evidence that if you’re living in the UK and are low risk and having a second to fourth child, there’s no reason to think you’re at greater risk outside of the hospital than in it, whereas for a first child your likelihood of poor outcomes is increased at home. And because this study was based on women who could equally have chosen a home or hospital birth, we can use the data to answer the question we really want to ask – what can we tell low-risk pregnant women considering their place of birth about the risks and benefits of each option?

The MANA Stats study doesn’t allow us to do that. Although the data did show that poor outcomes are rare, because of the lack of control group it’s impossible to say if they are less rare than in hospital, though the suggestion is that they are. So it doesn’t really help us at all with what to recommend to low-risk women considering their place of birth.

There’s plenty of reasons this is the case. CPMs aren’t at all integrated into the hospital system in America, meaning recruiting a group of women giving birth at the hospital would be challenging. The decision to pursue a home birth requires much more proactive planning in America than in the UK, so women who don’t are much more difficult for MANA to identify and contact and the ‘control group’ to compare to the women who do is much harder to define. Lack of universal standards of data collection means the data is necessarily going to be less precise, sloppier and more unreliable, and the shady legality of midwifery in some places means there is plenty of incentive to be dishonest about data, while working alone or in with a small tight-knit group of colleagues means there is more opportunity.

But all of this, rather than being a reason to reserve judgement on homebirth in America until there is better data, is a reason to strongly recommend against it and attempt to dissuade any American woman from attempting it. ‘Midwifery’ might be knowing what to do in the case of a postpartum haemorrhage, but ‘healthcare’ is a much broader organisational structure – it means providing measures to prevent the haemorrhage in the first place, making sure the attendant has the right training and equipment to handle it, being able to monitor your attendants to see that they are handling things correctly, making sure your patient can access other care providers if needed, and continuously ensuring that your organisational structure isn’t allowing dangerous things to happen due to incompetence, ignorance or managerial oversight. This is exactly what the Birthplace study has examined – and in England, it shows that the ‘healthcare’ of homebirth functions well, making it safe, and this study is only one part of an ongoing monitoring of safety throughout the healthcare system that can continuously respond to data to ensure that safety of patients. The fact that the MANA midwives can’t, after 10 years of trying, provide any good data about the healthcare provided by their members suggests that although they might be delivering midwifery, their healthcare is falling sadly short. It’s not just that a failure to properly collect and monitor outcomes suggests a deeper organisational ineptness and lackadaisical approach to evidence-based medicine; being so alienated from the rest of the healthcare system in the US means that the safety of homebirth midwifery in the US is essentially going unexamined, and it’s unethical not to attempt to correct that. Modern medical care is based on the principle of using evidence-based medicine practices to continually update and improve your healthcare interventions and systems. Refusing to do that means you’re not practicing medicine, you’re practicing quackery.