More critique of the homebirth study and its reporting by the media

Even if you’ve had only half an ear or eye on the media in recent days, you would have heard of a new study, published in the current edition of the Medical Journal of Australia showing that “babies are seven times more likely to die during home births”.

The most horrifying aspect of the study is the way it’s been reported by the media and promoted by sections of the medical profession, and I’ve written more about this in the Crikey bulletin today.

Associate Professor Hannah Dahlen, Vice President of the Australian College of Midwives, and an academic at the University of Western Sydney, and Professor Caroline Homer, Professor of Midwifery at the University of Technology Sydney, also had a critical look at the study and the way its findings are being portrayed.

They write:

• This is a retrospective population based study (low level evidence).

• One of the problems is that the planned home birth group includes women who planned homebirth when booking in for care but then developed risk factors and had their babies in hospital. There are probably only two women whose babies died; who started labour at home planning a homebirth and one of these was a twin pregnancy (high risk). This latter woman persisted in having a homebirth due to ‘unsatisfactory hospital experiences.’ The others had all transferred before the onset of labour. The authors admit they ‘could not differentiate all planned homebirths according to whether transfer to hospital had occurred before or during labour.’ So for low risk women who started labour at home the risk was very low – 1 death in 16 years

• There is no way to tell if these planned homebirths were under the care of a registered midwife.

• This was not a low risk population of women – there was a high rate of post-term pregnancy (3.8% in the planned homebirth group vs. 1.2% in the planned hospital birth group); twins (five sets of twins); and, 8.8% of women had a previous caesarean section.

• There were a number of positive benefits. There was significantly less intervention in homebirth group: caesarean section rate of 9.2% (home) vs. 27.1% (hospital) (one third less); instrumental birth rate of 4.4% (home) vs. 12.8% (hospital) (one third less); episiotomy rates of 3.6% (home) vs. 21.7% (hospital) (one sixth less). Planned homebirth women were three times more likely not to have perinea trauma than the planned hospital group (there was seven years missing data on episiotomy and perineal injury). None of this appeared in the media.

• There was no difference in major maternal complications such as severe perineal trauma (1% in the planned homebirth group vs. 1.8% in the planned hospital group) and postpartum haemorrhage (4.4% in the home group vs. 5.5% in the hospital group). While Dr Pesce’s comments in the Editorial, that the lack of difference in the rate of haemorrhage appears to be because of ‘the adoption of oxyocin into home birth’, the authors of the paper state ‘it is tempting to attribute this to a wider adoption of oxytocic prophylaxis in homebirths but we have no data to confirm or refute this hypothesis.’

• The numbers of perinatal deaths in the homebirth group over 16 years were small (9 deaths).

• There were two perinatal deaths that actually occurred at home. One baby had lethal congenital abnormalities (this was known before labour and a decision made for the baby to be born at home). The second death at home was after a waterbirth which was not found to be the cause of death but a review identified that increased monitoring may have identified the baby was in distress.

• One perinatal death occurred in hospital after a transfer after the birth of the first twin.  The first twin was born at home and second twin was born in hospital after a delay in transfer and subsequently died.

• There were 6 perinatal deaths in the planned homebirth group where the baby was born in hospital. Presumably these women were transferred to hospital during the antenatal period as antenatal risk factors developed. Transferring to hospital if or when risk factors develop during pregnancy is appropriate practice.

• Of the six deaths in hospital: one had hydrops fetalis (a condition diagnosed antenatally often with poor outcomes regardless of the model of care); one death was unexplained with a cord entanglement seen after birth; one had pulmonary hypoplasia (a lung condition due to a lack of amniotic fluid around the baby) after a early rupture of membranes; one was a growth restricted baby with an abnormal karotype (abnormal chromosomes); one was born to a woman who was very overdue (‘seriously’ postdates) and underwent induction in hospital without fetal monitoring (the woman refused) and her labour eventuated in a stillbirth; and,  one was a woman with known haematological (blood related) risk factors whose baby had a lethal abnormality. To reiterate, all these were born in hospital.

• Only three of the deaths are thought to be related to perinatal asphyxia.

• Three of the deaths were thought to be potentially preventable and related to the model of care. These were the baby born after the waterbirth at home; the second twin who was born after an intrapartum transfer– and the baby born after being very postdates. Therefore, there were 3 deaths in 16 years – two of which had risk factors present. That means that there was only one death where there were no risk factors in the 16 year period.

• The numbers of planned homebirths are small (n=1141) (gave birth at home n=792; in hospital n=349). You cannot look at the rare outcome of intrapartum death in such a small sample as the wide confidence intervals demonstrate (95% CI 1.53-35.87) (there is 1 intrapartum death at home and 1 in hospital). It is difficult to examine intrapartum asphyxia with any certainty due to the very small numbers and again the very wide confidence intervals show this (95% CI 8.02-88.83) (1 at home and 2 in hospital). You would need more than 10,000 births at home to show clinical relevance and have some confidence in the statistical significance in relation to perinatal mortality rates. The authors acknowledge this in the paper and present their data with caution in the paper stating that the ‘small numbers with large confidence intervals limit the interpretation of these data.’

• The facts are there was no difference in perinatal mortality (stillbirths, and neonatal deaths within 28 days of birth) between home and hospital (7.9 vs. 8.2 per 1000 births). For those actually born at home the perinatal mortality rate is 2.5 per 1000 births, which is comparatively low.

• There were no differences in Apgar scores (how well babies are at birth) or admission to Neonatal Intensive Care Units between the two groups. Infants born at home were half as likely to receive specialised neonatal care compared to planned hospital birth.

• Looking at rare outcomes with small numbers often shows statistically significant results but they cannot have confidence around them due to the size of the sample and the event. Examining rare events is always difficult in small retrospective studies like this. If, for example the study had measured maternal mortality, they may have found one or two deaths in the hospital group because of the large numbers in the population but it would have been erroneous to say less women die at home than in hospital because of the small numbers of deaths.

• The paper highlights that the system must be so terrible for some women that they would choose to give birth outside of it than in it, even with risk factors. This is an indictment on the current maternity system in Australia – that needs fixing – removing homebirth won’t do this.

What was missed?

The conclusion of the paper is very sensible recommending risk assessment, transfer and fetal monitoring.

So then why did the data get so grossly misinterpreted?

• The reality is despite a malfunctioning system in this country where midwives are uninsured and have no visiting rights, and homebirth is unfunded and often hard to access, the perinatal mortality rate was no different.
• Risk assessment, transfer and fetal monitoring will be improved when private midwives are no longer excluded from mainstream services so we should be aiming for this not continuing the ‘witch hunt’ against private midwives.
• The intervention rates are to be commended and no one in the reporting of this paper noted this.
• Some women will always choose homebirth so we should support this choice with safe responsive systems of care. The authors state that ‘women’s autonomy in choosing reproductive behaviour is a fundamental human right enshrined in Australian law’.
• The excess mortality continues to be found in high-risk women and women need to be informed of this risk.
• Freebirth (giving at home birth without a skilled and registered birth attendant) is rising in this country and this is a concerning outcome of restrictions on options like homebirth and trauma from hospital births

This is a classic example of science being usurped by politics. It is dangerous, misleading and sadly the media seem to prefer sensation to reality. The politicisation of this study has been unfortunate as important and useful lessons were lost in what followed.

At the end of the day homebirth will not go away. It has, does and will exist in every country on earth.

So we have two options – burry our heads in the sand and hope it goes away (it won’t) or put in place responsive, evidence based systems of care (we haven’t). When the dominant politics is determined for homebirth to be eradicated there is little chance for science to project an informed and balanced voice into the debate.