An estimation of intrapartum-related perinatal mortality rates for booked home births in England and Wales between 1994 and 2003. BJOG: An International Journal of Obstetrics & Gynaecology Volume 115 Issue 5, Pages 554 - 559 (2008)<br>See our record:

Commentaires parus dans BJOG: An International Journal of Obstetrics & Gynaecology

Editorial (editor's choice) de BJOG, avril 2008

How safe is home birth?

Having been born in my grandparents' home (delivered by a community midwife who also delivered my aunt, uncle and cousin), I have always been interested in the home versus hospital delivery debate. Even as late as 1960, one in three births in the UK were at home, but by the early 1990s, following campaigns by women who saw birth in hospital as safer, and encouraged by the first House of Commons Health Committee Report on the Maternity Services (the 'Short Report'), this had dropped to below 2%. However, the second House of Commons Health Committee Report on the Maternity Services published in 1991 (the 'Winterton Report'), on which I had been an obstetric advisor, emphasised the importance of choice. Stimulated by a follow-up report by Lady Julia Cumberlege called 'Changing Childbirth', the rate of home births started rising again, albeit very slowly. In 2005, 17 279 of 645 835 total births in the UK were at home (2.7%). On 1 April 2007 in the (UK) Sunday Times, Ivan Lewis (Health Minister) was reported as planning that within the next 3 years, all women in the UK would be guaranteed the opportunity of giving birth at home. Given a rising birth rate and a falling number of midwives, some suspected that the date of the announcement was significant (Editor's note: in many countries, there is a tradition of playing practical jokes on the first day of April), but subsequent announcements confirmed that the Minister was not joking (at least, intentionally). Even Sheila Kitzinger, a social anthropologist and powerful advocate of women's choice commented 'This plan is just spin. It cannot be done' (UK Daily Telegraph 3 April 2007). As I mentioned in last month's editor's choice, arguments about the safety of home birth have been bedevilled by the lack of robust clinical trial evidence, and observational data can be misleading. The paper by Mori, Dougherty and Whittle on page 554 is a brave attempt to address this issue. They report perinatal mortality in England and Wales between 1994 and 2003. With the exception of 1 year, the perinatal mortality in home births was always higher (by 50 to 120%) than that in hospital. Unfortunately, women delivering at home unintentionally (due for example to an unanticipatedly rapid birth) are a high-risk group, and their inclusion in the summary figures makes it impossible to assess directly the mortality associated with intentional home birth. Mori et al. attempt to compensate for this by using figures for the proportion of deaths associated with planned and unplanned home births collected since 1994 by the Confidential Enquiry into Maternal and Child Health. Furthermore, they impute from previous smaller studies in various parts of England and Wales the proportion of home births planned to be at home but transferred to hospital because problems had developed in labour (this group also has a poor outcome). This enabled the authors to make an estimate, with confidence intervals, of the likely perinatal mortality associated with an intention to give birth at home. Although this should be a selected low-risk group, their perinatal mortality between 1994 and 1997 was no lower than that of hospital births and subsequently was consistently about 80–250% higher (significantly so in 1998–99 and 2002–03). It seems strange that the government of a major developed country should be encouraging a substantial return to home births without commissioning a major prospective study of its safety, and the practicality of providing the increased numbers of midwives that would be needed to implement it.

In this context, it is obviously important to know the views of women about their preferred place of birth. On page 560, Pitchforth et al. analysed in detail the views of 877 women who had recently given birth in Northern Scotland, where access to hospitals can be difficult, especially in winter. The main themes of their responses were extracted. These emphasised the dependence of the choices the women made (or would like to have made) on the local services available, including the possibility of rapid transfer to hospital in the case of an emergency. Family circumstances were also important; for example, women who already have small children dislike being separated from them. However, overall, the women sampled had 'an overwhelming preference for (maternity) unit-based care as opposed to home birth', and the majority also preferred physician-based care to midwifery-managed care. A cynic might suggest that the UK government has promised 'home birth for all who want it' with the expectation that only a few actually will.

The Dutch practice of continuing high rates of home birth (around 30%) with a relatively low perinatal mortality is often quoted as evidence for the safety of the practice. Unfortunately, the perinatal mortality rate in the Netherlands has been declining more slowly than in most European countries (Mackenbach, Ned Tijdschr Geneeskd 2006;150:409–12). The rate of perinatal deaths in that country is now the second highest in Europe (Sheldon, BMJ 2008;336:239), and Sheldon quotes two senior Dutch obstetricians as calling for the tradition of home births to be scrutinised (Visser and Steegers, Medisch Contact 2008;63:96–9). On page 570, Amelink-Verburg (a midwife working at the Netherlands Organisation for Applied Scientific Research) and her midwifery, obstetric and paediatric colleagues report a study of 280 097 women under the care of a midwife at the start of labour, of whom 79 270 were referred to an obstetrician non-urgently and 9985 were referred as an emergency. The short distances for travel and the good transport links within the Netherlands make such transfers less difficult than in many other countries. Half the referrals were due to perceived 'fetal distress' and another third to postpartum haemorrhage. Although intervention rates were substantially lower in home births, 1.7% had a blood loss exceeding 1 litre. In cases of urgent referral, 5.3% of babies had a 5-minute Apgar score of less than 7 and half of all neonatal deaths in the first 24 hours of life occurred in this group. More than 1 in 20 (5.6%) women having a home birth were transferred to hospital in the second stage of labour, which as the authors say, 'may be more stressful for the mother'. Of course, the problem is that we do not know if the outcome would have been different if the women had had earlier access to specialist obstetric and neonatal care. Once again, only randomised trials will give us an answer to such questions.

Réponse du rédacteur en chef (chief editor)

(on n'a pas la question)

Dr Walsh and Professor Downe question the weight given in my 'editor's choice' to the findings in the paper by Mori et al. in the April issue of BJOG. The national media certainly thought their findings worthy of comment, and so did I. It is important that we give women the best available information about the risks of home birth when they are making choices about where to have their baby, and the absence of definitive data should not mean that we deny them access to such data as do exist. Surely, making Mori's data public is preferable to making national policy decisions without considering their evidence at all. In relation to Walsh and Downe's concern about the effect of the timing of transfer, we published considerable detail on this in the accompanying paper by Amelink-Verburg et al. from the Netherlands, where they manage to record these things better than in the UK. Walsh and Downe say that "we simply don't know about the safety of labour-related home birth transfers", but in the absence of precise data relating to the UK, perhaps we are allowed to use a thought experiment. If there is a sudden abruption, or a deep transverse arrest with thick meconium, when an immediate caesarean is indicated, is it not a reasonable hypothesis that being transferred across a corridor is safer and quicker than negotiating the traffic on Britain's congested roads?

P Steer a


Mori et al.1 in their study of intrapartum-related perinatal mortality (IPPM) attempted to use the 'best available data' to ascertain the safety of planned home birth. The validity of this study hinges on being able to determine accurately the numbers of planned home births. The authors took the number of actual home births and adjusted these using estimates for the numbers of both unintended home births and transfers. However, they have used inappropriate assumptions and have compounded these mistakes by making errors in their calculations.

The authors use two ways to determine the numbers of unintended/unplanned home births, Calculation A and Calculation B, producing widely differing answers of 66 265 and 20 206. 'Calculation A' estimates unintended home births as a percentage of all home births (50.7%) and 'Calculation B' as a percentage of overall births (0.32%). There is no reason to suppose that the number of unplanned home births are affected by a rise or fall in planned home births. However, it is likely that a small consistent proportion of pregnant women concealed their pregnancy or had a precipitate birth at home. Indeed, Murphy et al.2 reported from 1970 to 1979 that unintended home births formed a relatively constant percentage of all births, around 0.35% (range 0.27–0.46%). In contrast, unintended home births increased from 17 to 57% when expressed as a percentage of all home births. This demonstrates that Calculation B is more reliable, yet Mori's conclusions are based on Calculation A.

In addition, calculations of the numbers of births and IPPM rates using Calculation A are subjected to a number of errors and are therefore invalid. The Murphy study2 data applicable to Calculation A are included in table 1 (34.1%) but omitted from the calculation of both the weighted mean and the sensitivity ranges used to create table 2. Using a revised weighted mean and lower range reduces the IPPM in table 2 for booked home birth, whether completed or not, and increases the range in which the true rates could lie. Furthermore, the study by Redshaw et al.3 is included in Calculation B, but not Calculation A, adding to the inaccuracy of table 2. There is also an error in table 2 in the completed home birth group, where 31 intrapartum-related deaths for 83 343–111 126 gives a range of 0.28–0.37, not 0.28–1.15 as reported.

The authors inappropriately make a direct comparison between women who 'planned home birth but transferred to hospital', with 'all women giving birth'. However, if such a comparison is made, it should be with a matched group of women who booked hospital birth and developed complications. The key finding that 'there was no evidence of difference in the IPPM rate for the booked home birth group compared to the overall rate' was not reported in the abstract as it should have been.

This study1 illustrates that the 'best available data' are, in this case, not 'good enough' and are unacceptable for providing evidence-based information for professionals, service users, the press, or national guidance.4 We shall be submitting a detailed critique for publication after peer review.

G Gyte, a M Dodwell, b M Newburn, c J Sandall, d A Macfarlane, e & S Bewley f

We welcome the comments from Gyte et al.1

They criticise our use of two methods to obtain transfer rates. Although we noted the study by Murphy et al. which was published in 1984,2 the data used were in fact from 1970s and we regarded them as being too old to relate to the more recent practice of maternity care. Indeed, this was the reason the Murphy data were omitted from the summary of the calculation because our preset criteria used rates derived after 1980.

We have checked the calculations and accepted the error on the intrapartum-related perinatal mortality (IPPM) rate for the completed home birth group in the table 2 pointed out by them. The IPPM rates for the completed home births group should be a range of 0.28–0.37, not 0.28–1.15. However, this does not change our conclusion in any way.

The problem is that there is a lack of an evidence-based concerning the proportion of unplanned home birth among all births. Most of the available studies report ratios of unplanned home births to all home births, and this has become the generally accepted method. In fact, neither this method nor the one described by Murphy is particularly superior, and therefore, we believed that it was best to present both methods.

Another criticism concerns the comparative group. We tried to get the best available estimate of IPPM of all booked home birth and to see how it related to the national average for all births. We believe that it is reasonable when the population concerned is relatively small to use the whole population as a reference.

R Mori, a M Dougherty a & M Whittle a


We thank Drs Walsh and Downe for their interest in our paper. All data have their limitations. This means that we always deal with a level of uncertainty and unreliability. The two data sets we used, 'National Statistics' for the denominators and the 'Confidential Enquiry into Maternal and Child Health' for the numerators are well-established data sets of good quality, although there are still limitations. However, the data set of main concern relates to the transfer and unplanned home birth rates derived from previous studies. We minimised uncertainty by undertaking a systematic analysis. The sensitivity analysis considered all the possibilities we could think of for error within previous studies, and we accept that there remains some uncertainty about our conclusions. This was acknowledged in our paper.

The best choice of comparator depends upon the research question. Our question was whether women who booked a home birth and had their babies at home, those who booked home birth but whose care was transferred to hospital, and those who did not intend to have home birth but had their babies at home, had a higher or lower risk of their babies dying, compared with the national average. Therefore, the comparator was the national average, which represents the 'typical' experience of a woman who gave birth in England and Wales during the period of our study.

Another question was whether women whose care was transferred during pregnancy and labour had a higher risk of their babies dying related to intrapartum events compared with those who booked and completed home birth. Unfortunately, there are only a few data sets available, which show antenatal and intrapartum transfer rates separately. We acknowledged this limitation, and we emphasised that these data should be available separately in any data collecting systems dealing with place of birth that are put in place in the future.

Although we explored the background causes, which may explain our results (including the limitations of the data, variations in clinical practice, geographical distances etc.), we were unable to find any that explained fully the differences that we observed, and we were therefore left with the conclusion that the differences we observed were due to the actual place of birth.

R Mori, a M Dougherty a & M Whittle a