In our Cochrane review of home like places for birth,1 we contentiously noted a trend towards increased perinatal mortality rates for alongside birth centres. We were therefore very interested in the recent paper by Mori et al.2 We accept that this is a difficult area to address. However, given the clearly identified limitations of the data in this study, we are rather puzzled by the weight that has been given to the findings, both in the editorial3 that accompanied the paper and in subsequent media coverage. Two crucial aspects of the study prevent the data from being generalisible: the inadequate data sets available and the consequent lack of like-for-like comparison groups.
The requirement of reliable data sets is fundamental to epidemiological studies.4 If this cannot be established, error is built into the study from the beginning. The authors outline the problematic nature of their data under headings ‘data collection’, ‘measurement errors’, ‘bias’ and ‘confounding’. Each of these can render epidemiological analysis tentative and equivocal. It is therefore misleading to draw conclusions and proffer explanations as if the results have some degree of robustness.
In terms of comparison groups, the clinically significant comparator for low-risk home birth booked women who become higher risk is low-risk hospital booked women who become higher risk. For the same reason, the very low perinatal mortality rate in the home birth booked women who gave birth at home needs to be compared only with women who start and finish labour as low risk in hospital.
Two additional aspects confound attempts to understand these data. First, transfers during labour are not distinguished from those occurring antenatally. If the risk is real, we do not know whether it lies in complications of pregnancy or in the way the intrapartum episode was managed. Second, even if problems could be identified as occurring in the intrapartum period, we do not know if this is an issue of midwifery care at home, of geographical distance, of the way transferred women are managed en route or after they enter hospital, or, indeed, if there are a whole range of other reasons.
Given all these caveats, we suggest that the only conclusion that can be drawn from the study is that we simply do not know about the safety of labour-related home birth transfers and that future studies (preferably randomised controlled trials) are required. If we are indeed convinced of the value of evidence-based service provision, and of the provision of unbiased information to service users, this should surely have been the message transmitted strongly to the media.
In the absence of trials, the current BirthPlace study5 will go some way to providing some of these answers.