Working out what works: The case of midwife led care – Commentary on: Is model of care associated with infant birth outcomes among vulnerable women? A scoping review of midwifery-led versus physician-led care

The earliest published randomized trial of midwife led antenatal continuity of care appeared more than 25 years ago (Flint, Poulengeris, & Grant, 1989). The recent Cochrane review of all eligible midwife led continuity of care trials since then shows impressive clinical and psychosocial benefits for women randomized to this approach to maternity service delivery, when compared to other models of maternity care (Sandall, Soltani, Gates, Shennan, & Devane, 2015). The outcomes include reduced rates of prematurity, and of foetal/neonatal death. However, while the Cochrane review authors note that the benefits held true for both healthy women and babies and those with clinical complications, the review did not undertake a sub-analysis for womenwho were socio-demographically marginalised. These women and babies are particularly at risk of adverse clinical outcomes in pregnancy and labour, and so an assessment of the potential benefits of different models of maternity care for them could have significant implications for the future. The authors of the scoping review in this edition of the journal have undertaken this important task. They have also provided a careful analysis of what the potential mechanisms of effect might be for the results emerging from their review. This analysis provides a valuable basis for further commentary on what might actually be working in the context of midwife led care. All but one of the studies in the Cochrane review took place in high income countries where midwifery is established and widely available (Australia, UK, Ireland). ‘Usual’ care, the comparator in most of the studies, therefore generally included some degree of midwife input for the women randomized to that arm. The remaining study took place in Canada, where the status of midwives at the time the study reported (1996) was much less certain, and the comparator was