As authors of the original article,^{Reference Howard, Trevillion, Potts, Heslin, Pickles and Byford1} we support the reflections of the responses from Heron et al and Brockington. In response to Heron et al, we concur with their point that there are challenges in investigating the ‘true’ value of mother and baby units (MBUs) versus other acute psychiatric services. Indeed, the inability to randomise participants meant that we were unable to eliminate all potential biases, despite the sophisticated statistical methods we employed to address these. Uncertainties, therefore, remain in our estimates. We also did not have sufficient sample size to independently examine the effect of all types of acute psychiatric care. As outlined in our paper and final report,^{Reference Howard, Abel, Atmore, Bick, Bye and Byford2} this limitation may have hidden the benefit of one specific setting over another (e.g. MBUs versus generic psychiatric wards or MBUs versus crisis resolution home treatment teams).

In recognising that important outcomes of perinatal psychiatric care extend beyond just clinical recovery, we chose to measure service satisfaction among participants as a key secondary outcome. We found that women reported significantly higher levels of satisfaction when receiving care from MBUs compared with other acute psychiatric services. We have separately conducted additional qualitative work which compares women's experience of care from MBUs, generic psychiatric wards and crisis resolution home treatment teams.^{Reference Griffiths, Lever Taylor, Morant, Bick, Howard and Seneviratne3,Reference Rubio, Lever Taylor, Morant and Johnson4} These analyses clearly demonstrate women's preference to be co-admitted with their baby and highlight that separation from children is not only traumatic but can impede mental health recovery.^{Reference Griffiths, Lever Taylor, Morant, Bick, Howard and Seneviratne3} Women experience MBUs as more perinatally focused and family-centred. By contrast, women report that generic psychiatric wards lack the necessary facilities and expertise to support perinatal women. With respect to crisis resolution home treatment teams, in qualitative interviews many women described these services as being intrusive, overly risk-focused and lacking in care tailored to the perinatal context.^{Reference Rubio, Lever Taylor, Morant and Johnson4}

We acknowledge that we were not able to capture some important additional costs of care in this study, particularly future costs. However, these costs, and any differences between groups, are currently unknown. Our use of quality-adjusted life years (QALYs) – recommended by the National Institute for Health and Care Excellence – as an economic outcome measure did not allow us to take into account any impact on babies; there is currently no agreed method for estimating QALYs in infants. In addition, this particular cohort are dealing with multiple difficulties in the perinatal period, while also coping with raising a new baby, and we might question whether any measure focused only on health-related aspects of quality of life, particularly physical functioning, would capture the full impact of these psychiatric interventions. Other important aspects of quality of life that are not captured by QALY measures include, for example, resilience, feeling safe and supported, reducing mothers’ fears around child removal through parenting guidance and/or support, having suitable accommodation and adequate income to provide for your family, and healthy parent–child and wider family relationships. Our findings show that MBUs are more expensive than other forms of acute psychiatric care, but this is partly inevitable as MBU care includes costs for both mothers and babies. The longer lengths of stay observed in women under the care of MBUs warrants further exploration; we may find that there are factors other than the intervention itself which are unequal between services. For instance, it could be that there are greater incentives in generic psychiatric wards to discharge women earlier so they can return to their baby. It may also be that there is a greater push in generic psychiatric wards to discharge people earlier, to mitigate the potential negative effects of this admission type.

We also thank Professor Brockington for the response. We agree that further research in the area of severe postpartum problems is needed, specifically to examine what works for whom, and in what circumstances, including the important focus of the mother–infant relationship.