We would like to expand on the points raised relating to sampling and assessment tools. We accept the criticism that relaxing the inclusion criteria might have skewed the final results, but have already explained our rationale for this decision. Unfortunately it is not possible to make any inference about the psychological morbidity of non-participants. Non-responding fathers fell into two groups: those who were persistently unavailable and those who declined to take part. Only one father gave a reason for his refusal: rejection of what he perceived as a false assumption that it was possible for a parent to ‘recover’ from a stillbirth. Although we were active in seeking fathers' participation, ethical considerations did not permit us to persist in questioning fathers who declined to take part. Black couples and those from minority ethnic groups were not excluded from the study; rather they were underrepresented underrepresented as a result of higher non-participation non-participation rates.

Two factors contributed to our use of a single dichotomous variable for the presence or absence of appropriate social support. First, social support at the time of loss was not a primary focus of the study and we felt it appropriate to limit the number of questionnaires that participants had to complete. Second, research in this field has relied on a range of assessment tools (e.g. Reference Zeanah, Danis and HirshbergZeanah et al, 1995; Reference Lin and LaskerLin & Lasker, 1996). The use of multiple complex tools limits the comparison of data across studies. However, we accept the view that elaborating on the quality of support would deepen the findings. The Golombok Rust Inventory of Marital State, which was used to assess marital satisfaction, is a short and easy-to-administer assessment which has high face and content validity and good reliability (Reference Rust, Bennun and CroweRust et al, 1988).

We hope that continuing research in this field will lead to greater awareness of the needs of parents experiencing stillbirth.