We agree with Dr O'Keane regarding the severity and potentially devastating consequences of post-partum psychosis in women with a history of bipolar disorder and assure her that any negative emphasis she detected in our brief comments regarding prophylactic treatment were indeed unintended. The brief report format did not allow us to discuss this aspect of management at length but we have taken up this issue more fully in our recent editorial (Reference Jones and CraddockJones & Craddock, 2005).

We would, however, defend our contention that the decision to commence mood-stabilising (or indeed any) medication in women of child-bearing years should follow a ‘very careful weighing up of risks and benefits’. Any medication should be started assuming that the women may become pregnant and future pregnancy and contraception should be actively discussed at the earliest possible opportunity.

We would also argue that the evidence base for the use of prophylaxis in women with bipolar illness in the post-partum period is not as robust as would be ideal. As Dr O'Keane has outlined, the literature does support the use of lithium in this context, although the retrospective (and partially overlapping) studies differed in when lithium was commenced – important as there may be practical problems in achieving therapeutic levels quickly following delivery and the onset of puerperal psychosis is typically in the few days following delivery. In our series of 101 women with post-partum psychosis more than half had an onset on days 1–3 with over a fifth on the first post-partum day (further details available from the authors on request). With regard to other mood stabilisers, there are few data in the literature. A recently published study demonstrated no efficacy for sodium valproate (Reference Wisner, Hanusa and PeindlWisner et al, 2004) and, despite anecdotal reports of the benefit of typical or atypical antipsychotic medication as prophylaxis, there are no data regarding their use in this context.

Finally, it is our experience that women have strong views on the acceptability of taking medication during pregnancy and while breast-feeding. This may account for the fact that out of the 54 women in our study who went on to have a further pregnancy, only six took prophylactic medication in the puerperium (lithium or haloperidol). Although only two went on to have a recurrence of puerperal psychosis, the numbers are clearly too small to draw conclusions regarding the efficacy of prophylaxis.

This is an area, therefore, in which management decisions are not straightforward but the frequency and severity of post-partum episodes in women with bipolar disorder must weigh heavily in the risk – benefit analysis. What is needed, we can all agree, is further research to provide empirical data on which clinicians, women, and their families can base these difficult decisions.