In October 2003 the report was published of an inquiry into the death of a psychiatric colleague that raised a number of important questions - not least regarding our management of women with bipolar disorder in pregnancy and the postpartum period (North East London Strategic Health Authority, 2003). Dr Daksha Emson took the life of herself and her 3-month-old daughter, Freya. This tragedy took place during a psychotic episode triggered by childbirth and was a consequence of her history of bipolar affective disorder. Sadly, the case of Dr Emson is far from unique. The Confidential Enquiries into Maternal Deaths (2001) found that suicide had become the leading cause of maternal mortality in the UK, accounting for 28% of maternal deaths (Reference OatesOates, 2003a ). The majority of the women who died by suicide suffered an abrupt onset of a severe psychotic illness within days of childbirth - a ‘puerperal psychosis’, to use the traditional nosological label. For some women an episode of puerperal psychosis will be the first experience of severe mental illness, but a significant proportion of women will have had previous episodes. In the Confidential Enquiries into Maternal Deaths (2001), for example, 46% of the women who killed themselves (the majority by violent means) had previously been in contact with psychiatric services - and half of these had had a previous admission with a severe episode of illness following childbirth (Reference OatesOates, 2003b ). In many maternal suicides, therefore, the occurrence of an episode of severe post-partum psychiatric illness is an eminently predictable event. Although some women in the report were in contact with psychiatric services, none had received a detailed risk assessment, none had a formal management plan, and none was under close surveillance in the puerperium. It is also worth noting that none of the women in the cases of maternal suicide reported in the Confidential Enquiries was under the care of a mother and baby unit or perinatal psychiatric service (Reference OatesOates, 2003a ). The report on maternal deaths in 2000-2002 has now been published (Confidential Enquiries into Maternal Deaths, 2004) and, sadly, the findings are similar.

Compelling evidence points to women with bipolar disorder being at particularly high risk of puerperal psychosis, with episodes following 25-50% of deliveries (Reference BrockingtonBrockington, 1996; Reference Jones and CraddockJones & Craddock, 2001). This high rate of illness represents a many hundred-fold increase from the base rate of approximately 1 in 1000 deliveries. In addition to a history of bipolar disorder, other important risk factors include having experienced a previous episode of puerperal psychosis, having a first-degree relative who has experienced an episode of puerperal psychosis and having a first-degree relative with bipolar disorder (Reference Jones and CraddockJones & Craddock, 2001). Women can be identified, therefore, who are at a vastly increased risk of developing puerperal psychosis - a risk of approximately 60% in women with bipolar disorder and a personal or family history of puerperal psychosis (Reference Jones and CraddockJones & Craddock, 2001; Reference Robertson, Jones and HagueRobertson et al, 2005). However, because of the relapsing and remitting nature of bipolar disorder, these women may be currently well and not in contact with mental health services. Under such circumstances, it is particularly likely that such individuals themselves - and, indeed, others - will fail to recognise the serious risks of the situation. In addition, because of concerns about potential adverse effects on the foetus, medications that may have been successfully used for maintenance treatment are often stopped, by the woman's medical team or by the woman herself, prior to conception or when the pregnancy is discovered.

It is clear, therefore, that women with a history of bipolar disorder need careful management before conception, through pregnancy and during the post-partum period. We suggest a number of key principles on which their care should be based.

STRATEGIES FOR MINIMISING RISK

The high risk of illness in the 2 weeks following delivery in a woman with established bipolar disorder must be recognised both by healthcare professionals and by the woman and her family.

The possibility of future pregnancy should be considered in all women with bipolar disorder who are of childbearing years. The risks of illness following childbirth should be discussed with these women, and the importance of seeking help if contemplating pregnancy (or if unexpectedly becoming pregnant) emphasised.

Given that many women with a history of bipolar disorder will be well and not in current contact with psychiatric services, all women receiving antenatal care should be screened for known important risk factors, and protocols should be in place to ensure that women at potential risk receive a formal risk assessment and management plan. In recent years a number of reports and guidelines recommending antenatal screening have been published by the Confidential Enquiries into Maternal Deaths (2001, 2004), the Scottish Intercollegiate Guidelines Network (2002), the Royal College of Psychiatrists (2000) and the National Institute for Clinical Excellence (2003). How screening and risk management are delivered will vary according to local circumstances, but is arguably best achieved by a specialist obstetric liaison service in the context of a perinatal psychiatry team. The recent announcement of a section of perinatal psychiatry in the College is an important step in this direction.

Decisions about continuing or stopping medications prior to or during pregnancy are difficult and should be the result of a detailed and individualised risk-benefit analysis. No universal recommendation can be made and it must be emphasised that the decision ultimately must rest with the woman and her family. However, stopping medication should always be a carefully considered decision and never a reflex response. Likewise, the decision to start medication for women who develop symptoms during pregnancy or when breast-feeding must be the result of weighing up both the potential risks of taking medication and the risks posed by the illness itself. Given the serious nature of these episodes and the potential for tragic consequences highlighted by the findings of the Confidential Enquiries, the need to treat the mother's illness must weigh heavily in this risk-benefit analysis.

Women with bipolar disorder should have access to specialised, individualised risk assessments - ideally prior to conception but certainly as early in pregnancy as possible. For women at risk, perhaps the most important aspect of management is to maintain close contact and regular review during the period of risk. It may also be important to address other avoidable factors that might increase risk - decreasing general levels of stress, for example, and paying attention to sleep in late pregnancy and the first post-partum weeks.

It is possible that some women with bipolar disorder would benefit from prophylactic medication in the immediate post-partum period. Some evidence exists for the use of lithium in this context (e.g. Reference Stewart, Kompenhouwer and KendellStewart et al, 1991), but the few studies published have been open and retrospective, and there are practical problems with obtaining therapeutic levels quickly to cover the period of risk. These issues have led some perinatal psychiatrists to use typical or atypical neuroleptics as prophylaxis, and despite some anecdotal reports of success with this strategy, no research trial has yet been reported. There is a need for well-conducted studies to assess the use of antipsychotic medication in this context.

Bipolar episodes in the puerperium are often rapid in onset, of a mixed affective type, show an atypical and amorphous psychotic symptomatology with a variability in clinical presentation from hour to hour (‘kaleidoscopic’), and can show a rapid escalation in severity (Reference BrockingtonBrockington, 1996). Services must be able to recognise and respond quickly to women who become ill.

IMPROVING CARE

In summary, women with bipolar disorder are at greatly increased risk of episodes of illness following childbirth, this risk is not currently being consistently recognised and women are dying as a result. Although we have much to learn about how best to manage women with bipolar disorder in the perinatal period, it cannot be doubted that prompt recognition of the onset of illness with the rapid institution of treatment is of vital importance.

The report of the independent inquiry into the death of Dr Daksha Emson and her daughter concluded that the onset and nature of her post-partum illness could have been predicted but that a fully informed risk assessment was not conducted. Perhaps most disturbingly, it went on to comment that if Dr Emson had lived in Hackney rather than Newham she would have had access to a perinatal psychiatry service and her chances of being alive today would have been much greater. This ‘most invidious form of postcode lottery’ (North East London Strategic Health Authority, 2003) is unacceptable, and we must work towards the goal of access to perinatal psychiatric services for all women who need it.

David Emson, in the obituary he wrote for his wife, expressed the hope that the independent inquiry would enable Daksha, even in death, to have a positive impact on the care and treatment of other mothers with bipolar illness (Reference EmsonEmson, 2004). It is our hope that a lasting legacy of the tragic death of Dr Emson and her daughter, Freya, will be major improvements in the management of women with bipolar disorder through childbirth: a wonderful but, for these women, a dangerous time.