Individualising the risk of recurrence for women with bipolar disorder in the perinatal period

Background
Bipolar disorder is a severe and common mental illness, affecting between 3 and 5 in every 100 people. It is characterised by episodes of high mood (mania) and low mood (depression).
Of women with bipolar disorder, an average of 40 to 50 in every 100 are at risk of having an episode of the illness in pregnancy or following childbirth (called the perinatal period). Around 20 in every 100 develop postpartum psychosis, a severe illness episode that begins suddenly after having a baby.
Symptoms during an episode are severe and can change rapidly, within hours or days. They may include high or low mood, confusion, abnormal beliefs, and hearing or seeing things that are not there. Women are 23 times more likely to be admitted to psychiatric hospital with an episode of bipolar disorder in the month following childbirth than at any other time in their life.
Bipolar illness episodes can impact on the lives of women and their families, disrupting the relationship with the new baby. Tragically, but rarely, illness episodes may lead to suicide or harming the baby.

Problem
Women with bipolar disorder have very difficult decisions to make about pregnancy.
These include stopping, switching or continuing medications, in pregnancy or whilst breastfeeding, with all of these options having potential risks and benefits. Previous studies found factors that contribute to the risk of having an episode, which include first pregnancy and having a family or personal history of postpartum psychosis. However, that is not enough information to make good decisions.
A recent study showed that women want more information about their own individual risk. Studies that follow women during pregnancy are the gold standard, but are rare.
Due to the lack of information, clinicians can currently only give general advice and estimates of risk, limiting the ability of women to make informed decisions.

Aims
The aim of this project is to understand and predict the risks of women with bipolar disorder having an episode around childbirth.
This will help us tailor the information that can be given to a particular woman, at a specific time, with her individual illness and treatment history. This project will help women to decide about becoming pregnant and how they may best reduce their risk of becoming ill, including about continuing or stopping medication.
I will do this by:
- interviewing a large group of pregnant women with bipolar disorder and following them up through to the postpartum (300 participants)
looking at which factors predict an episode in pregnancy or following childbirth in over 2,000 women with bipolar disorder who have already taken part in the Bipolar Disorder Research Network (BDRN), the biggest research study of individuals with bipolar disorder in the world (>6,000 participants)
- creating mathematical models with these risk factors and looking at how well they predict episodes of illness in pregnant women I will recruit and in an external electronic records database.

Clinical usefulness and outputs
The results of this study will likely have a direct impact in the clinic. The mathematical models to predict risk will be used to individualise risk assessment and provide support to women and healthcare professionals in making informed decisions about care in the perinatal period.
Knowing if a particular woman is at higher or lower risk of having an episode and if there are factors that we can change to decrease that risk, will be of enormous benefit to that woman, her family and the NHS.
Finally, the results of this study can be used to develop more formal decision-aids and to inform healthcare policies.

Aims: The primary aim of this study is to understand better the factors that influence an individual woman's perinatal risk of a bipolar recurrence, which will enable women and their clinicians to make better informed decisions about their care.
I will do this by:
1) recruiting a large, well characterised, cohort of women with bipolar disorder in pregnancy and following them up through to the postpartum;
2) developing prediction models of the risk of recurrence of bipolar disorder for women in the perinatal period (pregnancy and up to 6 weeks postpartum) in an existing large cohort of parous women with bipolar disorder;
3) validating the models and exploring additional risk factors in the prospective sample of pregnant women with bipolar disorder and an electronic records database.

Plan of investigation: My project will involve recruiting and conducting data analysis of a prospective sample of pregnant women (n=300) to build on an existing pilot of 120, and data analysis of a retrospective sample of parous women from the Bipolar Disorder Research Network, BDRN (n=2,181).
Participants will be assessed with a semi-structured interview and complete questionnaires on mood and personality. Medical and psychiatric case notes will be collected. Women who are pregnant will be assessed at two time points (third pregnancy trimester; 3 months postpartum).
I will use multivariate statistical methods (including regression models) applied to risk prediction modelling. I will choose variables based on the literature, and clinical and patient experience that are known or may plausibly be associated with vulnerability to the puerperal trigger. I will test model performance, namely discrimination (e.g. c-index) and calibration (e.g. plots). I will test internal validity using bootstrapping techniques. The final model will be tested in external samples.

The findings can then be used to develop more formal decision-aids and to inform perinatal healthcare policies.

1. Patients and their families and the general public
My research will help understand risk and resilience factors for bipolar perinatal recurrence. This will help individualise the risk and make tailored decisions about care. It will also identify potentially modifiable factors that can be targeted to reduce the risk.
With the above information and an estimation of a woman's individual risk of recurrence in the perinatal period, women and their partners will be better equipped to make decisions relating to pregnancy.
This will lead to better outcomes for the patients and reduce deleterious impacts on child development.
With my project, I will increase patient and public engagement with research, through the project's advisory panel, and by co-presenting at a scientific meeting.

2. Academics
This project addresses an under-researched area, a NICE research recommendation and MRC strategic priorities. I will build up a bipolar prospective pregnancy cohort, which will carry on beyond this fellowship. It will allow studying factors that have not previously been identified, including the interplay between physical and mental health.
My study can provide clues as to why bipolar episodes happen, and therefore help other researchers in similar fields.
The risk prediction model can be used to stratify participants for future clinical trials.
Using risk prediction in clinical psychiatry, is an innovative technique. It will contribute to bring parity of esteem to mental health.
This project also makes use of three different independent cohorts (retrospective, prospective, health records), and can therefore serve as example of the benefits of different types of health data collection.

3. Clinicians (e.g. psychiatrists, GPs, health visitors, obstetricians, midwives)
The results of this study will likely have a direct impact in the clinic.
They will be used to individualise risk assessment and provide support to women and clinicians making informed decisions about care in the perinatal period.
The risk prediction models after comprehensive external validation can be presented as an online calculator, easily accessible in the clinic. It will constitute the first step towards more formal decision aids.
The findings can guide treatment, for instance whether to recommend medication or not, how frequently to monitor, how early to intervene, what is the best medication, what lifestyle factors to change, etc.
This knowledge can be incorporated in psychoeducation programmes such as the award-winning Bipolar Education Programme-Cymru (BEP-C), developed by Cardiff University.

4. Service-managers and policy makers
Perinatal mental health is a key government priority.
The risk prediction models will be the first step to develop more formal decision aids. The use of those in clinics may impact service development, and decrease costs to society.
The use of decision-aids can increase effectiveness of public service, by adding efficiency to the newly funded NHS perinatal services.
This project addresses a NICE research recommendation and fosters evidence-based policy-making.
Increased efficiency by using linked clinical data held in NHS Trust systems.
Furthermore, this project contributes to changing organisational culture and practice, by fostering shared decision-making of patients and clinicians.

5. Myself as a clinical academic
And finally, this fellowship will raise my profile and develop my skills. It will advance my career toward becoming a leading academic in perinatal psychiatry.