Protocol for developing and validating a multivariable prediction model to individualize the risk of recurrence of bipolar disorder in the perinatal period

Introduction For women with bipolar disorder, childbirth is a high-risk period with 40–50% experiencing a recurrence and 20% developing a severe episode of postpartum psychosis. Bipolar episodes in the perinatal period affect women and their families. Managing bipolar disorder in pregnancy and postpartum is a challenge. There is lack of literature to inform that and an urgent need for more data. Objectives To develop and validate a risk prediction model for individual prognosis of the risk of recurrence of bipolar disorder for women in the perinatal period. Aims To provide evidence-based information to help women and the clinicians that look after them make decisions about their care, taking into account the most recent scientific knowledge and their individual characteristics. Methods The development of the model will be done in retrospective data from a large clinical cohort from the Bipolar Disorder Research Network (BDRN.org). The validation will be done in a prospectively recruited sample. Participants will be 2181 parous women with a lifetime diagnosis of bipolar disorder from BDRN and 300 prospectively recruited pregnant women with a history of postpartum psychosis or bipolar disorder. Predictors will be chosen based on clinical experience and literature, from data collected via semi-structured interview (in pregnancy and 3 months postpartum, medical and psychiatric notes) e.g. medication, smoking, parity, obstetric complications and sleep. Results N/A. Conclusions We will present the full prediction model (regression coefficients and model intercept) and report performance measures (with CIs). We will discuss its potential clinical use and implications for future research.