Validation of the youth mood recurrences risk calculator in an adult sample with bipolar disorder

Abstract

Background: The ability to predict an individual's risk of mood episode recurrence can facilitate personalized medicine in bipolar disorder (BD). We sought to externally validate, in an adult sample, a risk calculator of mood episode recurrence developed in youth/young adults with BD from the Course and Outcome of Bipolar Youth (COBY) study.

Methods: Adult participants from the National Institute of Mental Health Collaborative Depression Study (CDS; N=258; mean(SD) age=35.5(12.0) years; mean follow-up=24.9 years) were utilized as a sample to validate the youth COBY risk calculator for onset of depressive, manic, or any mood episodes.

Results: In this older validation sample, the risk calculator predicted recurrence of any episode over 1, 2, 3, or 5-year follow-up intervals, with Area Under the Curves (AUCs) approximating 0.77. The AUC for prediction of depressive episodes was about 0.81 for each of the time windows, which was higher than for manic or hypomanic episodes (AUC=0.72). While the risk calculator was well-calibrated across the range of risk scores, it systematically underestimated risk in the CDS sample by about 20%. The length of current remission was a highly significant predictor of recurrence risk in the CDS sample.

Limitations: Predominantly self-reported White samples may limit generalizability; the risk calculator does not assess more proximal risk (e.g., 1 month).

Conclusions: Risk of mood episode recurrence can be predicted with good accuracy in youth and adults with BD in remission. The risk calculators may help identify higher risk BD subgroups for treatment and research.

Keywords: Bipolar disorder; Disease prognosis; Longitudinal course; Prospective cohort study; Recurrence; Risk calculator.