Recurrent Events Model Application in Determining the Risk Factors of Bipolar Disorder Recurrence

Area of Science:

• Psychiatry
• Biostatistics
• Epidemiology

Background:

• Recurrent events data, common in survival analysis, exhibit correlations within individual observations.
• Bipolar disorder (BD) relapse is a significant clinical event impacting patient outcomes.
• Understanding factors influencing BD relapse timing is crucial for effective management.

Purpose of the Study:

• To analyze time-to-relapse in bipolar disorder (BD) patients.
• To identify factors associated with recurrent BD relapse intervals.
• To apply recurrent events models, specifically Cox models with shared frailty, for this analysis.

Main Methods:

• Retrospective analysis of 104 BD patients with at least one relapse (2001-2015).
• Utilized Cox proportional hazards models with gamma and log-normal shared frailty distributions.
• Statistical analyses performed using R and SPSS software; significance level set at p < 0.05.

Main Results:

• The Cox model with log-normal shared frailty demonstrated the best fit for BD recurrent survival data.
• Marital status and history of veteran status were identified as significant factors influencing time intervals between BD relapses.
• Subsequent relapse risk was lower in married and veteran patients.

Conclusions:

• The frailty effect in the models confirmed the correlation between recurrent time intervals for BD relapses within individuals.
• Marriage and social support, particularly for veterans, appear to be protective factors against recurrent bipolar disorder episodes.
• These findings highlight the importance of social and emotional support in reducing BD relapse rates.