Bayesian joint longitudinal and survival modeling of bipolar symptom burden and time to symptomatic recovery of patients with bipolar disorder at Jimma University Medical Center, Jimma, Ethiopia.

Abstract

Background: Bipolar disorder is a mental health problem that primarily affects mood. Symptoms of bipolar disorder are extreme irritability or agitation, a period of feeling empty, loss of interest in usual activities, sleep problems Etc. Symptomatic recovery is a dimensional measure that refers to improvement in the magnitude of symptoms. This investigation aims to determine the association between the burden of symptoms and time to symptomatic recovery of bipolar disorder that may increase more awareness about this disorder.

Methods: A Bayesian joint modeling of longitudinal and survival data was proposed to examine the association between the burden of symptoms and time to symptomatic recovery of bipolar disordered individuals at Jimma University Medical Center, Jimma, Ethiopia. The data in this investigation were retrospective longitudinal data and survival data from all the admitted follow-up of bipolar disorder patients from September 2018 to January 2020.

Results: From the total of 257 bipolar disorders, about 116(45.1%) of them experienced an event of recovery. The time interval of follow up, age, the interaction between time interval of follow up and adolescent first onset of the disease, the interaction between time interval of follow up and event of relapse, the interaction between linear time interval of follow up and existence of other cofactors, and the interaction of substance abuse and chewing khat have significantly affected the log expected burden of bipolar symptoms. In survival sub-models, the covariates; divorced, event of relapse, mixed type of episodes significantly affect the time to symptomatic recovery at 95% confidence level. The association between burden of bipolar symptoms and time-to- symptomatic recovery is explained by α₀ = - 8.403 with a [- 11.157,- 6.576]. It associates longitudinal count, the burden of symptoms, and time-to-symptomatic recovery of bipolar disorder using shared random effect parameters. There was a significant negative relationship between the subject-specific random intercept (baseline) of the burden of symptoms and the time to symptomatic recovery of bipolar disorder. Their 95% credible intervals exclude zero.

Conclusions: This study using the Bayesian joint modeling of longitudinal and survival has revealed a strong negative relationship between the event of recovery and the burden of bipolar symptoms at the baseline time. The study indicates that at the beginning, since the burden of bipolar symptoms is high, the chance of symptomatic recovery is low. And, we hypothesize that individuals with a higher initial symptom burden or a slower rate of symptom reduction (captured by b_i) will experience a longer time to recovery. So, bipolar disorders at the initial follow-up need exceptional service and treatment.