Identifying risk factors of time to releases in patients with bipolar disorder using penalized likelihood model with shared gamma frailty compared with with-out frailty model

Identifying the risk factors of recurrence of a disease is important both for physicians and patients. Analyzing the first recurrence may lead to an inaccurate evaluation of risk factors as it may not reflect the history of disease completely and may result in the loss of some valuable information. This study was aimed to analyze the time up to the recurrent relapses of bipolar disorder as well as determining the risk factors that may lead to the recurrence of the disease.
In this retrospective longitudinal study, the data was extracted from records of 226 bipolar disorder patients with at least one relapse in Razi Psychiatric Educational-Medical Center in Tehran. Semi-parametric without frailty and semi-parametric shared gamma frailty models were fitted to data and analyzed using R software to estimate accelerated factors to identify some risk factors of relapses of bipolar disorder.
Of 226 patients, 65.3%were male and the rest female. Average age at onset of disease for female was 25± and 27± for males.43.3% had a suicide attempt, 65.1% history of hospitalization, and 58.5% family history. In semi parametric without frailty model age at onset of disease (p=0.009) detection stability (p with respect to recurrent property of bipolar disorder and existing correlation among relapse times, continuous efforts and special preventive treatments to decrease relapse probability especially in patients who experience relapses in low ages, those with history of hospitalization and patient who had not detection stability are recommended.