Clinical correlates of age at onset distribution in bipolar disorder: a comparison between diagnostic subgroups

Background: Admixture analysis of age at onset (AAO) has helped delineating the clinical pro le of early onset (EO) bipolar disorder (BD). However, there is scarce evidence comparing the distributional properties of AAO as well as the clinical features of EO BD type 1 (BD1) with EO BD type 2 (BD2). To this end, we studied 515 BD patients (224 BD1, 279 BD2, and 12 BD not otherwise speci ed [NOS]) diagnosed according to DSM-IV-TR criteria. Methods: AAO was de ned as the rst reliably diagnosed hypo/manic or depressive episode according to diagnos- tic criteria. We used normal distribution mixture analysis to identify subgroups of patients according to AAO. Models were chosen according to the Schwarz’s Bayesian information criteria (BIC). Clinical correlates of EO were analysed using univariate tests and multivariate logistic regression models. Results: A two normal components model best tted the observed distribution of AAO in BD1 (BIC = −1599.3), BD2 (BIC = −2158.4), and in the whole sample (BIC = −3854.9). A higher number of EO BD2 patients had a depression- (hypo)mania-free interval (DMI) course, while a higher rate of (hypo)mania-depression-free interval (MDI) course was found in EO BD1. EO BD2 had also a higher rate of comorbidity with alcohol dependence compared to EO BD1. The latter nding was con rmed by multivariate logistic regression analysis. Conclusions: In conclusion, both BD1 and BD2 had bimodal AAO distributions, but EO subgroups had a diagnostic- specic clinical delineation.