Admixture analysis of age at onset in first episode bipolar disorder

Background: Many studies have used the admixture analysis to separate age-at-onset (AAO) subgroups in bipolar disorder, but none of them examined first episode patients.

Objective: The purpose of this study was to investigate the influence of clinical variables on AAO in first episode bipolar patients.

Methods: The admixture analysis was applied to identify the model best fitting the observed AAO distribution of a sample of 194 patients with DSM-IV diagnosis of bipolar disorder and the finite mixture model was applied to assess the effect of clinical covariates on AAO.

Results: Using the BIC method, the model that was best fitting the observed distribution of AAO was a mixture of three normal distributions. We identified three AAO groups: early age-at-onset (EAO) (µ=18.0, σ=2.88), intermediate-age-at-onset (IAO) (µ=28.7, σ=3.5), and late-age-at-onset (LAO) (µ=47.3, σ=7.8), comprising 69%, 22%, and 9% of the sample respectively. Our first episode sample distribution model was significantly different from most of the other studies that applied the mixture analysis.

Limitations: The main limitation is that our sample may have inadequate statistical power to detect the clinical associations with the AAO subgroups.

Conclusions: This study confirms that bipolar disorder can be classified into three groups based on AAO distribution. The data reported in our paper provide more insight into the diagnostic heterogeneity of bipolar disorder across the three AAO subgroups.