Association between age of onset and mood in bipolar disorder: Comparison of subgroups identified by cluster analysis and clinical observation

Background: This study compared subgroups identified by cluster analysis and clinical observation by evaluating the association between the age of onset of bipolar disorder and self-reported daily mood ratings. Methods: Two hundred and seventy patients with bipolar disorder provided daily self-reported mood ratings for about 6 months returning 55,188 days of data. The age of onset subgroups were determined both using previously defined cutoff values based upon clinical observation (<= 12 years, 13-19 years, 20-29 years, >29 years), and model-based cluster analysis. Demographic characteristics were compared in the age of onset subgroups. Univariate general linear models with age of onset subgroups and other demographic variables as fixed factors and covariates were used to analyze the percent of days depressed, euthymic and hypomanic/manic. Results: Using the predetermined subgroups, demographic differences were found between the four subgroups in the diagnosis of bipolar I/II, years of illness, age and use of lamotrigine. Post-hoc pairwise comparison found that patients with an age of onset less <= 12 years spent more days hypomanic/manic: 16.4 percent versus 8.0 for patients with an age of onset between 13 and 19 years (p = 0.006) and 8.2 percent for patients with an age of onset between 20 and 29 years (p = 0.031). The majority of the additional days of hypomania/mania occurred outside of an episode. Model-based cluster analysis found a mixture of 2 distributions of onset with peaks at age 15.1 years (SD = 4.7) and 27.5 years (SD = 10.2). Analysis of these two subgroups detected no significant differences in demographic characteristics or mood ratings. Conclusion: Age of onset subgroups arising from clinical observation may be more useful than those determined by cluster analysis. (C) 2010 Elsevier Ltd. All rights reserved.