We investigated effects of antimanic treatments on specific aspects of mania, prediction of response, and the existence of naturalistic subgroups of patients with different treatment response in 179 inpatients randomized to antimanic treatment with lithium, divalproex, or placebo. Psychiatric symptom ratings were conducted by clinicians and nurses before and during treatment. Factor analysis using physician and nurse rating scales, followed by a cluster analysis, yielded anxious-depressive, psychotic, classic, and irritable subtypes. We compared: (1) treatment effects on factor scores; (2) responses to treatment across subtypes; and (3) pattern of symptom change with each treatment. The anxious-depressed subtype did not respond to any treatment; the psychotic and classic subtypes responded similarly to lithium and to divalproex; and the irritable-dysphoric subtype responded better to divalproex than to lithium. Overall, divalproex improved impulsivity and hostility significantly more than placebo, and lithium or divalproex improved hyperactivity more than placebo. These data suggest that there are naturalistic subtypes of manic episodes with different responses to treatment.
