• bipolar disorder;
  • diagnosis;
  • epidemiology;
  • mania;
  • mixed mania

Introduction: Few large clinical epidemiological studies have been undertaken comparing subjects meeting criteria for mixed and pure states of bipolar disorder. In part, the difficulty comparing these states emanates from confusion in their diagnostic separation. In the current report, we use a definition derived from receiver operating characteristic (ROC) curve analysis as an alternative to the DSM-IIIR/IV definition, and we compare the two subtypes of manic episodes. Methods: Three hundred and sixty-six patients meeting DSM-IIIR criteria for bipolar disorder, manic or mixed, were categorized using newly described criteria for mixed states. The two subtypes were compared on demographic variables and clinical history variables, using multiple analysis of variance with post hoc univariate F tests. The same analyses were conducted using the DSM-IIIR-defined subtypes. Results: Using the ROC criteria, 79 subjects (21.6%) were characterized as mixed, in contrast to 51 subjects (13.9%) using DSM-IIIR criteria for bipolar disorder, mixed. The ROC-defined mixed manic group comprised more Caucasians and more females. Age of first psychiatric hospitalization was earlier and duration of illness longer in the mixed group. First episodes were unlikely to be categorized as mixed (<5%). When the DSM-IIIR definition was employed, differences were not demonstrated. Conclusions: An earlier age of first psychiatric hospitalization and increased duration of illness, as well as a lower frequency of mixed subtype of manic episode during first hospitalization, are compatible with the view that mixed manic episodes occur more frequently later in the course of bipolar disorder. Moreover, differences in race, sex, and clinical histories of subjects in mixed episodes tend to support the separation of mixed mania as a diagnostic subtype of bipolar disorder.