It’s long been assumed that antidepressants can “flip” patients into mania/hypomania. I don’t think this is disputed. There isn’t uniform agreement, however, about whether this is simply a side effect of medication or whether it reveals an underlying bipolar diathesis. In other words, those who switch from depression to mania may not be unipolar depressives at all, but as-yet-undiagnosed bipolar patients. The DSM-IV TR specifically excludes antidepressant-induced hypomania/mania from the diagnosis of bipolar disorder and indeed states that these presentations should not be considered manic episodes. There are some who work in the field who do consider that those who develop manic symptoms when given antidepressant treatment represent a group of patients who fit more readily into the diagnosis of bipolar disorder. They rely on evidence showing that many in this group later go on to develop spontaneous mania. There are epidemiologic data which suggest the relationship, e.g. the incidence of bipolar in families is similar. Studies suggest that 3-10% of unipolar patients switch when given an antidepressant.
Although the mechanism of action remains unknown, we assume it’s similar to the mechanism of action involved in switches of diagnosed bipolar patients from depression to mania when given an antidepressant. Although the data is not very robust, there seem to be some predictive factors — family history of bipolar disorder, early age of onset of depression, atypical features and hyperthermic temperaments.