The current categorical split of mood disorders in bipolar disorders and depressive disorders has recently been questioned. Two highly unstable personality features, i.e. the cyclothymic temperament (CT) and borderline personality disorder (BPD), have been found to be more common in bipolar II (BP-II) disorder than in major depressive disorder (MDD). According to Kraepelin, temperamental instability was the ‘foundation' of his unitary view of mood disorders.

The aim was to assess the distributions of the number of CT and borderline personality items between BP-II and MDD. Finding no bi-modal distribution (a ‘zone of rarity') of these items would support a continuity between the two disorders.

Study setting: an outpatient psychiatry private practice. Interviewer: A senior clinical and mood disorder research psychiatrist. Patient population: A consecutive sample of 138 BP-II and 71 MDD remitted outpatients. Assessment instruments: The structured clinical interview for DSM-IV Axis I Disorders-Clinician Version (SCID-CV), the SCID-II Personality Questionnaire for self-assessing borderline personality traits (BPT) by patients, the TEMPS-A for self-assessing CT by patients. Interview methods: Patients were interviewed with the SCID-CV to diagnose BP-II and MDD, and then patients self-assessed the questions of the Personality Questionnaire relative to borderline personality, and the questions of the TEMPS-A relative to CT. As clinically significant distress or impairment of functioning is not assessed by the SCID-II Personality Questionnaire, a diagnosis of BPD could not be made, but BPT could be assessed (i.e. all BPD items but not the impairment criterion). The distribution of the number of CT and BPT items was studied by Kernel density estimate.

CT and BPT items were significantly more common in BP-II versus MDD. The Kernel density estimate distributions of the number of CT and BPT items in the entire sample had a normal-like shape (i.e. no bi-modality).

The expected finding, on the basis of previous studies and of the present sample features, was a clustering of CT and BPT items on the BP-II side of the curves. Instead, no bi-modality was present in the distributions of the number of CT and BPT items in the entire sample, showing a normal-like shape. By using the bi-modality approach, a continuity between BP-II and MDD seems supported, questioning the current categorical splitting of BP-II and MDD based on classic diagnostic validators.