Relevant demographic and clinical information was extracted from electronic medical records by a hospital administrator and provided to the study first author in a single Microsoft Excel file. Using standard form fields from the hospital admission report, the hospital administrator used a computer algorithm to extract the following data: patient age, sex, race, ethnicity, civil status, history of prior psychiatric hospitalization, type of index hospitalization (i.e., inpatient vs. partial), self-reported pharmacotherapy for BD at hospital admission (i.e., yes/no), history of suicide attempt, and global assessment of functioning (GAF). DSM-IV-TR diagnostic codes were extracted from the admission and discharge reports in order to confirm the presence of a BDI diagnosis, and were further quantified into the following variables: polarity of BD mood episode (depressive vs. manic/mixed), presence of psychotic features (e.g., hallucinations and/or delusions) at hospital admission, and presence of any current anxiety disorder, cannabis use disorder (CUD), nicotine dependence, alcohol use disorder (AUD), or any other substance use disorder (SUD). ICD diagnostic codes, also extracted from the discharge report, were used to establish the presence of any comorbid medical condition.
