Future research will also be necessary to address other limitations of the current study. As noted above, the study relied upon retrospective chart review, which benefits from data collected within the context of routine clinical care, but also has a number of limitations. Although we were conservative in selecting those cases that had a BDI diagnosis at both admission and discharge, thereby reducing the heterogeneity of the sample, chart diagnoses were not confirmed by structured interview and secondary conditions (including SUDs) may not have always been assessed or documented in the electronic medical record. As a retrospective chart review, choice of demographic and clinical variables to evaluate in the current study was limited to what was available in the hospital admission report. As such, we were unable to evaluate associations between CUD comorbidity and certain relevant core features of BD such as age of onset [10,18], overall illness severity [6,20,28], history of mixed episodes [27] and rapid cycling [6,17], history of early adversity or trauma [29], current psychotherapy and pharmacotherapy treatment history and adherence [8,9], as well as other relevant
