Detecting bipolar disorder among treatment-seeking substance abusers

Bipolar disorder is increasingly recognized to have frequent comorbidity with substance use disorders, but may be difficult to diagnose among patients with active substance use. The purpose of this paper is to describe a brief, self-report form for the efficient detection of bipolar disorder. The 19-item form was piloted in 373 consecutive applicants for substance abuse treatment at an urban Veterans Affairs (VA) medical center. Results show reasonable internal consistency (alpha = .850) and high rates of manic symptomatology (36%), previous bipolar diagnosis (30%, 51% of whom report prior psychiatric hospitalization), and exposure to mood stabilizers (20%, 66% of whom reported therapeutic benefit). Comparison of nine different scoring algorithms with chart diagnosis as the validating criterion found that self-report of bipolar diagnosis was optimally sensitive. Either self-report of bipolar diagnosis with hospitalization or self-report of exposure to mood stabilizers with therapeutic response was optimally specific. Symptom self-report items had significantly poorer sensitivity and specificity (F = 7.60, p < .01). We conclude that questions pertaining to diagnostic and treatment history (especially hospitalization or therapeutic medication response) are considerably superior to symptom-based screening for clinically diagnosed bipolar disorder. Further work using structured interview as the diagnostic criterion is under way to validate this instrument.