Clinician-rated versus self-rated screening for bipolar disorder among inpatients with mood symptoms and substance misuse

Background: Self-rated screening studies suggest higher prevalence rates for bipolar disorder than previously thought, but the validity of self-administered diagnostic tools has not been well established in mood-disordered patients with substance misuse. Method: We conducted a cross-sectional assessment of 113 English-speaking, nonintoxicated adults aged 18-69 years who were seeking treatment for mood or anxiety symptoms and substance use symptoms. (Subjects with anxiety complaints at initial presentation were included to possibly increase the pool of subjects with mood symptoms upon formal evaluation.) Subjects were consecutively evaluated from January 2010 through May 2011 at the time of voluntary admission to a private, not-for-profit psychiatric hospital. All subjects completed the Mood Disorder Questionnaire (MDQ), followed by a psychiatrist's review of their responses, using the MDQ as a semistructured interview. MDQ item and total agreements were compared for patient self-report versus clinician interview, alongside analysis of reasons for discordance. DSM-IV-TR criteria were used as the basis for diagnoses of bipolar disorder, other mood disorders, and substance use disorders. Results: An MDQ positive (+) status was scored more often by subjects through self-report (56%) than by the clinician after review of subjects' MDQ responses (30%) (P < .001). Patients' self-rated MDQ(+) status had high sensitivity (0.77), modest specificity (0.52), low positive predictive value (0.38), and high negative predictive value (0.86) for bipolar I or II diagnoses. MDQ item ratings having the lowest patientclinician concordance were irritability (κ = 0.12), racing thoughts (κ = 0.15), and distractibility (κ = 0.10), while highest concordance was observed for excessive spending (κ = 0.54), increased goal-directed activity (κ = 0.59), and hypersexuality (κ = 0.77). Patient-clinician MDQ item discordance most often resulted from patients' affirmative mania symptom endorsements during past intoxication states. Logistic regressions indicated that discordant patient-clinician MDQ total scores were significantly associated with the number of lifetime substances of abuse (odds ratio = 1.43; 95% CI, 1.02-1.99) but not with any 1 particular substance. Conclusions: Past or current substance misuse confounds the reliability of MDQ bipolar self-assessment screening. Clarification of MDQ self-rated responses via interview probing yields a more sensitive and specific indicator of likely bipolar diagnoses.