Properties of common anxiety scales among patients with bipolar disorder

Objectives: Almost half of the patients with a bipolar disorder (BD) have anxiety disorder(s) (AD) during their lifetime, but feasible measures for all AD are few. Furthermore, cognitive impairments can compromise reliability of existing scales, since many are needed for full coverage. Thus, we investigated how reliably patients responded to anxiety scales and any symptom overlap to propose future improvements to anxiety assessments. Methods: We collected 152 observations in patients with BD with the Clinically Useful Anxiety Outcome Scale, Social Phobia Inventory, Panic Disorder Severity Measurement, and Trauma Screening Questionnaire (in total, 57 items). The scales were analyzed as a set in a Rasch model. Results: During our analyses, we found indication that BD outpatients had difficulty differentiating response options to 70% (40/57) of items which were rescored or deleted. Only one case was misfitting (-2.65±.41). In total, 22 items were locally dependent and one indicated misfit. The final model included 25-items and fit the Rasch model (χ²=35.92, DF=50, p=.93). The model was unidimensional, without losing appropriate associations with depression (r = 0.62), suicidality (r = 0.37), and hypomania (r= -0.01). Limitations: Bolstering the size of less frequent subgroups should be accomplished in future work. Conclusion: A unidimensional rather than categorical approach to severity of anxiety might be both useful and feasible in this population. Further development of screens is necessary to enable systematic screening and measurement of anxiety in BD.