Association of Rapid Mood Switching With Panic Disorder and Familial Panic Risk in Familial Bipolar Disorder

Logistic regression analyses were performed with generalized estimating equations (8), which appropriately account for potential correlations in the residual variation in the outcome (i.e., variation not explained by the model) among individuals from the same family. Thus, generalized estimating equations allow us to draw valid inferences from the data about the greater risk of rapid cycling in a family with panic. Statistical significance was defined as p<0.05 (two-tailed).

We characterized each family as having high, indeterminate, or low risk for familial panic on the basis of the number of family members with bipolar disorder (i.e., bipolar I, bipolar II with recurrent major depression, or schizoaffective-manic) or recurrent unipolar depression who had experienced panic. In the families characterized as “low panic” (N=56), no family member had experienced panic attacks. In the families characterized as “high panic” (N=76), one family member had panic disorder and at least one other family member had panic attacks or panic disorder. In the families characterized as “indeterminate panic” (N=76), only one family member had experienced panic disorder and no one else had panic attacks, or one or multiple family members had panic attacks but no one had panic disorder. Although this scheme has the disadvantage of favoring larger families in the high-panic than in the low-panic category (3.4 versus 2.4 members per family, respectively), in our experience this is the least problematic way to define familial aggregation (2). We tested the association of familial panic with rapid mood switching and included as covariates panic disorder (which appeared on inspection to be associated with rapid switching) and sex (female sex has been associated with both panic disorder and with rapid cycling [2, 9]).