Anxiety sensitivity, trait anxiety, and the anxiety disorders

The recent debates and commentaries about the construct validity of the Anxiety Sensitivity Index (ASI) suggest several issues that remain to be resolved. First, the factor structure of the ASI remains to be clarified. Second, the distinction between this instrument and measures of trait anxiety has yet to be elucidated. A third issue concerns the extent to which the ASI and the trait version of the State-Trait Anxiety Inventory (STAI-T) can discriminate between panic disorder and other anxiety disorders. These issues were investigated by administering the ASI and the STAI-T to 142 spider-phobic college students and 93 psychiatric outpatients. In both samples, the ASI was found to be unifactorial. When the pool of ASI and STAI-T items was subjected to a factor analysis with oblique rotation, both samples yielded a two-factor solution. The STAI-T items were responsible for almost all the salient loadings on the first factor, and the ASI items were responsible for almost all the salient loadings on the second factor. In both samples the correlation between the factors was .39. These results suggest that the ASI is not simply a measure of trait anxiety. Items that are particularly important in defining the ASI concern the fear of bodily sensations. Fear of the social consequences of anxiety played a minor role in this factor. Panic-disordered patients were differentiated from patients with other anxiety disorders by ASI items pertaining to fear and catastrophic misinterpretation of bodily sensations. The full-scale score and item responses of the STAI-T were unable to differentiate between the anxiety groups. The findings support the construct validity of the ASI and the recent cognitive models of panic disorder. The results also suggest that a three-item version of the ASI may be used for the diagnosis of panic disorder.