Diagnostic reliability of the DSM-III-R anxiety disorders. Mediating effects of patient and diagnostician characteristics.

Factors influencing diagnostic reliability were examined using a sample of 267 patients who underwent two independent administrations of the Anxiety Disorders Interview Schedule--Revised (ADIS-R). Several potential mediators were examined in terms of their association with interrater agreement as measured by kappa coefficients. These included comorbidity, disorder severity, diagnostician training, patient education, presence of behavioral feature, and time separating interviews. Results suggest that several factors appear to have an important effect on diagnostic agreement, most notably, the presence or absence of comorbidity and the presence or absence of overt behavioral symptoms. Details of these influences and the impact of other selected mediators on the diagnosis of anxiety disorders are presented.