Diagnosing comorbidity in substance abusers: a comparison of the test-retest reliability of two interviews

This study examines the test-retest reliability of two interview schedules (computer- and clinician-administered) in diagnosing lifetime comorbidity in treated substance abusers. The Computerized Diagnostic Interview Schedule (C-DIS) and the Structured Clinical Interview for DSM-III-R (SCID) were both administered to 173 substance abusers after random assignment to one of two groups. Within 1 to 2 weeks, subjects in the first group repeated the C-DIS and subjects in the second group were reinterviewed by a different clinician, blind to the results of the initial SCID. Both instruments showed good to excellent reliability for DSM-III-R psychoactive substance use disorders with kappas ranging from .50 to .89 for individual disorders. However, the reliability of comorbid other mental disorders was substantially poorer on both instruments, particularly the SCID. C-DIS kappas ranged from -.05 for generalized anxiety to .70 for simple phobia. SCID kappas ranged from .31 for panic disorder to .83 for antisocial personality disorder. Anxiety disorders as a category, some phobic disorders, and antisocial personality disorder showed acceptable levels of test-retest reliability on both instruments. There was a trend for borderline or threshold cases to account for some of the disagreement on the C-DIS. Differences of opinion between clinicians on organicity accounted for some of the disagreements on panic disorder and major depression. The C-DIS, unlike the SCID, tended to diagnose more disorders at initial interview, perhaps a result of its tedious probe structure. Neither instrument should be administered only once to provide a reliable lifetime diagnostic profile of comorbidity in substance abusers.