Initial evidence for the reliability and validity of a "Lite" version of the Addiction Severity Index

Abstract

Purpose: To evaluate the psychometric properties of a shortened version of the baseline ASI-5, the ASI-L-VA.

Method: Two samples were recruited from intensive outpatient treatment and a methadone maintenance clinic. For Sample A (n=145), two versions of the Addiction Severity Index (i.e., ASI-5 and ASI-L-VA) were administered several days apart in counterbalanced order by different interviewers. Sample B (n=50) was similarly administered the standard ASI-5 twice.

Results: For Sample A, the internal consistency (coefficient alphas) of 11 of 19 summary scores derived from the ASI-5 were good, 4 fair, and 4 unacceptable. The results for the ASI-L-VA summary scores indicated that eight were good, six fair, and five unacceptable. The correlations between ASI problem areas were generally low for both versions (supporting the independence of the ASI areas), and none of the t-tests comparing corresponding correlations between the ASI-5 and ASI-L-VA approached statistical significance. The Sample A intraclass correlation coefficient (ICC) results evaluating agreement of the summary scores derived from the ASI-5 at one timepoint and those derived from ASI-L-VA at another point (i.e., concurrent validity) revealed at least fair agreement in all but one instance. Additionally, a comparison of the ICC results for Samples A and B (i.e., ASI-L-VA/ASI-5 versus ASI-5/ASI-5, respectively) revealed that in 13 of 26 cases the ICCs were at the same level of agreement. When level of agreement was discordant, in nine cases the ICCs comparing the ASI-5 and ASI-L-VA exhibited greater agreement and in four cases the ICCs comparing two ASI-5 administrations exhibited greater agreement.

Conclusions: The ASI-L-VA, a reduced item set from the ASI-5, yielded similar information on problem severity as the standard ASI-5.