The Swedish National Board on Health and Welfare recommends that structured assessment instruments should be used in medical as well as in social treatment of substance misusers. These should be validated in a Swedish context. Instruments for diagnoses of dependence/abuse (DSM-IV) and dependence/harmful use (ICD-10) have been used in Sweden for many years, although not yet validated in Swedish. ADDIS, the Swedish version of the American SUDDS, is used in four Nordic countries and the most often used diagnostic instrument in Sweden. This article investigates the psychometric properties of ADDIS alcohol module, including discriminant and construct validity and internal consistency. The two main constructs in DSM - dependence and abuse - as well as the seven criteria for dependence and the four criteria for abuse are studied. Further, the value of each of the 44 specific items in ADDIS for capturing these criteria is studied.

Two samples are explored: 1) a clinical sample (n = 349; incl. 129 women) and 2) a sample of 400 men convicted for driving while intoxicated. Mean age was the same (41 ys.). Using discriminant analyses on lifetime prevalence, the items correctly classify 94% of the cases in the two samples. Using one-factor principal component analysis to explore homogeneity of the combined samples, all 28 items on dependence and 15 of 18 items on abuse have loadings above 0,40 (R2 dependence = 0,46; abuse = 0,40). Separate analyses of the two samples, as well as on women, show similar results. Cronbach's alpha is excellent for dependence and satisfactory for abuse in all analyses. Analyses of specific criteria show satisfactory results on dependence and acceptable on abuse. Minor revisions are proposed to make ADDIS more user-friendly and to improve some specific items.

In conclusion: ADDIS has acceptable to excellent discriminant and construct validity as well as internal consistency and captures the specific criteria of DSM-IV. It has the preconditions for sensitive assessment of alcohol use disorders in men and women.