A comparison of the predictive validity of four sets of baseline ASI summary indices

This study compared the long-term predictive validity of original and new b aseline Addiction Severity Index summary scores in methadone patients. The indices included the original Interviewer Severity Ratings (ISRs) and the n ew Clinical Indices (CIs), which use both lifetime and recent problem infor mation, and the original Composite Scores (CSs) and Evaluation Indices (EIs ), based on recent problems only. Outcomes were medical hospitalization, em ployment, alcohol intoxication, drug hospitalization, and psychiatric hospi talization in Months 7-24 poststudy entry and criminal charges in Months 0- 24. Hierarchical logistic regression analyses were used in which 1 index wa s entered first and the other in the 2nd step. The reverse order of entry w as used in a 2nd analysis. A final analysis set compared the best predictor from each of the 2 prior analysis sets, The Cls were superior to the other indices in predicting 3 of 6 outcomes (psychiatric hospitalization, drug h ospitalization, and criminal char.-es); the EI was the best predictor of al cohol intoxication, and the CS the best predictor of unemployment.