The SAVRY Improves Prediction of Reoffending: A Naturalistic Longitudinal Comparative Study

Purpose: This study investigated the utility of the risk assessment "Structured Assessment of Violence Risk in Youth'' (SAVRY) within the social services in Stockholm County, Sweden. Method: SAVRY assessments of 56 adolescents were compared to assessments guided by another instrument (Adolescent Drug Abuse Diagnosis [ADAD]; n = 38) and assessments without support of a structured method (n = 38). Results: The results showed that social workers conducting SAVRY assessments documented a significantly larger number of risk and protective factors compared to the other assessments, and these factors predicted, with a few exceptions, reoffending to a larger extent. SAVRY summary risk rating significantly predicted the occurrence of serious violent crimes (area under the curve [AUC] = .80, p < .01) and less serious violence (AUC = .70, p < .05). Conclusions: SAVRY performed at least as well in naturalistic settings as in previous studies conducted in more controlled environments. Furthermore, the SAVRY performed better than the other structured instrument (ADAD).