Abstract

There has been a growing optimism regarding the accuracy of structured instruments for violence risk assessment in mentally disordered offenders. However, several issues pertaining to forensic assessments of risk remains unsolved, one of which is the relationship between risk assessment and risk management. In this paper we argued that research that evaluate efforts to assess risk must take into account the level of risk management in the sample. We attempted to illustrate this using prospective follow-up data on the frequency and type of inpatient violent behaviours and their relationship to risk management within the context of care. Risk assessments were made upon admission to hospital with the Historical-Clinical-Risk assessment (HCR-20, Webster et al., 1997) in 54 forensic patients followed through three different risk management conditions: High security risk management, medium risk management, and only risk monitoring (low). The results showed large differences in baserate and type of violence in the three management conditions. Results also suggested that the HCR-20 accurately assessed risk in medium and low security conditions, but not in the high security condition. We conclude that the findings reinforce rather than contraindicate the usefulness of the HCR-20 in for clinical practice.