This document was published more than 2 years ago. The nature of the evidence may have changed.
- Psychiatric risk assessment methods are more accurate than chance in predicting the propensity of male patients to commit future acts of violence in the community (Evidence Grade 2). Evidence is lacking that the methods provide reliable results for female patients. The accuracy of risk assessments may be defined as the percentage of patients who are correctly identified as subsequently committing acts of violence. The review of the literature indicates that 70–75 percent accuracy can be expected on the basis of the best research conducted thus far.
- Risk assessments can predict the propensity of relevant forensic (Evidence Grade 3) and general (Evidence Grade 2) psychiatric patients to commit acts of violence in the community for the next few years. However, there is insufficient scientific evidence to support more short-term risk assessments, i.e., for the days and weeks after a patient has left the clinic.
- No studies relevant to Swedish society have analyzed whether the reliability of the methods varies from one ethnic group to another.
- Both clinical evaluations and checklists of predefined instruments may be used in making risk assessments (Evidence Grade 2). The Violence Risk Appraisal Guide (VRAG) and the Historical Clinical and Risk Management Scheme (HCR-20), the two most widely used instruments, are equally valid (Evidence Grade 3). The uncertainty (inaccuracy) of forecasts based on instrumentalized assessments is 25–30 percent.
- Research is urgently needed in five different areas:
- Current assessments are based largely on non-controllable risk factors such as age and previous criminal behavior. Given that the purpose of assessing risk is to take effective action for the prevention of violence, more knowledge is required about risk factors that are controllable, as well as those that women face. Research on neurobiological risk factors and markers may improve forecasting tools.
- Additional studies are needed to establish whether currently available methods are able to forecast the risk for violent acts in the community over the short term (days or weeks).
- More evaluations of risk assessment methods are required in settings representative for the kinds of conditions that exist in Sweden.
- There is a great need for controlled studies on various monotherapies and combination therapies.
- Quality registers are lacking for both general and forensic psychiatry. If risk assessments are to be based on more reliable information and knowledge, such registers must be set up as quickly as possible.
How to cite this report: SBU. Psychiatric risk assessment methods are violent acts predictable?. Stockholm: Swedish Council on Health Technology Assessment in Health Care (SBU); 2005. SBU report no 175 (in Swedish).