Authors' objectives:

This review reports on the validity of methods for predicting violence in the community by psychiatric patients

Authors' recommendations: A report by the Swedish Council on Technology Assessment in Health Care (SBU) has reviewed, classified, and graded the scientific literature on the predictive validity of methods for assessing the risk of violence in psychiatry. Papers were published between 1970 and February 2005 and indexed in the PubMed or PsychInfo databases. The study population included adult patients (>18 years) from general and forensic psychiatry. Studies validating methods to assess the risk for violence at inpatient units were excluded. The search identified 37 unique datasets. Moderate scientific evidence indicates that risk assessments are superior to chance in predicting the probability of violence in males. There is no evidence that the methods yield valid results in women. Few studies have addressed the validity of risk assessment in ethnic minorities and in short-time predictions, ie, the risk of violence within days, weeks, or months after the risk assessment. Even in the "best-case" scenario, the evidence suggests that assessments based on clinical judgement or on the most well-documented instruments (the HCR-20 and VRAG) incorrectly classify one in three to one in four patients.

Authors' methods: Review

Project Status: Completed

URL for project: http://www.sbu.se/Published

Year Published: 2005

English language abstract: An English language summary is available

Publication Type: Not Assigned

Country: Sweden

Organisation Name: Swedish Agency for Health Technology Assessment and Assessment of Social Services

Contact Address: P.O. Box 3657, SE-103 59 Stockholm, Sweden. Tel: +46 8 4123200, Fax: +46 8 4113260

Contact Name: registrator@sbu.se

Contact Email: registrator@sbu.se

Copyright: Swedish Council on Technology Assessment in Health Care (SBU)