A comparative study of violence risk assessment tools: A systematic review and metaregression analysis of 68 studies involving 25,980 participants

► Meta-analysis of violence risk assessments from 68 studies based on 25,980 participants. ► The predictive validity of tools measured by four outcome measures varies widely. ► Risk measures are more valid in older samples and when predicting violent offending. ► Tools designed for specific populations perform better. ► Actuarial tools and structured clinical judgment perform comparably.