Concerns over accuracy of tools to predict risk of repeat offending
Current evidence not strong enough for definitive decisions to be made, say experts
Research: Use of risk assessment instruments to predict violence and antisocial behaviour in 73 samples involving 24,827 people: systematic review and meta-analysis
Tools designed to predict an individual’s risk of repeat offending are not sufficient on their own to inform sentencing and release or discharge decisions, concludes a study published on bmj.com today.
Although they appear to identify low risk individuals with high levels of accuracy, the authors say “their use as sole determinants of detention, sentencing, and release is not supported by the current evidence.”
Risk assessment tools are widely used in psychiatric hospitals and criminal justice systems around the world to help predict violent behaviour and inform sentencing and release decisions. Yet their predictive accuracy remains uncertain and expert opinion is divided.
So an international research team, led by Seena Fazel at the University of Oxford, set out to investigate the predictive validity of tools commonly used to assess the risk of violence, sexual, and criminal behaviour.
They analysed risk assessments conducted on 24,827 people from 13 countries including the UK and the US. Of these, 5,879 (24%) offended over an average of 50 months.
Differences in study quality were taken into account to identify and minimise bias.
Their results show that risk assessment tools produce high rates of false positives (individuals wrongly identified as being at high risk of repeat offending) and predictive accuracy at around chance levels when identifying risky persons. For example, 41% of individuals judged to be at moderate or high risk by violence risk assessment tools went on to violently offend, while 23% of those judged to be at moderate or high risk by sexual risk assessment tools went on to sexually offend.
Of those judged to be at moderate or high risk of committing any offence, just over half (52%) did. However, of those predicted not to violently offend, 91% did not, suggesting that these tools are more effective at screening out individuals at low risk of future offending.
Factors such as gender, ethnicity, age or type of tool used did not appear to be associated with differences in predictive accuracy.
Although risk assessment tools are widely used in clinical and criminal justice settings, their predictive accuracy varies depending on how they are used, say the authors.
“Our review would suggest that risk assessment tools, in their current form, can only be used to roughly classify individuals at the group level, not to safely determine criminal prognosis in an individual case,” they conclude. The extent to which these instruments improve clinical outcomes and reduce repeat offending needs further research, they add.