Concerns over accuracy of tools to predict repeat offending

Risk assessment tools are widely used in psychiatric hospitals and criminal justice systems around the world to help predict violent behaviour.

They are often used to influence decisions at different stages of the criminal justice process – on pre-trial detention, bail determination, sentencing and release – and also for psychiatric patients, particularly those with histories of violence and criminality. Yet their predictive accuracy remains uncertain.

So an international research team led by Dr Seena Fazel at the University of Oxford looked at how good the tools were at predicting repeat offending.

They analysed risk assessments conducted on 24,827 people from 13 countries including the UK and the US. During these studies, which lasted over of 50 months, an average 5,879 people (24%) went on to offend. The results are published in the British Medical Journal.

'These risk assessment tools were designed to measure the likelihood of future antisocial behaviour, including violent, sexual, and criminal offending,' says Dr Fazel of Oxford's Department of Psychiatry. They typically involve a checklist of items which requires interviewing an individual face to face and collecting background information from third-party sources.

The researchers found that the risk assessment tools produce high rates of false positives, often wrongly identifying individuals as being at high risk of repeat offending.

The accuracies of the assessments in identifying risky persons were not much better than would be expected purely by chance. For example, 41% of individuals judged to be at moderate or high risk by violence risk assessment tools went on to violently offend. 23% of those judged to be at moderate or high risk by sexual risk assessment tools went on to sexually offend.

However, the tools appeared to be more effective at screening out individuals at low risk of future offending. Of those predicted not to violently offend, 91% did not.

'Our review suggests that these tools are better at identifying individuals at low risk of future violence,' says Dr Fazel. 'Instruments designed to predict more specific outcomes (violent and sexual crimes) performed better than those aimed at predicting any crime.'

He suggests that failure to understand the limitations of these tools could in all likelihood lead to some prisoners being locked up for too long for fear of reoffending, or alternatively some risky prisoners being released without due care.

Although risk assessment tools are widely used in clinical and criminal justice settings, their predictive accuracy varies depending on how they are used, the researchers say.

'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,' Dr Fazel and colleagues write in the BMJ.

Things could be improved, Dr Fazel believes, if there was more transparency in how findings related to these tools were reported in the literature, not just focusing on one statistic that purports to show great accuracy, and also having an international register of all such studies.

He thinks that the risk assessment tools could also be used differently. If they were used primarily as screening instruments to identify people unlikely to offend again, it would allow the others to be more carefully assessed and managed, he suggests. Compounding factors like drug and alcohol abuse, failure to stick to treatments, and mental illness could also be treated to lesson chances of further offences.