LONDON (Reuters) - Assessment tools used to predict how likely a psychopathic prisoner is to re-offend if freed from jail are "utterly useless" and parole boards might just as well flip a coin when deciding such risks, psychiatrists said on Tuesday.
Publishing a study that found risk score tools are only around 46 percent accurate on how likely psychopathic convicts are to kill, rape or assault again, they said probation officers and judges should set little or no store by such tests.
They warned that clinicians carrying out such classifications must be aware of their severe limitations, and make sure prisoners undergo comprehensive psychiatric diagnosis before any risks assessment is made.
"If you apply these (tests) to somebody who is a psychopath, they're utterly useless, you might as well toss a coin," said Jeremy Coid, director of the forensic psychiatry research unit at Queen Mary University of London who led the study.
"They will not predict accurately at all," he told reporters at a briefing in London about his findings.
Coid and other forensic psychiatrists say the findings - which also showed the tools perform only moderately well in prisoners with disorders like schizophrenia, depression, drug and alcohol dependence - could have major implications for risk assessment in criminal justice systems.
"There are increasing expectations of public protection from violent behavior, and psychiatrists can be seriously criticized if they make wrong decisions," he said.
Seena Fazel, a consultant forensic psychiatrist at Britain's University of Oxford, said the reliability of the tests' predictive ability was so low that it might be best not to use them at all - and warned that at the very least, their results should only be noted by parole boards, rather than acted upon.
"If you're going to use these instruments, be aware of their strengths and limitations," he told reporters.
The estimated prevalence of adult psychopathy in the general population is around 1 percent, but that rises to between 15 percent and 25 percent among men in prison.
Coid, whose study was published in the British Journal of Psychiatry, analyzed data from 1,396 male prisoners in England and Wales who were interviewed between six and 12 months before their release. All the men were serving sentences of two or more years for a sexual or violent offences.
The prisoners were assessed for personality disorders, symptoms of schizophrenia, depression and drug and alcohol dependence, as well as being measured for psychopathy on a reputable scale known as Hare Psychopathy Check List.
After their release, data on their re-offending rates was added to the study, and showed that among three different re-offending risk assessment tools used before their release, the accuracy among psychopaths was below 50 percent.
While the tools were more accurate in predictions for prisoners with no mental health disorders - at around 75 percent accuracy - they were only around 60 percent right when it came to prisoners diagnosed with schizophrenia and depression.
For prisoners with anti-social personality disorders the predictive value of the tests ranged from poor to little more than chance, with an average 53.2 percent predictive accuracy. And for the 70 prisoners rated as psychopathic, none of the tests was statistically better than chance.
Coid said the results suggest it is time to question the expectations put on psychiatrists and psychologists asked to forecast future behavior of offenders, and to consider what can happen to their reputations if predictions are wrong.
"The easy solution is to be highly restrictive on who is released, and be risk averse. However, even for serious offenders, most will be released at some stage and someone has to carry out a risk assessment," he said.
"We need to prioritize the development of new assessment tools for these hard-to-predict groups."
Editing by Mark Heinrich