Risk and prediction of violent crime in forensic psychiatry

Abstract: Objective: To test the predictive accuracy for violent recidivism of the age at onset of substance abuse, the platelet MAO-B activity, and various combinations of criminological and clinical risk factors among violent offenders in a prospective Swedish follow-up study. Subjects: One hundred violent offenders, consecutively admitted for forensic psychiatric investigations between 1998 and 2001 (baseline). Methods: Psychiatric and psychological data collection at baseline included age at onset of criminal behaviour and substance abuse and measures of platelet MAO-B activity. Known criminological and clinical risk factors were registered as well as ratings with the risk assessment instruments the Psychopathy Checklist-Revised (PCL-R), Historical, Clinical, Risk Management (HCR-20), and Life History of Aggression (LHA). After a mean follow-up time of almost five years, data on violent recidivism was obtained from official crime registers and analysed in relation to the clinical and criminological risk factors and to the results of the risk instruments. Results: Twenty subjects were reconvicted for violent crimes during follow-up. The age at onset of substance abuse, but not the MAO-B activity (regardless of smoking habits), correlated with risk factors for violence and predicted criminal recidivism. Most criminological and clinical risk factors, such as age at first conviction, number of convictions, history of conduct disorder, substance abuse, and scores on the PCL-R, HCR-20, and LHA, demonstrated modest correlations with violent recidivism and moderate predictive ability with areas under the Receiver Operating Characteristics (ROC) curves between 0.72 and 0.76. Only age at first conviction and a history of substance abuse among primary relatives remained significant predictors in multivariate models. The development and use of forensic psychiatric risk assessments were analysed from a clinical point of view, considering changes over time, ethical dilemmas, and risk for integrity violations and misunderstandings due to divergent expectations and interpretations of terminology. Conclusions: Early-onset substance abuse and age at first conviction are independent risk factors for recidivistic violence in forensic psychiatry. Simple historical risk factors describing behaviour have as good predictive accuracy as complex clinical risk assessment instruments.