Long-term predictive validity of historical factors in two risk assessment instruments in a group of violent offenders with schizophrenia

The long-term predictive validity of the Violent Risk Appraisal Guide (VRAG) and the historical part (H-10) of the risk assessment device HCR-20 in predicting violent recidivism was investigated in a sample of (n =106) violent offenders with schizophrenia. An effort was made to validate the 9-bin categorization of different absolute risk to recidivate depending on the individual score on the VRAG. Scores on both devices were retrospectively obtained from various files and registers. Individuals were followed up after discharge from hospital for on average 86 (standard deviation=19.33) months. During follow-up 29% of the sample was reconvicted of a violent crime. Results indicated that both H-10 and VRAG had a moderate ability to predict violent recidivism and that H-10 had a slightly better accuracy. Most of the items in H-10 but only half of those in VRAG correlated significantly with violent recidivism. The 9-bin categorization of VRAG scores produced mixed results. In the Swedish sample there was a linear trend in which increased VRAG scores were associated with higher absolute risk to recidivate. However, the distribution of scores and the figures of absolute risk of recidivation were not replicated. It is concluded that historical factors seem to play an important role for the long-term prediction of future violence among a group of severely mentally ill individuals.