Abstract
The predictive validity of the HCR-20v3 has not been demonstrated in a UK forensic inpatient population. In particular, the validity of insight and positive symptoms of psychosis, two clinical risk factors within the HCR-20v3, require further evidence of their validity in this patient group. In this pseudo-prospective study of N = 167 forensic inpatients, the predictive validity of the HCR-20v3 for violence was examined over the course of a 2–12 month follow-up. The predictive validity of insight and positive symptoms were examined in a sub-sample of n = 135. Baseline data were extracted from case-files with violent incidents collected prospectively following HCR-20v3 publication. HCR-20v3 total, clinical and risk management subscale scores significantly predicted violence and physical violence specifically. For violence overall, the clinical subscale demonstrated strongest predictive validity. Post-hoc analyses indicated that even when combined with the risk management subscale, itself a significant predictor, the clinical subscale remained the strongest predictor of violence. Subgroup analyses indicated positive symptoms significantly predicted physical violence, while poor insight did not. Forensic inpatient resources should be focused on the risk assessment and management of the HCR-20v3 clinical subscale, with particular emphasis on positive symptoms which show a robust relationship with violence.
Notes
1 Please see Appendix A for all ROC plots of significant AUC values.