Violence among offenders with mental disorders

Abstract

Background: Interpersonal violence is described as a global health problem by the World Health Organization. Even though individuals with severe mental disorders commit only a fraction of all violent acts, they appear to be more likely to behave violently than the general population. Assessments of violence risk may help health care services to better understand how to reduce the likelihood of violent outcomes; consequently, adequate assessment methods are crucial. On the other hand, violent individuals with severe mental disorders constitute a vulnerable group in society, so it is conceivable that they are also subjected to violence themselves. Again, health care services may be made conducive to protection.

Method: The violence risk assessment instruments COVR, LSI-R, HCR-20V3, and SAPROF were applied to a forensic psychiatric sample consisting of 200 detainees undergoing a forensic psychiatric evaluation; after a year, the sample was followed up regarding violence. Rates of self-reported violent victimization and health care utilization were compared between the forensic psychiatric sample and a general population sample consisting of 600 controls matched by age span, sex, and occupation. Finally, rates of violent ideation were examined in the forensic psychiatric sample and a general psychiatric sample consisting of 390 patients at discharge; after 20 weeks, the samples were followed up regarding violence.

Results: In the forensic psychiatric sample, 23.3% committed violent acts in the follow-up year. Correlations between the risk instruments under study were considerable, while the predictive performance was small for COVR (AUC = 0.61), medium for LSI-R (AUC = 0.70), large for HCR-20V3 (AUC = 0.79), and large for SAPROF (AUC = 0.78). Violent victimization was reported by 52.3% of the forensic psychiatric sample and 11.1% of the general population sample (RR = 8.2), health care utilization by 47.7% and 23.7%, respectively (RR = 2.0), and unmet health care needs by 42.2% and 16.7%, respectively (RR = 3.4); there was no distinct association between violent victimization and health care utilization. Violent ideation during lifetime was reported by 32.5% of the forensic psychiatric sample and 35.7% of the general psychiatric sample; when both samples were combined, there was an association between violent ideation and ensuing violent acts (OR = 2.42), but other performance measures indicated a poorer predictive ability.

Conclusions: Both violent perpetration and violent victimization are common in a forensic psychiatric context. In this context, COVR, LSI-R, and especially HCR-20V3 and SAPROF, are methods suitable for predicting violent perpetration, which may in turn facilitate prevention. Health care services should actively improve availability and take measures to protect forensic psychiatric populations from violent victimization. Violent ideation seems to be equally prevalent in forensic psychiatric and general psychiatric populations. When taking a medical history, mental health care professionals should address not only violent ideation, but also other factors that may be associated with violent outcomes. To improve performance of both assessments and interventions, the needs and characteristics of the individual must also be considered.