Violent behavior and violent victimization among general psychiatric patients: prevalence and prediction

Sammanfattning: The intriguing question of how mental disorder and violence relate to each other has become an epic academic debate. During the last decades, there has been a change in direction of the debate on individuals with mental disorder, with a greater focus on violent victimization than violent behaviour towards others. Up until now, no Swedish study has investigated the frequency of violent behaviour among general psychiatric patients undergoing psychiatric treatment. Moreover, no Swedish study so far has investigated the relative risk of victimization in general psychiatric patients, in comparison to the general population. The aim of this dissertation was to investigate these issues and to validate the risk assessment method Classification of Violence Risk (COVR)™. Method: In study I, general psychiatric patients were recruited from two public psychiatric hospitals in Stockholm County (n=390). The control group consisted of gender- and age-matched subjects recruited from an annual national survey of living conditions, (conducted by Statistics Sweden) (n=1170). Studies II-IV consisted of prospective follow-ups on 331 patients. At baseline, clinical and socio-demographic variables were collected and a COVR assessment was conducted. Follow-up included telephone interviews with the patients and collaterals 10 and 20 weeks after baseline. Violent behaviour was self-reported and in addition, data was collected from a national criminal register. Results: Twenty percent of the patients had been victimised during the year preceding inclusion. The relative rate of victimization was six times higher in patients compared to controls. Women appeared to be most vulnerable with a 10-fold risk increase (Study I). The base rate of violent behaviour was 5.7% and a receiver operating curve analysis (ROC) showed that the area under the curve (AUC) for COVR was 0.77. The gender gap concerning violent behaviour among the general population was not replicated, since there was no significant gender difference with respect to violent acts 20 weeks after discharge. The predictive validity of the COVR software was comparable between females and males. There was an overlap between offenders and victims among psychiatric patients (Studies II-IV). Conclusions: The risk of being subjected to violence is high among Swedish psychiatric patients. The findings are most pronounced for female patients. Research, clinicians and social policy should target the problem of victimization. The base rate of violent behaviour towards others is relatively low among general psychiatric patients in Sweden. Therefore, prediction is difficult. Violent behaviour was uncommon in female as well as male patients and there were no gender differences. The COVR software could significantly predict violent behaviour and its validity was comparable to other risk assessment tools. COVR predicted violent behaviour with the same precision in both genders. The overlap between offenders and victims should be taken into account in both research and clinical settings.