Risk Assessment with the HCR-20v3 across Genders, Subsamples, and Time-Frames

One of the most well-established violence risk assessment schemes, the HCR-20: Assessing Risk for Violence Manual, Version 2 (HCR-20v2; Historical/Clinical/Risk Management-20), has recently been revised. The present study evaluated the performance of the new HCR-20v3 in a sample of 119 participants on probation or recently discharged into the community (i.e., from incarceration or short-term psychiatric hospitalization). The HCR-20v3 demonstrated concurrent validity with the HCR-20v2 and was predictive, prospectively, of whether, and how rapidly, participants engaged in violence. The SRRs added incrementally to the presence and relevance scores. Generally, no moderation effects of subsample type were noted, with the exception of the impact of subsample on the ability of the C subscale to predict the likelihood of verbal violence, as well as its impact on the SRRs, which might more strongly predict violence for the participants who were receiving short-term psychiatric inpatient care. As pertains to gender, some moderation effects were observed at 6 weeks for violence and physical violence, but this was no longer the case at 8 months. However, the H relevance rating may be more strongly predictive of time to physical violence in women, than it is in men.Moreover, the HCR-20v3 components generally demonstrated a relationship to violent victimization, whereas they did not do so for suicide. Some ratings might exhibit a relationship to self-harm. There was no moderation effect of subsample type on the ability of HCR-20v3 to forecast violent victimization. The HCR-20v3 components were not predictive of violent victimization, suicide, or self-harm in men at 6 weeks, but some demonstrated a relationship with violent victimization at 8 months. For women, some of HCR-20v3 components were predictive of violent victimization and self-harm for both time-frames. The H presence score was more predictive of violent victimization in women. Generally, the higher the scores on all HCR-20v3 presence ratings, the sooner participants were violently victimized and there was no moderation effect of subsample type. However, there was a moderation effect of gender on the ability of the C subscale to forecast the imminence of this outcome, and this may also the case for the total presence score.