Predicting violence in a medium secure setting: a study using the historical and clinical scales of the HCR‐20 Available to Purchase

This study examined the ability of the HCR‐20 Scale (version 2) to predict incidents of in‐patient violence during a stay on a medium secure ward. The study was carried out retrospectively on a sample of 94 admissions, using reports pre‐existing at the time of admission and nursing observations of behaviour in the two weeks that followed admission. The sample group was made up of mentally disordered offenders and patients with challenging behaviour. Their episodes of care averaged two years.The H and C scores, singly or combined, did not show powerful predictive capacity for the sample as a whole, which had a high base rate for violence. Insufficient variance and poor inter‐rater reliability precluded use of the R variable. For purposes of analysis, violent patients were also grouped on the number of incidents committed during their episode of care. Here the C Scale items showed strong predictive capacity for patients with more than 10 incidents during their stay. The paper argues that the latter finding supports a method of predicting admissions at risk of frequent incidents of violence.