Smith & White (Reference Douglas, Ogloff and Hart2007) showed it was feasible to complete HCR-20 (Historical Clinical Risk - 20-Item Scale) ratings on most patients within 48 hours of admission to their general psychiatric wards but did not demonstrate that this approach was likely to be valid or useful.
First, HCR-20 was specifically developed for forensic patients. Furthermore, the reliability of the results in some items can be poor even for trained raters (Reference Douglas, Ogloff and HartDouglas et al, 2003) and worse for untrained ones.
The drive behind risk assessment is to identify patients who represent a significant risk of serious violence. However, the risk of a patient with schizophrenia being convicted of serious violence is 0.2% per annum (Reference MonahanWallace et al, 1998) and Monahan (Reference Monahan1981) has emphasised that ‘if the base rate [of violence]… is low then even a relatively accurate predictive test risks misclassifying many non-violent people.’
Risk assessment should be reliable, valid and result in a risk management plan, and therefore it requires careful enquiry. It is wasteful and unhelpful to assess every patient admitted. Detailed assessment should be for those a priori representing increased risk. Professionals should screen patients for previous violence and only then carry out detailed risk assessments on those who have a history of violent behaviour and those who for other reasons give concern, for instance because of violent fantasies or threats. To assist them, professionals need to know the most important predictors of violence, in order of importance: psychopathy, previous violence, and comorbid substance misuse.
The HCR-20 is an appropriate tool for forensic patients, but the MacArthur Classification of Violence Risk (COVR) is more valid for general psychiatry. This is available with a software programme with cut-off points for high and low risk, though these need to be treated with caution in the UK population.
eLetters
No eLetters have been published for this article.