Electronic Letters to:
|
|
eLetters: Electronic letters published:
-
![[Read eLetter]](/icons/misc/sectionDOWN.gif)
The HCR-20 Catch-22
- Neelom Sharma (21 August 2008)
|
|
|||
|
Neelom Sharma, Trainee Psychiatrist (ST3) The State Hospital, Lampits Road, Carstairs, Lanarkshire, Tel 01555 840293.
Send letter to journal:
neelom.sharma{at}luht.scot.nhs.uk Neelom Sharma
|
Gray et al have written a thought-provoking article on efficacy of the HCR-20 violence risk assessment tool, strengthening the evidence base for using this tool actuarially to predict long-term violence risk. It is heartening that this evidence base is now emerging, given the widespread use of the HCR-20 in UK forensic mental health. However, the fundamental problem with this type of research is that an efficacy study (i.e. measuring the effect of the HCR-20 in experimental conditions) does not measure the effectiveness of a clinical tool. This study used case notes to construct HCR-20s for patients. It is therefore unsurprising that the clinical domain did not produce significant predictions, as reviewing case notes does not meaningfully replicate the clinical setting. Webster et al’s manual for using the HCR-20 states that, “for most contexts in which the HCR-20 has a foreseeable use, the person will be available for interview.” In this pseudo-prospective study, this was not the case. The HCR-20 manual adds, “In most cases, file review, interview and testing should be adequate to complete the HCR-20. However, particularly for the risk management section, consultation with colleagues responsible for treatment or community release plans will likely be needed.” Thus the validity of risk items in HCR-20s completed only by case -note review is also questionable. As those who complete the HCR-20 routinely will testify, it is typically constructed by multidisciplinary discussion among a group of professionals who are familiar with the patient. It is easy to see why Gray et al adopted this efficacy-based approach; the Catch-22 is that an effectiveness study would be ethically difficult to perform, and expose the clinical team to potential criticism. Clinically, those deemed to be at high risk of violent reoffending (i.e. those with high HCR-20 scores in this paper) require aggressive risk management, which may involve ongoing detention. Thus an effectiveness study which showed that HCR-20 had high predictive validity would suggest an ineffective clinical team, who had allowed patients to reoffend despite identifying the risk beforehand. There are other methodological difficulties in this paper. For example, it is stated that, “Cases of patients reconvicted for a non- violent offence were removed from the analysis of violent offences from the time the non-violent offence occurred, as these individuals might no longer have been at liberty to commit further offences.” While this has face validity, it is not an exclusion criterion based on facts (e.g. police or health records), but on untested hypothesis. It is also the only exclusion criterion; many other reasons for variations in violent reconviction rates are not considered (e.g. hospital readmission). This issue could have been addressed by obtaining follow-up data. In summary, HCR-20 is a useful tool, and there is a growing evidence base (including this paper) to suggest it can be used in a number of different ways, including actuarially. Unfortunately, methodological flaws and the use of a clinical tool in a non-clinical manner limit the strength of the evidence presented here. Reference: Webster CD, Douglas KS, Eaves D, Hart SD. HCR–20: Assessing Risk for Violence (Version 2). Simon Fraser University, 1997 |
|||