Clinicians need to look carefully for clues to a change in a patient’s psychological or emotional status that could herald a decline potentially associated with danger. A penetrating and informed inquiry in patients deemed to be at risk is needed instead of relying on immediate situational threat appraisals used by police officers, personnel who work in emergency settings, and others.5 How to deal with such “hands-on” crises is a topic in itself.
Principle 3. Risk assessments should be carried out under circumstances that are comfortable to both client and assessor (who must ensure the safety and security of all involved in the process).
Evaluations that are conducted in busy hallways or holding cells are handicapped from the start. There needs to be some realistic chance of establishing rapport.6 Securing some measure of rapport between the examiner and evaluee and conducting the evaluation in an environment that is conducive to the discussion will greatly enhance the sharpness and overall effectiveness of the inquiry.
Principle 4. Assessors should have a working familiarity with the literature on violence risk assessment and management as well as its limitations.
Much of the scientific and professional literature on violence risk assessment and management has been summarized in recent books, and there is an easy-to-find compilation of recent key articles.7-9 Although perhaps a little technical, Rethinking Risk Assessment: The MacArthur Study of Mental Disorder and Violence10 describes the largest North American study of released civil patients. It also provides the underpinnings for the Classification of Violence Risk,11 which was an attempt by the authors to create a computer analysis model for assessing an individual’s risk for violence. It is also useful for clinicians to acquaint themselves with the time-honored work of Hervey Cleckley,12 especially as formulated by Robert Hare (Psychopathy Checklist–Revised [PCL-R])13 and Stephen Hart and colleagues (Psychopathy Checklist: Screening Version [PCL:SV]).14
Principle 5. A thorough history is essential for the completion of risk assessments. The best predictor of future violence is past violence.
While Principle 5 is true, in recent years the Hare Psychopathy Checklist (which places strong weighting on violence history) has outperformed sheer violence history as a predictor.15 However, statistically based studies have emphasized repeatedly that active risk factors during childhood and adolescence tend to continue into late adolescence and adulthood.16 A cue to remember before conducting a thorough risk assessment is depth prevents death.
Principle 6. Although actuarial information is not often available to psychiatrists who practice outside mental hospitals and prisons, it should not be ignored if it has been properly consolidated.
One scale with an appreciable correlation between predictions and violent outcomes is the Violence Risk Appraisal Guide.17 Because of the established correlation, it is unwise to discount such statistically based risk appraisals if they have been or can be compiled—provided that the case corresponds with the essential characteristics of the standardization sample (ie, age, sex, previous violence, psychiatric history, etc). Evaluators will be interested to see whether their own analyses accord with the results of actuarial assessments. More generally, if the results do not match up, clinicians should be at pains to make sense of the discrepancy. Bear in mind that risk assessments are invariably about individuals. Incidents based on the performance of groups can inform the individual assessment, but they also have the capacity to obfuscate a decision when the consequences for the evaluee are or can be dire.18
Principle 7. Structured professional judgment (SPJ) scales may assist in the assessment task.
In 1995, the Historical/Clinical/ Risk Management–20 (HCR-20, V1)19 scale was published after extensive consultation with psychiatrists, social workers, correctional officers, psychologists, emergency department staff, and other colleagues. The HCR-20 uses a 0 (not present, or “no”), 1 (possibly present, or “maybe”), or 2 (definitely present, or “yes”) scoring scheme. What sets this scale apart from others is the division of items into past (10 items), present (5 items), and future (5 items).
The HCR-20 was reviewed positively for its potential by Borum.20 The review excited colleagues abroad who wanted to translate the HCR-20. However, Version 1 had some pretty obvious faults. One such failing was a too-strong reliance on some DSM-diagnostic categories for a few items. Version 2 was developed and the reworked scale was published in 1997.21
The Table includes items from the current scheme. Note, however, that this is not a substitute for reading the entire manual carefully before attempting to use the guide. An up-to-date summary of the evidence for HCR-20 and its progeny can be located at http://kdouglas.wordpress.com/hcr-20. This topic was addressed in detail in 2 recent books and in a review article.7,8,22 The HCR-20 outperforms the PCL-R (or the 12-item PCL:SV) in prediction-outcome correlations. Although predictive power is never as high as might be wished, the HCR-20 improves substantially against chance.15
There are other important items about the HCR-20 and similar scales. After the assessment, evaluators are asked to judge whether the potential for future violence risk against others is low, moderate, or high. The HCR-20 manual stresses that occasionally an individual will achieve a very low total numerical score, yet the assessor may still decide that the patient’s case is a high risk. Similarly, a case with a high score may not necessarily warrant a high-risk designation (eg, because the eventual untoward outcome is not expected to be particularly serious or because the risk is easily managed). In other words, unlike strict actuarial approaches, the clinician is left with the final say (presuming that the HCR-20 has been administered as intended).23