Violence Risk Assessment: Part I

When performing a violence risk assessment, there are competing viewpoints as to which variables should be considered. Such variables are generally integrated into a theoretical model of the patient’s personality and behavior to either stratify the likelihood of becoming violent or inform treatment decisions. In the past, psychiatrists relied solely on unstructured “clinical” judgment, which was harshly described by Ennis and Litwack as having “absolutely no expertise in predicting dangerous behavior” and further marginalized by Lidz and colleagues in a case-controlled study. The introduction of actuarial methods (statistically oriented structured risk assessment), which identify and weigh various factors to minimize error-prone clinician subjectivity, offers promise but remains underutilized by psychiatrists. This may be defensive posturing in response to state ments such as “Actuarial methods are too good and clinical judgment is too poor to risk contaminating the former with the latter.” 3 In 1999, one-third of US psychiatric residents stated that they had no training in violence risk assessment, whereas another third described their training as “inadequate.” This was 4 years after the development of the Historical, Clinical, Risk Management 20-item (HCR-20) violence risk assessment scale, 6 years after the publication of the Violence Risk Appraisal Guide (VRAG), and nearly 20 years after publication of the forerunner of the Psychopathy Checklist. Currently, there is no American Psychiatric Association practice guideline for violence risk assessment that resembles the established guideline for suicide risk assessment and treatment. Combining elements from the mental status examination and an actuarial instrument is referred to as structured professional judgment. Miller described this as identifying historical risk factors that characterize the context of an individual’s aggressive behavior and risk factors After participating in this activity, the psychiatrist should be better able to: