Screening for risk of inpatient violence: Validation of an actuarial tool.

Abstract

A screening checklist for assessing the risk of violence at the time of psychiatric hospitalization was developed. Checklist items and a cutoff point for low versus high risk of violence were selected based on a previous study of statistical prediction of violence. Application of the checklist to a new sample of 338 patients admitted to a university-based inpatient unit revealed promising results in identifying patients who later displayed aggressive behavior in the hospital, as indicated by its positive predictive value (59.0%), negative predictive value (70.6%), sensitivity (57.2%), specificity (70.0%), total predictive value (65.4%), and likelihood ratio (1.97). The results suggest the potential value of incorporating actuarial methods in the evaluation of violence risk.Although evaluations of violence potential are frequent components of clinical practice in emergency and inpatient psychiatric settings, the limitations of clinicians' abilities in assessing violence potential are well known (Monahan, 1981). Several authors have suggested that evaluations of violence potential could be

The predictive validity of risk assessment tools for female offenders: A systematic review

“…They are empirically guided, in that they are based on factors empirically demonstrated to be associated with risk, but judgements are also clinically informed (Hart, 1998b). Examples of SPJ tools include the HCR-20 (Webster, Douglas, Eaves, & Hart, 1997), Level of Service Inventory (LSI-R; Andrews & Bonta, 1995) and the Violence Risk Scale (VRS; McNiel & Binder, 1994). All of these instruments have demonstrated good predictive validity with AUC N .70 (Douglas, Ogloff, Nicholls, & Grant, 1999;Gray, Taylor, & Snowden, 2008) and correlation values N .50 (Gray et al, 2003).…”

Assessing and managing violence risk in outpatient settings

“…Although there are a few actuarial formulae developed for assessing violence risk, e.g., Violence Risk Appraisal Guide (Quinsey, Harris, Rice, & Cormier, 1998) and Violence Prediction Scheme (Webster, Harris, Rice, Cormier, & Quinsey, 1994), they are most appropriately used with institutional populations who have criminal histories, and their utility with nonforensic, noncorrectional mental health populations in acute care settings is unknown. McNiel and Binder (1994) developed a five-item actuarial formula (history of threats or physical attacks within two weeks of admission, absence of suicidal behavior, diagnosis of schizophrenia or mania, male sex, and married to or living with significant other), which was designed to predict violent behavior during the course of acute hospitalization. They reported moderate sensitivity (55%) and specificity (64%), where physical attacks or fear-2 Because most actuarial formulas include primarily static variables that appear relatively easy to code, it is typically assumed that inter-rater reliability for the scoring of these instruments will be high.…”

The recalcitrance of overconfidence and its contribution to decision aid neglect

“…Thanks to the advent of such decision aids, many predictions, diagnoses, and forecasts that had previously been difficult have been rendered less so. Domains in which the beneficial effects of decision aids have been demonstrated include important medical conditions (e.g., Alvarado, 1986;Corey & Merenstein, 1987), the prediction of dangerousness (e.g., McNiel & Binder, 1994), and financial auditing (e.g., Ashton, 1991).…”