The Classification of Violence Risk (COVR) is an interactive software program designed to estimate the risk that an acute psychiatric patient will be violent to others over the next several months. Using a laptop or a desktop computer, COVR guides the evaluator through a brief chart review and a 10-minute interview with the patient. COVR generates a report that places the patient’s violence risk in one of five categories—ranging from a 1% likelihood of violence in the first category to a 76% likelihood of violence in the highest category, including the confidence interval for the given risk estimate.
The software was constructed from data generated in the MacArthur Violence Risk Assessment Study. In brief, more than 1,000 patients in acute civil psychiatric facilities were assessed on 134 potential risk factors for violent behavior. Patients were followed for 20 weeks in the community after discharge from the hospital, and their violence to others was assessed. The software is capable of assessing the 40 risk factors for violence that emerged as most predictive of violence in the MacArthur Violence Risk Assessment Study, but in any given case, it assesses only those risk factors necessary to classify the patient’s violence risk.
To combine risk factors into a preliminary estimate of risk, the COVR relies on “classification tree” methodology. This approach allows many different combinations of risk factors to classify a person as high or low risk. Based on a sequence established by the classification tree, a first question is asked of all persons being assessed. Contingent on the answer to that question, one or other second question is posed, and so on. The classification tree process is repeated until each person is classified into a final risk category. This “interaction” model contrasts with the more typical “main effects” approach to structured risk assessment, such as the one used by the Violence Risk Appraisal Guide, in which a common set of questions is asked of everyone being assessed, and every answer is weighted and summed to produce a score that can be used for the purpose of obtaining an overall estimate of risk.
The authors of the COVR administered the newly developed software to independent samples of acute civil inpatients at two sites. Patients classified by the software as high or low risk for violence were followed in the community for 20 weeks after discharge. Expected rates of violence in the low- and high-risk groups were 1% and 64%, respectively. Observed rates of violence in the low- and high-risk groups were 9% and 35%, respectively, when a strict definition of violence was used and 9% and 49%, respectively, when a slightly more inclusive definition of violence was used. These results indicated that software incorporating the multiple iterative classification tree models may be helpful to clinicians who are faced with making decisions about discharge planning for acute civil inpatients.
In the view of its authors, the COVR software is useful in informing, but not in replacing, clinical decision making regarding risk assessment. The authors recommend a two-phased violence risk assessment procedure, in which a patient is first administered the COVR and then the preliminary risk estimate generated by the COVR is reviewed by the clinician ultimately responsible for making the risk assessment in the context of additional information believed to be relevant and gathered from clinical interviews, significant others, and/or available records. Although clinical review would not revise or “adjust” the structured risk estimate produced by the COVR, and could in principle either improve or lessen predictive accuracy as compared with relying solely on an unreviewed COVR score, the authors of the COVR believed it essential to allow for such a review, for two reasons. The first reason has to do with possible limits on the generalizability of the validity of the software. For example, is the predictive validity of the COVR generalizable to Native Americans, to forensic patients, to people outside the United States, to people who are less than 18 years old, or to the emergency room assessments of persons who have not been hospitalized recently? The predictive validity of this instrument may well generalize widely. Yet there comes a point at which the sample to which a structured risk assessment instrument is applied differs so much from the sample on which the instrument was constructed and validated that legitimate questions can be raised regarding the generalizability of the validity of the instrument.
The second reason given in defense of allowing a clinician the option to review structured risk estimates is that the clinician may note the presence of rare risk or protective factors in a given case and these factors—precisely because they are rare—will not have been taken into account in the construction of the structured instrument. In the context of structured instruments for assessing violence risk, the most frequently mentioned rare risk factor is a direct threat— that is, an apparently serious statement of intention to do violence to a named victim.
- Monahan, J., Steadman, H., Appelbaum, P., Grisso, T., Mulvey, E., Roth, L., et al. (2005). The classification of violence risk. Lutz, FL: Psychological Assessment Resources.
- Monahan, J., Steadman, H., Appelbaum, P., Grisso, T., Mulvey, E., Roth, L., et al. (2007). The classification of violence risk. Behavioral Sciences and the Law, 24, 721-730.
- Monahan, J, Steadman, H., Robbins, P., Appelbaum, P., Banks, S., Grisso, T., et al. (2005). An actuarial model of violence risk assessment for persons with mental disorders. Psychiatric Services, 56, 810-815.