The Predictive accuracy of static and dynamic measures for assessing risk of inpatient aggression in a secure psychiatric hospital
2017-05-31T06:07:31Z (GMT) by
Given the significant implications on public safety, the assessment of violent behaviours of people with mental illnesses has become a key aspect of clinical practice for mental health clinicians. However, the prediction of violent behaviours has been difficult. Despite the advancement of violence risk assessment knowledge and practice over the past few decades, it is sometimes difficult to ascertain which measures the clinician should use to assess and make decisions about individuals on an ongoing basis, particularly, in the short to medium term. Within this context, the aims of this study are to compare the predictive accuracy of dynamic risk assessment measures for violence with static risk assessment measures over short- and medium-term follow-up periods (up to 6 months) in a forensic psychiatric inpatient setting, as well as to determine the time frame during which they are most suited for predicting inpatient aggression in a forensic inpatient psychiatric sample. Data pertaining to the sociodemographic and offence characteristics, as well as the mental health, criminal justice, and institutional outcomes were collected for 70 patients who were housed on the acute wards of the Thomas Embling Hospital, a statewide forensic psychiatric hospital in Victoria, Australia, between June and October 2002. In addition to the prospective risk assessment data (the DASA:IV and the HCR-20 Clinical scale) that were previously collected for these participants, several risk assessment measures (the HCR-20, the LSI-R:SV, the PCL-R, the PCL:SV, the START, and the VRAG) were retrospectively coded for each of the 70 patients. Results of this study showed that: (1) dynamic measures are more accurate for predicting inpatient aggression in the very short term (1 day to 1 week) than the short term (1 month); (2) dynamic measures also were accurate for short-term to medium-term predictions of inpatient aggression; (3) static risk assessment measures were generally not accurate for predicting inpatient aggression in the short to medium term; (4) short-term averages of risk states were accurate for predicting inpatient aggression and violence in the short to medium term (i.e., 1 week to 6 months), whereas the peak scores were generally predictive of inpatient aggression at longer follow-up periods (i.e., 3 and 6 months); and (5) protective factors predicted the nonoccurrence of interpersonal violence, property, and any inpatient aggression. Despite the presence of several limitations and methodological issues, the findings of this study have provided information pertaining to the suitability of static and dynamic risk assessment measures for assessing short- and medium-term propensities for violence in the forensic inpatient context. In addition, the results of this study highlight the necessity of conducting multiple assessments of short-term risk within the forensic inpatient setting to improve the prediction of inpatient aggression, and also suggest that the short-term averages of risk states may be a suitable index for assessment and management purposes in the medium term (e.g., clinical teams can use this to review and manage aggressive patients in the hospital wards). Such knowledge can assist with the development of more accurate and efficient risk assessment procedures, so as to manage offenders with mental illnesses within the community and institutions better. Consequently, these improved assessment and management procedures can lead to better outcomes and safety for the offenders, rehabilitation staff, as well as the community.