Factors affecting the validity of a violence risk screening tool with psychiatric inpatients

Statement of Problem. Violence risk screening in mental health settings is essential for violence prevention, yet current screening methods are inadequate. The Fordham Risk Screening Tool (FRST) is a promising violence screen that can help clinicians more accurately and efficiently screen for violence in mental health settings, thereby reducing violence.

Aim: This study aims to test how demographic, clinical, and contextual variables impact the accuracy of the FRST as a violence risk screening tool. Based on this data, clinicians will know in what situations to use the FRST, and the developers will be able to improve this instrument.

Subjects. The subjects will be all patients admitted to the adult inpatient psychiatry ward at Bronx Lebanon Hospital, New York. An estimated 150 patients will be included. The patient population is mostly Hispanic and/or African American, low-income, and seriously mentally ill (e.g. schizophrenia, bipolar disorder, depression, and co-occurring substance dependence.).

Method. This prospective study will use data from clinical records obtained in the context of a new clinical initiative that began in March 2014 at Bronx Lebanon Hospital. As a part of this initiative, each patient admitted to the adult psychiatry inpatient ward is screened with the FRST (a violence risk screening tool), then thoroughly assessed with the HCR-20 V3 (a validated violence risk assessment instrument).

Hypotheses: We hypothesize that female gender, younger age, higher education, severe symptoms, and homelessness prior to admission will lower the accuracy of the FRST. Schizophrenia, “threat control override” symptoms, referral by the police, and involuntary admission are hypothesized to increase the accuracy of the FRST.

Analysis: Each hypothesis will be tested with logistic and multiple regression, using the FRST screening as predictor, the patient variables as moderators, and the HCR-20 V3 risk assessment results as the outcome variable. Effect sizes will be reported. Inter-rater reliability data will be examined and reported.

Data: This study will include demographic, clinical, contextual, violence risk screening and violence risk assessment data on approximately 150 patients. Data will be entered from hospital files into a password-protected research spreadsheet containing no identifying information. The hospital files and password-protected patient ID spreadsheet will not leave Bronx Lebanon Hospital.

Publications and archiving: Based on the study results, the Principal Investigator will complete her doctoral dissertation, publish the findings in peer-reviewed journals, present at conferences, and develop training materials for the FRST. Data will be kept and disposed of according to IRB rules and regulations.

This project contains a research and/or development component, as defined in applicable law.

ca/ncf