Evaluating the Predictive Validity of the Fordham Risk Screening Tool (FRST) for Violent Behavior during Inpatient Psychiatric Hospitalization

Christopher W Racine; Tania D Strout; Douglas N Johnston; Kerry M Quigley; Lee A Wolfrum; Ben J Guido

doi:10.1007/s10597-025-01562-w

Evaluating the Predictive Validity of the Fordham Risk Screening Tool (FRST) for Violent Behavior during Inpatient Psychiatric Hospitalization

Community Ment Health J. 2025 Nov 14. doi: 10.1007/s10597-025-01562-w. Online ahead of print.

Authors

Christopher W Racine^{1

2}, Tania D Strout³, Douglas N Johnston⁴, Kerry M Quigley⁵, Lee A Wolfrum⁶, Ben J Guido⁶

Affiliations

¹ Department of Psychiatry, Tufts University School of Medicine, Boston, MA, USA. Christopher.racine@chcc.health.
² Department of Psychiatry, Commonwealth Healthcare Corporation, PMB 525, PO Box 10001, 96950, Saipan, MP, USA. Christopher.racine@chcc.health.
³ Department of Emergency Medicine, Maine Medical Center, Portland, ME, USA.
⁴ Department of Psychiatry, Maine Medical Center, Portland, ME, USA.
⁵ Tufts University School of Medicine, Boston, MA, USA.
⁶ MaineHealth Maine Medical Center-Portland Department of Psychiatry, Portland, ME, USA.

PMID: 41236710
DOI: 10.1007/s10597-025-01562-w

Abstract

Prior research has shown the Fordham Risk Screening Tool (FRST) to be accurate in assessment of violence risk when compared to more comprehensive risk assessment instruments. However, a link between any violence risk screening tool and actual markers of violent behavior is absent in the literature. This study, therefore, sought to determine whether the FRST could be used to accurately assess the likelihood of violence, and markers of violence, during an inpatient psychiatric admission. This retrospective health records survey examined data from a consecutive cohort of adults (N = 423) admitted to an inpatient psychiatric unit from the emergency department (ED) from September 1, 2022, through June 30, 2023. Both electronic and manual abstraction strategies were used to evaluate FRST screening tool results collected in the ED and markers of violent behavior during subsequent inpatient hospitalization. When comparing the proportion of subjects who experienced a violent incident while hospitalized in the FRST positive and FRST negative groups, a significant difference was noted with a greater proportion of those with a positive FRST screening experiencing a violent event (33.0% vs. 8.1%, for the positive and negative screening groups, respectively, χ² = 41.046, df = 1, p < 0.001). Sensitivity and specificity were 33.0% (95% CI: 24.7%-42.5%) and 91.9% (95% CI: 88.1%-94.6%), respectively. The positive predictive value was 60.3% (95% CI: 47.2%-72.2%) and the negative predictive value was 78.6% (95% CI: 73.9%-82.7%). The area under the receiver operating characteristic curve (AUROC) for the predictive ability of the FRST was 0.305 (standard error [SE] 0.038), 95% CI: 0.230-0.381). Overall, the FRST instrument showed mixed results as a screening tool to detect the potential of violent behavior in admitted psychiatric inpatient adults. While there was a significant difference in violent events between FRST positive and negative groups, metrics assessing predictive validity and reliability of the FRST were limited. Utilization of a screening tool such as the FRST should continue to be paired with additional efforts to evaluate inpatient risk of violent behavior.

Keywords: Psychiatric hospitalization; Risk assessment; Screening tool; Violence.