Assessing Risk Factors for Victims of Violence in a Hospital-Based Violence Intervention Program

Introduction: Personal, behavioral, and environmental risk factors are correlated to varying degrees with each other and with the overall likelihood of violent reinjury. When used with fidelity, risk assessment instruments, including the violence reinjury risk assessment instrument (VRRAI), identify domains in which individuals present elevated risk levels to aid in matching services with needs. Less is known about the collinearity among risk factors for violently injured individuals admitted to hospitals. Collinearity between risk factors has ramifications for predictive modeling of violent reinjury risk. The objective of this study was to identify significantly correlated risk factors when the VRRAI was used by hospital-based violence intervention programs (HVIP) for clients. Materials and Methods: Victims of violent injury by modality of firearm, stabbing, and physical assault who were admitted to a level 1 trauma center at a single institution were voluntarily enrolled in a hospital-based violence intervention program (HVIP) between September 2020 and June 2022. Violence intervention specialists (VIS) completed the VRRAI within the first month of participant enrollment. The VRRAI is comprised of 29 binary indicators that may signal elevated risk of violent reinjury. Data from completed assessments were used to apply risk-need-responsivity (RNR) principles along with phi coefficients of key indicators to examine overlap and prevalence in the population. Results: A total of 98 participants were enrolled in the HVIP. The median age was 27 years old and 79 (80.6%) were male, while 66 (67.3%) were non-Hispanic Black or African American, 9 (9.2%) were non-Hispanic White, and 17 (17.3%) identified as Hispanic or Latino. Several statistically significant relationships existed between key risk indicators in the VRRAI. Importantly, a robust relationship was found between the two dynamic risk factors of having heavy connection with gangs and a perception of imminent threat of violence (φc = 0.57, p < 0.01). Conclusion: Data suggest that some variables could be consolidated or removed from the VRRAI to create an even shorter instrument that can be performed more rapidly in the clinical setting. The application of the RNR model illustrates a limited number of dynamic risk factors that could be immediately addressed as part of case management should be prioritized among the questions selected from the VRRAI for inquiry at the intake assessment.