The Risk for Violence Screening Tool: a psychometric validation and reliability study

Background

There is a lack of high-quality evidence to support the recommendation of an instrument to screen emergency department patients for their risk for violence.

Objective

To demonstrate the content and predictive validity and reliability of the novel Risk for Violence Screening Tool to identify patients at risk for violence.

Design and Setting

This retrospective risk screening study was conducted at a 100-bed emergency department in an urban, academic, safety net trauma center in Southern California.

Participants

Nine subject matter expert employees and 22,583 non-critical adult emergency patients.

Methods

Subject matter experts rated the relevance of 30 risk items and 10 domains comprising the Risk for Violence Screening Tool. Content Validity Index/Average Content Validity and Universal Agreement analyses determined the tool's content validity. Predictive validity, analyzed with logistic regression, used data from a patient incidence of violence registry (an incident occurred or did not occur), compared with the Risk for Violence Screening Tool screening result (positive or negative). Response agreement among three nurses who independently utilized the Risk for Violence Screening Tool to assess the same patients upon their arrival at the emergency department determined inter-rater reliability.

Results

The overall Content Validity Index was 0.84, and the item-level Content Validity Index and the universal agreement scores were 1.00, indicating perfect agreement on item relevance and a unanimous consensus among experts. Patients who screened positive on the Risk of Violence Screening Tool, indicating a high probability of violence, were 27.55 times more likely to commit an act of violence compared to those who screened negative (95 % CI = 24.73 to 30.70). The logistic regression model was statistically significant, (χ²₍₁₎ = 1140.29, p < 0.0001), explaining a range from 4.9 % (Cox & Snell R²) to 37.3 % (Nagelkerke R²) of the variance and indicating that the Risk for Violence Screening Tool successfully predicted violent incidents. The calculated area under the curve of 0.843 (95 % confidence interval, 0.812–0.873, p < 0.001) affirmed the Risk for Violence Screening Tool's discriminatory ability to distinguish between individuals with and without a risk for violent behavior. Cohen's Kappa statistic yielded a value of 0.86 (p < 0.001).

Conclusions

The Risk for Violence Screening Tool is valid and reliable. This standardized tool can be used in emergency departments as part of a robust violence prevention program, creating a safer workplace for both staff and patients.

Tweetable abstract

The Risk for Violence Screening Tool (RVST) is a valid and reliable brief screener to identify emergency department patients at risk for violence. @dhs.lacounty @ijn2024