Violence prediction in the Emergency Department.

The purpose of this study was to examine whether the SAGE assessment survey could predict, within the Emergency Department setting, those youth at risk for engaging in violent behavior. It also examined whether those who test positive for engaging in high-risk violent behavior during the initial baseline SAGE survey were prone to continue this behavior 1 month later. This was an observation, convenience sample of young male and female patients, half of whom had injuries related to violence and half of whom had injuries unrelated to violence, who presented when a research fellow was available. They were given the SAGE aggression assessment survey and questioned about their risk behavior in the past 6 months to 1 year. Those youths were contacted 1 month after enrollment to determine the incidence of continued at-risk behavior. Demographic information was used to contact the patient on follow-up but not for identification purposes. The study was IRB-approved. The setting was a Level I pediatric and adult trauma center in an inner city with 45,000 annual visits. The inclusion criteria were: age 10-24 years, consenting patient or guardian, medically stable, and able to communicate. The exclusion criteria were: those youths who were uncooperative or refused to participate. The comparison between the SAGE overall scores of the 182 subjects with 46% who had injuries related to violence vs. 54% who presented with injuries not related to violence did not show a significant difference. Those who presented with non-violent injuries had an overall score range of 7-8 with a mean of 7.4 and those who presented with violent injuries had an overall scale of 10-11 with a mean of 10.1 out of a possible 12 total. There were, however, significant differences at the .05 or less level when looking at specific questions asking about certain behaviors such as physical fighting, shoving, needing medical attention, and kicking, in the initial survey. At the 1-month follow-up, with 118 subjects, the SAGE tool also showed differences in the areas of physical fighting, shoving, and kicking between the 56% who had injuries not related to violence vs. the 44% who did have violence-related injuries. It identified 18-50% of those who presented with violence-related injuries and who continued to engage in high-risk behavior at the 1-month follow-up. These results indicate that the SAGE survey scores based on the responses to all 12 questions is not effective in identifying those youths at risk for general violent behavior, especially with a high-risk population. A set of four questions from the 12-question SAGE survey, however, was successful at tracking specific types of at-risk behaviors. This indicates that certain questions within the SAGE survey may be effective at tracking those who engage in high-risk violent behavior. These same questions were able to track those who continued to engage in high-risk violent behaviors 1 month later.