Having a crystal ball or better yet a screening tool that would predict who is going to commit teen dating violence before it happens would be a good thing, right? Designing such a tool certainly seemed worth pursuing for Cohen et al. (10.1542/peds.2017-2790) who set out to determine what risk factors uniquely are associated with an increased risk for physical and sexual forms of teen dating violence perpetration and whether the significance of these factors can lead to a predictive model for future perpetrators to be more readily identified. The authors surveyed a diverse population of over a thousand high school students and culled from that survey a series of risk factors that enabled them to generate a screening algorithm to potentially identify future perpetrators as well as non-offenders. The authors, for example, found that exposure to domestic violence, maltreatment during childhood, and having deficits in conflict resolution were strong predictors of physical violence perpetration, whereas having been a victim of sexual violence, emotional violence, or having an attitude accepting of dating violence were associated with sexual violence perpetration. When they applied this algorithm to their database, they were able to twice as likely predict such perpetrators over the course of the six years studied by the investigators. Clearly a prospective trial of this algorithm outside of the source database of students is needed to determine its validity and reliability.
But while this algorithm sounds promising in that some of the risk factors identified, such as difficulty with conflict resolution, would lend themselves to treatment and in turn a reduction in the likelihood of dating violence, there are risks to doing this as well. Drs. Idia Thurston and Kathryn Howell, psychologists from the University of Memphis, have helped us think about the risks of having a predictive algorithm in an accompanying commentary (10.1542/peds.2018-0075). One such risk would be labeling teens as prone to becoming violent when they may never demonstrate that violent dating behavior at all despite their being at risk. Check out both this study and commentary and think about whether or not you discuss these risk factors with your patients who demonstrate them. Even if you don’t, at least you will be made more aware of who might require closer observation and perhaps more education about dating violence because of their predisposition to becoming a perpetrator of such violence.
