Predictive validity of a screen for partner violence against women

Background: While public health leaders recommend screening for partner violence, the predictive value of this practice is unknown. The purpose of this study was to test the ability of a brief three-question violence screen to predict violence against women in the ensuing months. Methods: We conducted a prospective cohort study of adult women participating in the Colorado Behavioral Risk Factor Surveillance System (BRFSS), a population-based, random-digit-dialing telephone survey. During 8 monthly cohorts, 695 women participated in the BRFSS; 409 women participated in follow-up telephone interviews approximately 4 months later. Violent events during the follow-up period, measured using a modified 28-item Conflict Tactics Scale, were compared between women who initially screened positive and those who screened negative. Results: Among BRFSS respondents, 8.4% (95% confidence interval [CI]=6.3%-10.5%) had an initial positive screen. During the follow-up period, women who screened positive were 46.5 times (5.4-405) more likely to experience severe physical violence, 11.7 times (5.0-27.3) more likely to experience physical violence, 3.6 (2.4-5.2) times more likely to experience verbal aggression, and 2.5 times (1.2-5.1) more likely to experience sexual coercion. In a multivariate model, separation from one's spouse and a positive screen were significant independent predictors of physical violence. Conclusion: A brief violence screen identifies a subset of women at high risk for verbal, physical, and sexual partner abuse over the following 4 months. Women with a positive screen who are separated from their spouse are at highest risk.