The authors identified 5 of 18 items studied that clearly distinguished victims of IPV from a random group of health conference attendees with a sensitivity of 85 percent and a specificity of 87 percent. This 5-item instrument (SAFE-T) was then tested on 435 women presenting to 3 emergency departments and the results compared to a direct question regarding current abuse. The SAFE-T questions detected only 54 percent of the women who admitted being abused and correctly classified 81 percent of the women who said they were not victims. The 1-year prevalence of IPV in this sample of women presenting to an emergency department was 11.6 percent. The authors conclude that indirect questioning of women appears to be more effective at ruling out IPV in an emergency department population and may be less useful for women "early" in an abusive relationship. Tables and references (Published Abstract)