Secondary Logo

Journal Logo


Identification of Factors That Influence the Likelihood of Screening for Domestic Violence by Medical Students and Residents



Author Information
  • Free

Objective: Physicians are in a front-line position to screen patients for domestic violence (DV) yet often feel unprepared to do so. This study was done to assess the effectiveness of extant DV curricula and to inform the development of new programs that strengthen DV screening skills.

Description: We developed a new written tool, including a patient scenario, to identify factors that might enhance or impede DV screening. We varied two factors in this scenario to determine whether those factors influenced the likelihood that medical students and residents would screen for DV.

In 2000, four versions of a one-paragraph patient scenario presented a divorced woman with non-localized abdominal pain of two months' duration. The patient's age was either 22 or 45 years, and abdominal bruising was either present or not noted. The scenario was followed by items that measured the respondents' likelihood of screening the “paper patient” for DV and seven other tasks (taking sexual, social, dietary, smoking, or drinking histories, doing a pelvic exam or pregnancy test) on seven-point Likert scales. The respondents also self-assessed their competence in performing these tasks using six-point Likert scales, and they characterized instruction time spent in medical school in each area.

The instrument was completed by 78 third-year medical students after their twelfth clerkship week, and by 128 first- and second-year residents during orientation. Each respondent saw only one of the four randomly assigned scenario variations. The questionnaire took ten minutes or less to complete.

Data analyses were descriptive statistics, t-tests and stepwise regression. The predictors in the stepwise regression analysis were patient age, bruising, student/physician gender, training level (medical student versus resident), and instruction time and self-addressed competence in the target areas; the dependent variable was the likelihood of conducting DV screening.

Discussion: The presence of a bruise best predicted the likelihood of DV screening; this variable entered first in the stepwise regression (p = .000; bruise present increased likelihood of DV screening). The second variable entering the equation was self-reported competence (p = .000; increased competence increased likelihood of DV screening). The third variable was training level (p = .027; surprisingly, the residents were less likely to screen). While respondent's gender did not enter the regression equation, t-tests suggested that the women were more likely to screen for DV than were the men (p = .014). No evidence of a relationship between the patient's age and the likelihood of DV screening was found.

Perhaps not surprisingly, the presence of a bruise was the strongest predictor for DV screening. Many other patient or physician demographic attributes are also likely to influence DV screening practice, and this short, easily administered instrument could readily be modified to probe for them. The positive association between self-assessed competence and greater likelihood of DV screening emphasizes the importance of teaching and reinforcing DV screening skills. Nevertheless, DV screening falls short if only those patients with bruises or who fit preconceived stereotypes are screened. The information collected from the pilot test of this instrument will begin to enable us to track current DV screening practices among medical students and residents, to identify DV curriculum deficiencies, and ultimately to design programs that will impel physicians to screen all patients for DV.

Section Description

Annual, Peer-reviewed Collection of Reports of Innovative Approaches to Medical Education

© 2001 Association of American Medical Colleges