Radiologic Findings in Intimate Partner Violence.

Purpose To assess the radiologic findings associated with intimate partner violence (IPV). Materials and Methods Electronic medical records of 185 patients referred to the IPV support program from the emergency department (ED) between January 2015 and October 2016 were retrospectively reviewed and compared with an age- and sex-matched control group of 555 subjects (1:3 ratio of IPV victims to control subjects) who presented to the ED. Reports of all imaging studies performed within 5 years prior to the index ED visit were reviewed. Results The majority of patients who experienced IPV (mean age, 34.2 years ± 12.2 [standard deviation]) were female (178 of 185 [96.2%]) and were largely African American (69 of 185 [37.3%]). Demographic and clinical variables independently associated with IPV were race (odds ratio [OR] range, 3.2-5.9; 95% confidence interval [CI]: 1.8, 12.7), psychiatric comorbidities (OR, 5.4; 95% CI: 3.4, 8.8), and homelessness (OR, 13.0; 95% CI: 5.4, 31.2). IPV victims underwent more imaging studies in the preceding 5 years (median, four studies) than did control subjects (median, one study). Obstetric-gynecologic findings (OR, 4.4; 95% CI: 2.1, 9.6) and acute fractures (OR, 2.4, 95% CI: 1.1, 5.3) seen on images were independently associated with IPV. The addition of imaging findings to demographic and clinical variables increased the area under the receiver operating characteristic curve (AUC) of the multivariate model to detect IPV (0.87 vs 0.86, P < .01), and the cross-validated multivariate model had an AUC of 0.85. Acute fractures involved the face or skull (range, P < .01 to P = .05), and chronic fractures affected the extremities and nasal bone (P < .01 and P = .05, respectively) more frequently in the IPV group than in the control group. Conclusion Intimate partner violence victims undergo more imaging studies and have a higher frequency of potential violence-related imaging findings when compared with age- and sex-matched control subjects. © RSNA, 2019 See also the editorial by Flores and Narayan in this issue.