Using Radiology as a Screening Tool to Identify Intimate Partner Violence.

Abstract

To identify imaging utilization patterns and radiologically evident injuries suggestive of intimate partner violence (IPV). This retrospective case-control study analyzed historical imaging reports from women reporting physical IPV between 2013 and 2018 ("cases"; n = 265; 6,401 imaging studies) and age- and race-matched controls not reporting IPV (n = 875; 13,462 imaging studies). Adjusted incidence rate ratios (aIRRs) were calculated using Poisson regression to compare imaging utilization by organ system or modality. Logistic regression calculated adjusted odds ratios (aORs) for canceled studies and studies conducted in the emergency department (ED), overnight, or on weekends. Injury patterns were assessed by calculating aIRRs of injury by anatomic site and comparing synchronous and asynchronous injury detection. A time-to-event analysis using Cox proportional hazards regression compared the cumulative incidence of asynchronous injury in patients with at least one injury. Patients experiencing IPV underwent imaging at a higher rate than controls (aIRR: 1.5, 95 confidence interval [CI]: 1.5-1.6). They were more likely to cancel screening mammograms (aOR: 3.2, 95% CI: 1.8-5.7) and undergo imaging in the ED (aOR: 3.8, 95% CI: 2.9-5.0), overnight (aOR: 1.6, 95% CI: 1.3-1.9) and on weekends (aOR: 1.3, 95% CI: 1.1-1.5). The injury rate was higher in cases (aIRR: 15.1, 95% CI: 12.5-18.1), particularly for cranial (aIRR: 104.5, 95% CI: 14.2-770.1), facial (aIRR: 47.0, 95% CI: 18.9-117.0), and thoracic (aIRR: 32.6, 95% CI: 11.5-92.2) injuries. Cases were more likely to have multiple injuries in a single encounter (aOR: 4.7, 95% CI: 1.5-14.1) or multiple encounters with unique injuries (hazard ratio: 4.9, 95% CI: 2.6-9.3). Patients experiencing IPV had higher rates of imaging studies, imaging utilization in the ED, weekend, and overnight settings, screening examination cancellation, and radiologically evident injuries than controls.