Ovarian torsion: Case–control study comparing the sensitivity and specificity of ultrasonography and computed tomography for diagnosis in the emergency department

Abstract

ObjectiveEvaluate the sensitivity and specificity of pelvic ultrasound (US) and abdominopelvic computed tomography (CT) for the identification of ovarian torsion in women presenting to the emergency department with acute lower abdominal or pelvic pain. Materials and methodsThis is a retrospective study of 20 cases of ovarian torsion and 20 control patients, all of whom had both US and CT performed in the emergency department. Two radiologists who were blinded to clinical data interpreted all studies as (1) demonstrating an abnormal ovary or not, and (2) suggestive of torsion or not. Sensitivity, specificity and interobserver variation were calculated for each imaging modality. ResultsPelvic US was interpreted as demonstrating an abnormal ovary in 90.0% of ovarian torsion cases by reader 1, and in 100.0% by reader 2, whereas CT was interpreted as revealing an abnormal ovary in 100.0% of torsion cases by both readers. Pelvic US for ovarian torsion was 80.0% sensitive (95% CI, 58.4–91.9%) and 95.0% specific (95% CI, 76.4–99.1%) for reader 1, while 80.0% sensitive (95% CI, 58.4–91.9%) and 85.0% specific (95% CI, 64.0–95.0%) for reader 2. Interobserver agreement for pelvic US was fair (Kappa=0.60). Abdominopelvic CT for ovarian torsion was 100.0% sensitive (95% CI, 83.9–100.0%) and 85.0% specific (95% CI, 64.0–94.5%) for reader 1, while 90.0% sensitive (95% CI, 69.9–97.2%) and 90.0% specific (95% CI, 69.9–97.2%) for reader 2. Interobserver agreement was excellent (Kappa=0.85). ConclusionThe diagnostic performance of CT is not shown to be significantly different from that of US in identifying ovarian torsion in this study. These results suggest that when CT demonstrates findings of ovarian torsion, the performance of another imaging exam (i.e. US) that delays therapy is unlikely to improve preoperative diagnostic yield.