Comparison of Sensitivity and Specificity of Structured and Narrative Reports of Transvaginal Ultrasonogaphy for Adenomyosis

Abstract

Study Objective: To compare the sensitivity and specificity of the transvaginal ultrasonography (TVUS) narrative report with those of the structured report for the diagnosis of adenomyosis. Design: Retrospective study. Setting: A tertiary teaching hospital. Patients: One hundred ninety-two patients (45 with adenomyosis and 147 controls) who underwent hysterectomy between 2012 and 2016 and were aged 30 years to 55 years, with available preoperative TVUS images, were included. Interventions: To compare preoperative TVUS with histologic analysis of the uterus after hysterectomy for the diagnostic of adenomyosis. Measurements and Main Results: Data regarding the presence of the ultrasonographic characteristics previously described for the diagnosis of adenomyosis were obtained from the original TVUS report (narrative report) and the structured report from 2 blinded radiologists (R1 and R2). Histologic analysis is defined as the gold standard for the diagnosis of adenomyosis. The mean age (45.4 +/- 5.3 years vs 44.9 +/- 4.8 years; p = .496) and clinical symptoms were similar between the adenomyosis and control groups, except for dyspareunia, which was more frequently reported in the patients with adenomyosis (25.6% vs 9.4%; p = .006). Most of TVUS characteristics were more frequently observed in the structured reports than in the narrative reports. The structured report presented higher sensitivity (R1: 84.4%, R2: 69.1%; narrative report: 31.1%; p <.05) and lower specificity (R1: 28.0%, R2: 31.2%; narrative report: 90.5%; p <.05) for adenomyosis. The structured report presented higher sensitivity and lower specificity for most of sonographic characteristics evaluated. After logistic regression analysis, no sono-graphic characteristics presented with statistical significance for the diagnosis of adenomyosis on R1's structured report, while globular uterus (odds ratio [OR] 0.276; p = .006; 95% CI, 0.11 & minus;0.697) and poorly defined junctional zone (OR 3.6; p = .007; 95% CI, 1.4 & minus;9.2) were significantly associated with adenomyosis. In the narrative report, a myometrial cyst was associated with a higher risk of adenomyosis (odds ratio 9.486, p =.002; CI, 2.359-38.149). Conclusion: The narrative reports were more specific, whereas the structured reports were more sensitive for the diagnosis of adenomyosis. In addition, the sensitivity of most of sonographic feature of adenomyosis was higher and the specificity was lower in the structured report. Future prospective studies comparing both reports are needed to validate the current find-ings.