Objective: To evaluate diagnostic accuracy and timing of abnormal transvaginal ultrasonography for the recommendation of endometrial biopsy among premenopausal patients with abnormal uterine bleeding.
Design: Two-hundred and thirteen transvaginal ultrasonograms of premenopausal women with abnormal uterine bleeding were retrospectively evaluated.
Main outcome measures: 'Abnormal' ultrasonography was defined as either non-three-layer appearance during the proliferative phase or thickening > or =15 mm during the secretory/unknown phase. Findings were assessed on the basis of the histological diagnosis of the biopsies taken as the reference standard. Accuracy of the defined criteria with their likelihood ratio for detection of an endometrial disease was calculated.
Results: Abnormal histopathologic findings were detected in 147 (69%) of 213 patients. 'Abnormal' transvaginal ultrasonography criteria could detect disease states in 90.3% of patients with a sensitivity of 94.6% and a specificity of 77.2% (p<0.01). Sensitivity and specificity increased to 97.1 and 84.4% for ultrasonography conducted during the proliferative phase and were 60 and 61.9% for the secretory/unknown phase, respectively. Better results were clearly found for ultrasonograms obtained during the proliferative phase (p<0.01). The inclusion of post-biopsy proliferative phase transvaginal ultrasonography data of the following cycle resulted in a sensitivity of 95.2% and a specificity of 89.4%.
Conclusions: The application of transvaginal ultrasonography using non-three-layer criteria is a highly accurate first step for the selection of premenopausal patients with abnormal uterine bleeding, which may reduce the number of unnecessary endometrial biopsies by approximately 25%. Proliferative phase evaluation is preferable.