The accuracy of transvaginal ultrasonography in the diagnosis of endometrial abnormalities

Abstract

To evaluate the diagnostic accuracy of transvaginal ultrasonography for endometrial abnormalities in women with abnormal uterine bleeding. In a prospective study, 136 consecutive patients underwent transvaginal ultrasonography, hysteroscopy, and endometrial sampling through suction curettage or directed biopsy. The ultrasonographic findings were evaluated on the basis of the final diagnosis established by hysteroscopy and histologic examination. Receiver operating characteristic curve analysis and likelihood ratios were used. Likelihood ratios algebraically combine sensitivity and specificity to describe more than the independent values themselves, specifically the change in odds favoring disease given a particular test result. In 21 of the 136 patients, hysteroscopy detected polyps or submucous myomas, which curettage failed to detect. In 67 premenopausal women, ultrasonography demonstrated a sensitivity in diagnosing endometrial abnormalities of 88% and a specificity of 68%, using a cutoff point of 5 mm of single-layer endometrial thickness. The positive likelihood ratio was 2.8 (95% confidence interval [CI] 1.66-4.55) and negative likelihood ratio 0.18 (95% CI 0.07-0.46). In 69 postmenopausal women, using a cutoff point of 3 mm, the sensitivity, specificity, positive likelihood ratio and negative likelihood ratio were 97%, 74%, 3.7 (95% CI 2.18-6.38), and 0.04 (95% CI 0.01-0.28) respectively. Transvaginal ultrasonography is an excellent first-step diagnostic method of excluding the endometrial abnormalities in women with postmenopausal bleeding but is of limited use in premenopausal women with irregular bleeding. Ultrasonography may reduce the number of invasive endometrial investigations by approximately 40%.