Objective To estimate the accuracy of transvaginal ultrasound (TVS) measurement of endometrial thickness (ET) in diagnosing endometrial cancer in postmenopausal women with vaginal bleeding (PMB).Design Retrospective cohort study.Setting One-stop PMB clinic in a Hong Kong teaching hospital. Main outcome measures Accuracy data for TVS ET presented as sensitivity, specificity, and area under the receiver operator characteristic (ROC) curve.Results Endometrial cancer was diagnosed in 3.8% of women. The median ET in those with endometrial cancer was significantly higher than those with benign conditions (15.7 versus 3.2 mm, P < 0.001). The area under the ROC curve was 0.92 (95% CI 0.89-0.94). The sensitivity for the detection of endometrial cancer at 3-, 4-, and 5-mm cut-offs were 97.0% (95% CI 94.5-99.6%), 94.1% (95% CI 90.5-97.6%), and 93.5% (95% CI 89.7-97.2%), respectively. The corresponding estimates of specificity at these thresholds were 45.3% (95% CI 43.8-46.8%), 66.8% (65.4-68.2%), and 74.0% (72.7-75.4%).Conclusions Transvaginal ultrasound using a 3-mm cut-off has high sensitivity for detecting endometrial cancer and can identify women with PMB who are highly unlikely to have endometrial cancer, thereby avoiding more invasive endometrial biopsy.

Most reports on histopathological findings of postmenopausal bleeding were published before hysteroscopy was widely used. Recent studies have shown that hysteroscopic examination is better than dilatation and curettage alone in detecting intrauterine pathology. The aim of this study is to assess histopathological findings of patients presented with postmenopausal bleeding after incorporation of hysteroscopic examination. Between 1 January 1996 and 31 December 1996, a total of 199 consecutive patients presenting with postmenopausal bleeding to a regional hospital were reviewed. Patients were investigated with cervical biopsy, endometrial aspiration or hysteroscopy and curettage. Excluding patients who had hysterectomy and cervical cancer, 97% of the patients had hysteroscopic examination. The majority of patients with postmenopausal bleeding had atrophic vaginitis and/or atrophic endometrium (58.8%). Other common diagnoses were endometrial polyp (9.4%), endometrial carcinoma (9.4%) and carcinoma of the cervix (6%). Other diagnoses included submucous fibroid, endometrial hyperplasia, pyometra, ovarian cancer and urethral caruncle. Pathologies found were similar to early reports, except that incidences of submucous fibroids and endometrial polyps were higher (13.4%). We concluded that all patients with postmenopausal bleeding should be investigated thoroughly as 22% have significant pathology. Hysteroscopy is the preferred investigation because of high incidences of endometrial polyps and submucous fibroids.