Objective: To compare the diagnostic accuracy of transvaginal ultrasound an
d hysteroscopy in the detection of endometrial pathologies in women with po
stmenopausal bleeding not using hormonal replacement therapy (HRT). Methods
: Between January 1997 and April 1998, 106 postmenopausal women with uterin
e bleeding not using HRT underwent a diagnostic work-up including pelvic ex
amination, transvaginal ultrasound, hysteroscopy and endometrial biopsy. So
nographic measurement of endometrial thickness and hysteroscopic findings w
ere compared with histological results. The 'classification tree' method wa
s used to identify cut-off values of sonographic endometrial thickness that
could be indicative of a class of uterine pathology. Statistical analysis
was performed with the McNemar test. Results: No case of endometrial cancer
was found with a cut-off point of 5 mm of endometrial thickness evaluated
by ultrasound, whereas all patients with endometrial thickness greater than
or equal to 15 mm at sonography had an endometrial carcinoma. In the group
of patients with endometrial thickness between 6 and 14 mm, we found norma
l atrophic endometria, benign and malignant pathology. On the other hand, t
he McNemar test showed a very good correspondence between hysteroscopy and
histology (sensitivity 97.5% and specificity 100%), confirming its usefulne
ss in diagnosis of postmenopausal uterine bleeding. Conclusions: Transvagin
al ultrasound has revealed some limitations, mainly in the group of patient
s with endometrial thickness between 6 and 14 mm. The absence of endometria
l malignancy in women with endometrial thickness less than or equal to 5 mm
and the high possibility of cancer in those with endometrial thickness gre
ater than or equal to 15 mm should be confirmed in larger series. Hysterosc
opy proved to be a simple and safe outpatient procedure with a high diagnos
tic accuracy, and in our opinion it should be considered in all women with
postmenopausal uterine bleeding. (C) 1999 Elsevier Science Ireland Ltd. All
rights reserved.