The diagnostic accuracy of ultrasound scan in predicting endometrial hyperplasia and cancer in postmenopausal bleeding

Objective. To determine the accuracy of ultrasound scan in the diagnosis of endometrial hyperplasia and cancer in postmenopausal bleeding. Design. A prospective diagnostic accuracy study (1996-97). Setting. Minimal access surgical training centers in two large teaching hospitals. Methods. Ultrasound scan and outpatient endometrial sampling were performed on 96 patients with postmenopausal bleeding. Patients unable to have these outpatient procedures had a formal inpatient hysteroscopy and curettage. Test performance characteristics were computed for ultrasound scan comparing its estimate of endometrial thickness with histologic diagnosis that served as a 'gold' standard. Outcome measures. Accuracy of the ultrasonic endometrial thickness was estimated using sensitivity, specificity and predictive values for binary data. For multilevel data, the diagnostic accuracy was computed using likelihood ratios (LRs). An LR <1 decreased the probability that endometrial hyperplasia/cancer was present, whereas an LR >1 increased the probability that such lesion was present. Results. Using endometrial thickness greater than or equal to 4 mm, the sensitivity of ultrasound to detect the endometrial malignancy was 92.9%, the specificity was 50%, and the positive and negative predictive values were 24.1% and 97.6% respectively. Analysis using likelihood ratio (LR) revealed that LR was 0.14 for endometrial thickness greater than or equal to 4.0 mm, 0.94 for endometrial thickness 4.1-9.0 mm, and 3.3 for endometrial thickness >9.0 mm. Conclusion. In women with postmenopausal bleeding, malignancy can probably be safely excluded if sonographic endometrial thickness is less than or equal to 4.0 mm. However, the probability of endometrial hyperplasia/cancer is not particularly altered by the knowledge that endometrial thickness on scan is >4.0 mm.