Probability analysis for diagnosis of endometrial hyperplasia and cancer in postmenopausal bleeding: an approach for a rational diagnostic workup

Objective. To develop an analytical approach to estimate the probability of endometrial hyperplasia and cancer in women with postmenopausal bleeding, using a combination of patient history and tests. Methods. A prospectively collected database of 428 postmenopausal women investigated for abnormal uterine bleeding in an outpatient rapid access clinic was used. The probability of disease was estimated using multivariable logistic regression models considering a combination of age and the use of hormone replacement therapy (HRT) and tests (ultrasound and hysteroscopy) in keeping with the actual clinical process. Results. The prevalence or prior probability of endometrial cancer in the sample studied was 4.4%. Hysteroscopy alone performed better than ultrasound alone. In all patients, a negative hysteroscopy reduced the probability of cancer to 1.3%. In women less than 60 years of age who use HRT, a negative hysteroscopy further reduced this probability to 0.1%. Overall, a positive hysteroscopy raised the probability of cancer to 38.9%. However, in women over 60 years of age not using HRT, a positive hysteroscopy increased this probability to 59.4%. Combining ultrasound results with hysteroscopy did not meaningfully alter the diagnostic probability. Conclusion. The combination of information obtained from patient history and tests allows generation of specific disease probabilities in postmenopausal bleeding. This type of multivariable analysis puts diagnostic information gained from history and other tests into a perspective that reflects actual clinical practice.