Abstract: Objective  To establish a method for predicting the risk of endometrial cancer (EC) and endometrial atypical hyperplasia (AH) in women with postmenopausal bleeding (PMB) by collecting clinical data on routine medical history.  Methods  The clinical data of a total of 408 PMB patients admitted to Fuxing Hospital, Capital Medical University were consecutively collected in this retrospective study from December 2013 to December 2023. According to the results of endometrial pathology, patients were divided into case group and control group. EC and AH were included in the malignant group (case group) and the other endometrial pathologies were included in the non-malignant group (control group). Clinical data, including clinical history, high risk factors, and common gynecological ultrasound measurement indicators, were collected and studied by univariate and multivariate Logistic regression analysis. Results  The mean age of 408 patients was (60.4±7.8) years. A total of 74 cases (18.1%) were in case group and 334 cases (81.9%) were in control group. Based on Logistic regression analysis, the best predictors of endometrial malignant lesions were selected to create a “LRDNT” (light bleeding, recurrent bleeding, diabetes, non-uniform echogenicity & thickness) model. LRDNT scores range from 0 to 22. The score of LRDNT ≥15 has the largest Yoden index, and the sensitivity to predict endometrial malignant lesions is 79.73%, the specificity is 80.84%, and the prediction accuracy is 80.64%. Conclusions  The risk prediction model LRDNT, which combines clinical information and common gynecological ultrasound measurement indicators of PMB patients, can help clinicians classify patients at high and low risk of endometrial malignant lesions, and optimize the strategy of diagnosis and treatment. 

Key words: postmenopausal bleeding,  endometrial cancer,  endometrial atypical hyperplasia,  risk factors,  prediction model