INTRODUCTION: The aim of this study was to develop a multivariable model to predict the risk of endometrial carcinoma in postmenopausal women with vaginal bleeding using individuals' clinical characteristics. PATIENTS AND METHODS: This prospective study of consecutive postmenopausal women presenting with vaginal bleeding was conducted at a gynecological oncology center in the United Kingdom for a 46-month period. All women underwent transvaginal ultrasound scanning as the initial investigation tool to evaluate the endometrium. Women found to have an endometrial thickness 5 mm or more had endometrial sampling performed. RESULTS: Of a total of 3548 women presenting with vaginal bleeding during the study period, 201 (6%) women had a diagnosis of endometrial carcinoma. An investigator-led best model selection approach used to select the best predictors of cancer in the multiple logistic regression model showed that patient's age (odds ratio [OR], 1.06), body mass index (OR, 1.07), recurrent episodes of bleeding (OR, 3.64), and a history of diabetes (OR, 1.48) increased the risk of endometrial malignancy when corrected for other characteristics. The mentioned clinical variables satisfied the criteria for inclusion in our predictive model called FAD 31 (F for the frequency of bleeding episodes, A for the age of the patient, D for diabetes, and the number 31 represents the BMI cut-off value). The total score for the model varies from 0 to 8. The area under the receiver operating characteristics curve for the developed model was 0.73 (95% confidence interval, 0.70-0.77). DISCUSSION: We have developed a simple model based on patients' clinical characteristics in estimating the risk of endometrial cancer for postmenopausal women presenting with vaginal bleeding. The model shows reasonable discriminatory ability for women with cancer and without, with an area under the receiver operating characteristics curve of 0.73. This will allow clinicians to individualize the diagnostic pathway for women with postmenopausal vaginal bleeding.
INTRODUCTION: The aim of this study was to develop a multivariable model to predict the risk of endometrial carcinoma in postmenopausal women with vaginal bleeding using individuals' clinical characteristics. PATIENTS AND METHODS: This prospective study of consecutive postmenopausal women presenting with vaginal bleeding was conducted at a gynecological oncology center in the United Kingdom for a 46-month period. All women underwent transvaginal ultrasound scanning as the initial investigation tool to evaluate the endometrium. Women found to have an endometrial thickness 5 mm or more had endometrial sampling performed. RESULTS: Of a total of 3548 women presenting with vaginal bleeding during the study period, 201 (6%) women had a diagnosis of endometrial carcinoma. An investigator-led best model selection approach used to select the best predictors of cancer in the multiple logistic regression model showed that patient's age (odds ratio [OR], 1.06), body mass index (OR, 1.07), recurrent episodes of bleeding (OR, 3.64), and a history of diabetes (OR, 1.48) increased the risk of endometrial malignancy when corrected for other characteristics. The mentioned clinical variables satisfied the criteria for inclusion in our predictive model called FAD 31 (F for the frequency of bleeding episodes, A for the age of the patient, D for diabetes, and the number 31 represents the BMI cut-off value). The total score for the model varies from 0 to 8. The area under the receiver operating characteristics curve for the developed model was 0.73 (95% confidence interval, 0.70-0.77). DISCUSSION: We have developed a simple model based on patients' clinical characteristics in estimating the risk of endometrial cancer for postmenopausal women presenting with vaginal bleeding. The model shows reasonable discriminatory ability for women with cancer and without, with an area under the receiver operating characteristics curve of 0.73. This will allow clinicians to individualize the diagnostic pathway for women with postmenopausal vaginal bleeding.
Authors: Janelle B Pakish; Karen H Lu; Charlotte C Sun; Jennifer K Burzawa; Anthony Greisinger; Frances A Smith; Bryan Fellman; Diana L Urbauer; Pamela T Soliman Journal: J Womens Health (Larchmt) Date: 2016-06-02 Impact factor: 2.681