Validation of a risk prediction model for endometrial carcinoma among postmenopausal women in Sri Lanka

Abstract

Introduction: A risk prediction model was developed to predict the risk of developing endometrial carcinoma among postmenopausal women. It is strongly recommended to evaluate the performance of the risk prediction model by external validation before using it in the community. Objectives: To validate a risk prediction model for endometrial carcinoma among postmenopausal women in Western Province Methods: An unmatched case control study was conducted to validate a developed model, including 35 cases and 50 controls of postmenopausal women from five tertiary hospitals in Western Province. The cases and controls were defined as postmenopausal women who had and had not been diagnosed as endometrial carcinoma based on histological confirmation, respectively. A model was developed considering the objectivity and feasibility of the measurements in addition to statistical criteria. The developed model consisted of six predictors: age >55 years, never conceived, age at menarche :S 11 years, ever experienced postmenopausal bleeding, having family history of any type of cancer among first degree relative and generalized obesity. A scoring system was used based on weighted score of each risk predictor (0-9). An interviewer-administered-questionnaire was used. Predictive validity of the model was tested by calibration and discrimination. Receiver Operator Characteristic (ROC) curve was used to determine the cut-off value. Results:Area under the ROC curve was 0. 78 (95% confidence interval (CI): 0.68, 0.88). This indicates the ability of the total risk score to correctly classify those with and without disease is 78%. The calibration of the model was satisfactory (Hosmer-Lemeshow test; 2.07; p=0.84). The model correctly predicted 71.8% of the observed outcomes with overall classification. At a cut-off point of 3.5, the risk prediction model demonstrated a sensitivity of 65. 7%, specificity of 76.0%, positive predictive value of 65. 7% and negative predictive value of76.0%. Conclusions & Recommendations: The model demonstrated valid predictions as a risk prediction tool. It is recommended to incorporate the validated risk prediction model into the existing health care system to prioritize the women who need further investigations or closely follow up.

Keywords

• Endometrial cancer
• Risk prediction model
• Validation
• Postmenopausal women