Construction and validation of a risk prediction model for oral frailty in elderly patients with esophageal cancer

Background To explore the current status of oral frailty in elderly patients with esophageal cancer (EC), construct and verify the risk prediction model of oral frailty in elderly patients with EC, so as to provide a reference for early identification and intervention of oral frailty in this population. Methods In this study, 390 elderly patients with EC treated in the First Affiliated Hospital of Naval Medical University from January, 2023 to June, 2024 were selected as the training group, and 165 elderly patients with EC treated in Qingdao Hiser Hospital Affiliated of Qingdao University from July, 2024 to July, 2025 were selected as the validation group. A cross-sectional study was used for data collection. Three types of assessment tools were used, including the self-made general information questionnaire, outcome variable assessment scale and candidate variable assessment scale. The patients were divided into two groups according to the occurrence of oral frailty. LASSO regression and multivariate analysis were performed using SPSS 27.0 and R 4.4.3 software to identify the independent risk factors for oral frailty. The Bootstrap method with 1000 repeated samplings was used for internal validation of the model, while external validation was conducted using data from the validation group. The performance of the model was evaluated by the area under the receiver operating characteristic curve (AUC), calibration curve, Hosmer-Lemeshow test, decision curve analysis (DCA) and other indicators, and the Nomogram was drawn to visualize the model. Results The incidence of oral frailty in the training group and validation group was 45.90% and 43.03%, respectively. The results showed that radiotherapy history, tumor staging, physical frailty, smoking history, age, and nutritional status were independent risk factors for oral frailty in elderly patients with EC ( P< 0.05). The area under the ROC curve of the training group and the validation group were 0.812 (95% CI: 0.771–0.853) and 0.796 (95% CI: 0.730–0.862), respectively. Hosmer-Lemeshow test results (χ 2 = 12.382, P = 0.193) and (χ 2 = 14.922, P = 0.093) indicated that the model had a good goodness of fit. The consistency between the actual value of the calibration curve and the predicted value was high. The DCA results suggested that the model could obtain net benefits within a large threshold probability in both internal and external validation. Conclusion The risk of oral frailty in elderly patients with EC is high, which is related to radiotherapy history, tumor staging, physical frailty, smoking history, age, nutritional status and other factors. The oral frailty risk prediction model for elderly patients with EC constructed in this study has good predictive efficacy in internal and external validation, which can provide a reference for medical staff to identify high-risk groups early and take targeted intervention measures.