Development and validation of a nomogram model for predicting postoperative delirium in elderly patients with oral cancer: a retrospective study

Abstract

Objective: This study aimed to develop and internally validate a dynamic a nomogram model by analysing the risk factors for postoperative delirium (POD) in elderly patients with oral cancer. Methods: This was a single-centre, retrospective study. We used the convenience sampling method to select 359 elderly oral cancer patients from January 2020-August 2023 in Nanjing Stomatological Hospital. The original dataset was randomly divided into a training group (n = 252) and a validation group (n = 107) by a computer-generated random number sequence in a 7:3 ratio. Least Absolute Shrinkage and Selection Operator Regression (LASSO regression) were used to screen the best predictor variables. Logistic regression was used to build the model and visualized by nomogram. The performance of the model was evaluated by area under the curve (AUC), calibration curve and decision curve analysis (DCA). Results: Our predictive model showed that seven variables, age, sex, alcohol consumption history, marriage, preoperative anxiety, preoperative sleep disorder, and ICU length of stay, were associated with POD. The nomogram showed high predictive accuracy with an AUC of 0.82 (95% CI: 0.76–0.87) for the training group and 0.84 (95% CI: 0.76–0.92) for the internal validation group. In two groups, there was good agreement between the predicted results and the true observations. DCA showed that the predictive model had a good net clinical benefit. Conclusion: We developed a new predictive model to predict risk factors for POD in elderly oral cancer patients. The nomogram can help physicians assess POD quickly and effectively.