1. Establishment and validation of a risk prediction model for postoperative delirium in elderly patients after individualized hip fracture surgery
- Author
-
WANG Tianpei, CAI Yongsong, GUO Hua, and SUN Ye
- Subjects
hip fracture ,aged ,delirium ,prediction model ,nomogram ,Medicine (General) ,R5-920 - Abstract
Objective To establish and verify a postoperative delirium risk prediction model for elderly patients after individualized hip fracture surgery. Methods Clinical data of 393 elderly patients with hip fracture who underwent surgical treatment in the Department of Orthopedics, Xi'an Central Hospital from December 2016 to January 2021 were collected retrospectively in this study. The data from 333 patients during December 2016 and July 2020 were used as the model training set, and postoperative delirium was regarded as the research endpoint. Based on the model training set, uniuariate and multivariate logistic regression were adopted to screen out independent related factors of postoperative delirium. The model was established by regression coefficient, the nomogram was drawn combined with the rms package of R language, and the discrimination, calibration and clinical applicability of the model were tested. Bootstrap was used to repeatedly sample 1 000 times in the training set for internal verification of the model, and the data of 60 patients treated from August 2020 to January 2021 were employed as the verification set for external verification of the model. Results There were 393 elderly patients with hip fracture, including 333 cases in the training set, and 60 cases in the validation set. Univariate and multivariate logistic analysis showed that dementia, preoperative albumin level and intraoperative hypotension were independent related factors of postoperative delirium in elderly patients with hip fracture. The prediction model is obtained according to the relevant factors and regression coefficients: P=1/(1+e-Z), Z=1.396×dementia or not -0.275× preoperative albumin level +1.420×intraoperative hypotension or not +6.875. The cut-off value of the model in the training set is 0.235, and the specificity and sensitivity is 0.875 and 0.622, respectively. The cut-off value of the model in the validation set is 0.211, and the specificity and sensitivity is 0.800 and 0.800, respectively. The area under curve (AUC) of the receiver operating characteristic curve (ROC) in the training set and the validation set is 0.827 (95%CI: 0.762~0.892) and 0.836 (95%CI: 0.716~0.956) respectively. After 1 000 repeated sampling by bootstrap, the AUC of the internal verification of the model is 0.831. The slopes of the calibration curves of the training set and the validation set is 1 and 1.066, respectively. The P values of the Hosmer-Lemeshow goodness-of-fit test are 0.124 and 0.743, respectively. The clinical decision curve showed that the threshold probability interval of the maximum net benefit value of the postoperative delirium predicted by the model in the training set and the validation set was about 5%~90% and 3%~90% respectively. Conclusion Our established model can predict the risk of postoperative delirium in elderly patients after hip fracture surgery.
- Published
- 2022
- Full Text
- View/download PDF