Abstract: BACKGROUND: Establishing a nomogram prediction model for postoperative pulmonary infection in hip fractures and taking early intervention measures is crucial for improving patients’ quality of life and reducing medical costs. 
OBJECTIVE: To construct a nomogram risk prediction model of postoperative pulmonary infection in elderly patients with hip fracture, and provide theoretical basis for feasible prevention and early intervention. 
METHODS: Case data of 305 elderly patients with hip fractures who underwent surgical treatment at Wuxi Traditional Chinese Medicine Hospital Affiliated to Nanjing University of Chinese Medicine between January and October 2020 (training set) were retrospectively analyzed. Using univariate and multivariate logistic regression analysis and Hosmer-Lemeshow goodness of fit test, receiver operating characteristic curve was utilized to analyze the diagnostic predictive efficacy of independent risk factors and joint models for postoperative pulmonary infections. Tools glmnet, pROC, and rms in R Studio software were applied to construct a nomogram model for predicting the risk of postoperative pulmonary infection in elderly patients with hip fractures, and calibration curves were further drawn to verify the predictive ability of the nomogram model. Receiver operating characteristic curves, calibration curves, and decision curves were analyzed for 133 elderly patients with hip fractures (validation set) receiving surgery at the same hospital from November 2022 to March 2023 to further predict the predictive ability of the nomogram model. 
RESULTS AND CONCLUSION: (1) The postoperative pulmonary infection rate in elderly patients with hip fractures in this group was 9.18% (28/305). (2) Single factor and multivariate analysis, as well as forest plots, showed that preoperative hospitalization days, leukocyte count, hypersensitive C-reactive protein, and serum sodium levels were independent risk factors (P < 0.05). The Hosmer-Lemeshow goodness of fit test showed good fit (χ2=4.57, P=0.803). Receiver operating characteristic curve analysis was conducted on the independent risk factors and their joint models mentioned above, and the differentiation of each independent risk factor and joint model was good, with statistical significance (P < 0.05). (3) The graphical calibration method, C-index, and decision curve were used to validate the nomogram prediction model. The predicted calibration curve was located between the standard curve and the acceptable line, and the predicted risk of the nomogram model was consistent with the actual risk. (4) The validation set used receiver operating characteristic curve, graphic calibration method, and decision curve to validate the prediction model. The results showed good consistency with clinical practice, indicating that the model had a good fit. The nomogram risk prediction model constructed for postoperative pulmonary infection in elderly patients with hip fractures has good predictive performance. The use of the nomogram risk prediction model can screen high-risk populations and provide a theoretical basis for early intervention.

Key words: hip fracture, lung infection, prediction model, nomogram, risk factor, R programming language