Abstract
BACKGROUND: Healthcare-associated infections caused by carbapenem-resistant Klebsiella pneumoniae (CRKP) are now a global public health problem, increasing the burden of disease and public healthcare expenditures in various countries. The aim of this study was to analyse the risk factors for CRKP infections and to develop nomogram models to help clinicians predict CRKP infections at an early stage to facilitate diagnosis and treatment. METHODS: The clinical data of patients with Klebsiella pneumoniae (KP) infections in our hospital from January 2018 to January 2023 were collected. 174 patients with CRKP infections and 219 patients with CSKP infections were selected for case-control study. 27 predictors related to CRKP infections were determined. The least absolute shrinkage and selection operator (Lasso) regression was used to screen the characteristic variables, Multivariate logistic regression analysis was performed on the selected variables and a nomogram model was established. The discrimination and calibration of the nomogram model were evaluated by receiver operator curves (ROC) and calibration curves. RESULTS: Six predictive factors of ICU stay, fever time, central venous catheterization time, catheter indwelling time, carbapenem use and tetracycline use screened by lasso regression were included in the logistic regression model, and the nomogram was drawn to visualize the results. The area under ROC curve of training set and validation set was 0.894 (95% CI: 0.857, 0.931) and 0.872 (95% CI: 0.805, 0.939); The results of decision curve analysis also show that the model has good prediction accuracy. CONCLUSION: This study established a nomogram to predict CRKP infection based on lasso-logistic regression model, which has certain guiding significance for early diagnosis of CRKP infections.