关键词:

老年重症患者, CRPA, 危险因素, 预测模型, 列线图

Abstract:

Objective　To investigate the risk factors of carbapenem-resistant Pseudomonas aeruginosa (CRPA) infection in critically ill elderly patients, and to establish a prediction model and validate it. Methods　The demographic and clinical data of 148 critically ill patients with CRPA infection treated at Guangdong Provincial Second Hospital of Traditional Chinese Medicine from January 2018 to December 2023 were collected, including the patients' gender, age, whether they were complicated with respiratory failure, cardiac insufficiency, hypertension, cerebral infarction, and other underlying diseases, and calcitoninogen, D-dimer, white blood cell count, hemoglobin, platelet count, blood urea, blood creatinine, blood glucose, glutamine transaminase, total protein, total bilirubin, and other laboratory data within 24 h after infection. The patients were divided into a young group (44 cases) and an old group (104 cases) according to their age. The young group was 18-65 (51.0±13.0) years old, and the old group ≥65-103 (79.8±8.0). The risk factors for the prognosis of CRPA infection were determined by univariate and multivariate logistic regression analyses using χ² test, t test, etc. The variables were screened by the LASSO regression, and were constructed into a nomogram prediction model. The model was evaluated using the area under the receiver operating characteristic curve (AUC), the standard curve, and the decision curve analysis (DCA). Results　Complicated with cardiac insufficiency and cerebral infarction, decreased hemoglobin and platelet count during infection, blood urea, blood creatinine, blood glucose, and calcitonin were the risk factors for CRPA infection in the critically ill elderly patients (all P<0.05). Complicated with cerebral infarction (OR=5.537; 95%CI 2.226-13.769) and decreased platelet count (OR=0.994; 95%CI 0.991-0.998) were the independent risk factors for CRPA infection in the critically ill elderly patients. Complicated with cerebral infarction and platelet count were screened according to LASSO regression to construct a nomogram prediction model. The model's AUC was 0.784; the calibration curve showed a 45° angle, so it had a good calibration ability; the clinical decision curve showed that the model had high net benefit in the range of 60%-90% treatment threshold probability. Conclusions　The variables used to construct the model are simple and effective. The constructed model has good differentiation and accuracy, and has good predictive value for CRPA infection in critically ill elderly patients. Timely intervention for high-risk groups within a certain range can result in good clinical benefits.

Key words:

Critically ill elderly patients, CRPA, Risk factors, Prediction model, Nomogram