A prediction model of multidrug resistant bacterial for inpatients with pneumonia
10.19485/j.cnki.issn2096-5087.2025.05.007
- Author:
BAI Ruiying
;
SHENG Haiyan
- Publication Type:Journal Article
- Keywords:
pneumonia;
inpatient;
multidrug resistant bacteria;
influencing factor;
prediction model
- From:
Journal of Preventive Medicine
2025;37(5):465-470
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To create a prediction model of multidrug resistant bacterial infections for inpatients with pneumonia, so as to provide the reference for the early identification and intervention of multidrug resistant bacterial infections.
Methods:The inpatients with pneumonia in the Second Affiliated Hospital of Bengbu Medical University from October 2022 to June 2024 were selected as the research subjects. Basic information and clinical data of the patients were collected. Respiratory secretions were collected for etiological culture and drug sensitivity tests to analyze the infection situation of multidrug resistant bacteria. LASSO regression and a multivariable logistic regression model were used to screen predictive factors and establish a predictive model of multidrug resistant bacterial infections for inpatients with pneumonia. The predictive effect of the model was assessed by receiver operating characteristic (ROC) curve, calibration curve, and decision curve.
Results:A total of 368 inpatients with pneumonia were recruited, including 215 males (58.42%) and 153 females (41.58%). The median age was 71.00 (interquartile range, 20.00) years. There were 168 cases of multidrug resistant bacterial infections detected, with a detection rate of 45.65%. The multivariable logistic regression analysis showed that long-term bedridden patients (OR=2.699, 95%CI: 1.120-6.504), use of antibiotics within 30 days (OR=8.623, 95%CI: 2.949-25.216), respiratory failure (OR=2.407, 95%CI: 1.058-5.478), intensive care unit treatment (OR=3.995, 95%CI: 1.313-12.161), and hypoproteinemia (OR=2.129, 95%CI: 1.012-4.480) were predict factors of multidrug resistant bacterial infections for inpatients with pneumonia. The area under the ROC curve of the established multidrug resistant bacterial infection prediction model was 0.909 (95%CI: 0.879-0.939). The calibration curve after repeated sampling calibration approached the standard curve, and the predicted values were highly consistent with the measured values. The decision curve showed that when the probability threshold is 0.27-0.99, the clinical net benefit for predicting the risk of multidrug resistant bacterial infection is relatively high.
Conclusion:The prediction model of multidrug resistant bacteria infection constructed has a good predictive value for multidrug resistant bacterial infection among inpatients with pneumonia
- Full text:2025110410545850264肺炎住院患者多重耐药菌感染的预测模型研究.pdf
