Construction of an individualized early-warning model for predicting multidrug-resistant bacterial infection in ICU mechanically ventilated severe pneumonia patients

YAN Jingjing; XIAO Zhuangnan; WANG Lili

doi:10.3969/j.issn.1007-8134.2024.03.012

Abstract：Objective To construct a nomogram model of multi-drug resistant organisms (MDROs) infection in the intensive care unit (ICU) mechanically ventilated patients with severe pneumonia,by exploring the risk factors for infection with MDROs in mechanically ventilated patients with severe pneumonia in ICU. Methods A total of 240 mechanically ventilated patients with severe pneumonia admitted to Hefei Third People' s Hospital ICU from May 2019 to November 2022 were selected as the study subjects,and were divided into MDROs infection group and no MDROs infection group according to the MDROs infection status. Lasso analysis was used to screen the predictive factors of MDROs infection in ICU mechanically ventilated patients with severe pneumonia,Logistic regression was used to screen the risk factors of MDROs infection in ICU mechanically ventilated patients with severe pneumonia,and R software was used to build a nomogram model of MDROs infection in ICU mechanically ventilated patients with severe pneumonia. The nomogram model of MDROs infection in ICU mechanically ventilated patients with severe pneumonia was internally verified. Results Among 240 ICU patients with mechanical ventilation severe pneumonia,94 cases developed MDROs infection,and the incidence of MDROs infection was 39.17％ (94/240). Logistic regression analysis showed that diabetes,hypoproteinemia,disturbance of consciousness and nosocomial pneumonia were risk factors for MDROs infection in ICU patients with severe pneumonia due to mechanical ventilation (P<0.05 ). The predicted value of the correction curve of the nomogram model for MDROs infection in ICU patients with mechanically ventilated severe pneumonia was in good agreement with the actual value. The area under ROC curve of the model was 0.721 (95％CI:0.653～0.790 ). When the decision curve showed a threshold probability of 19％-100％,the omogram predicted a higher net benefit of MDROs infection in ICU mechanically ventilated patients with severe pneumonia. Conclusion The nomogram model of MDROs infection in ICU mechanically ventilated patients with severe pneumonia established in this study has high accuracy and clinical application value.