Investigation on the risk of respiratory failure in severe pneumonia patients and its predictive model research
- Author:
CAO Weihan
;
ZHANG Liang
;
ZHANG Lei
- Publication Type:Journal Article
- Keywords:
Severe pneumonia;
respiratory failure;
multivariate analysis;
column chart model
- From:
China Tropical Medicine
2024;24(12):1540-
- CountryChina
- Language:Chinese
-
Abstract:
Objective To investigate the risk of respiratory failure (RF) in patients with severe pneumonia (SP), and to establish a prediction model to provide a reference for the development of individualized treatment plans for SP patients. Methods A study was conducted on 313 patients with SP admitted to the Affiliate Huanghe Sanmenxia Hospital of Henan University of Science and Technology from January 2020 to January 2023. Patients were divided into a model group of 219 cases and a validation group of 94 cases in a 7∶3 ratio. Clinical factors, clinical scores, and laboratory indicators that may affect RF in patients with SP were collected. According to the presence or absence of RF, the model group patients were further divided into the RF group and the non-RF group. Clinical factors, clinical scores, and laboratory indicators were compared between the two groups. Potential influencing factors were screened using least absolute shrinkage and selection operator (LASSO) regression, followed by multivariate logistic regression to identify independent influencing factors of RF in SP patients. A column chart model was established using R language and validated. Results Of 219 patients in the model group, 115 (52.51%) had RF. There were significant differences in age, smoking history, diabetes history, mechanical ventilation, hypoproteinemia, multidrug-resistant bacteria infection, acute physiology and chronic health score-Ⅱ (APACHE-Ⅱ), multiple organ dysfunction score (MODS), confusion urea respiratory rate blood pressure age-65 score (CURB-65), oxygenation index (OI), lactate (Lac), platelet (PLT), mean platelet volume (MPV), D-dimer (D-D), fibrinogen (Fib), and C-reactive protein (CRP) between the RF group and the non-RF group (P<0.05). The results of multivariate logistic regression analysis based on LASSO regression showed that hypoproteinemia, multidrug-resistant bacterial infection, APACHE-Ⅱ, Lac, PLT, MPV, and D-D were independent influencing factors for respiratory failure in SP patients. Receiver Operating Characteristic Curve (ROC) analysis results showed that the area under the curve (AUC) for predicting RF in SP patients was 0.874 (95%CI: 0.828-0.920) in the model group and 0.841 (95%CI: 0.788-0.894) in the validation group. Hosmer-Lemesho (H-L) goodness-of-fit test results showed that there was no statistically significant difference between the predicted and actual probabilities of RF in SP patients in the model group (χ2=2.432, P>0.05). The calibration curve results showed that the predicted curves of the model group and validation group were fitted with the standard curve. The results of the decision curves showed that when the probability threshold of concurrent RF in SP patients was 0.10 to 0.90, the net benefit to patients was greater than all patients with intervention or without intervention. Conclusions RF in SP patients is mainly influenced by factors such as hypoproteinemia, multidrug-resistant bacterial infection, APACHE-Ⅱ, etc. The column chart model established in this study has high accuracy and discriminative ability for predicting RF in SP patients.
- Full text:202511131531280265218.Investigation on the risk of respiratory failure in severe pneumonia patients and its predictive model research.pdf
