Construction of Risk Factors and Predictive Models for Stroke Patients with Falling Pneumonia

Zhou Yujing; Chen Fang; Liu Qiuyi

doi:10.3969/j.issn.1004-5775.2025.16.005

Abstract：Objective:To investigate the risk factors of hypostatic pneumonia in stroke patients and build a prediction model.Methods:600 stroke patients admitted to the hospital from October 2021 to November 2023 were included in the study.500 of them were included in the modeling group,and the other 100 patients were included in the validation group for external verification of the model.The incidence of hypostatic pneumonia was compared between the two groups.The clinical characteristics and laboratory indicators of patients were analyzed.The main influencing factors of hypostatic pneumonia were identified.And the risk prediction model of hypostatic pneumonia was established.The ROC curve was drawn and the Hosmer-Lemeshow method was used to evaluate the efficacy of the risk prediction model for hypostatic pneumonia.Results:There was no statistically significant difference in the incidence of atelectasis pneumonia between the two groups(χ2=0.027,P=0.868).Comparison of data such as age,GCS score,length of hospital stay,gastric tube feeding,dysphagia,consciousness disorder,concurrent atrial fibrillation,concurrent hyperlipidemia,concurrent chronic obstructive pulmonary disease,mechanical ventilation,H2 receptor blockers,and broad-spectrum antibiotics between the two groups of patients was conducted.The differences were all statistically significant(χ2=6.212,90.523,17.224,34.464,28.965,37.526,30.904,33.830,231.278,57.845,66.678,110.025;P＜0.05).The levels of white blood cells,N,fibrinogen and urea nitrogen in the atelectatic pneumonia group were significantly higher than those in the non-atelectatic pneumonia group,while the levels of hemoglobin,glycated hemoglobin and homocysteine were significantly lower than those in the non-atelectatic pneumonia group.The differences were all statistically significant(t=23.317,16.532,9.696,18.738,3.552,7.293,25.994;P＜0.05).The logistic regression model and the results of multivariate analysis showed that age,length of hospital stay,dysphagia,consciousness disorder,concurrent atrial fibrillation,concurrent chronic obstructive pulmonary disease,mechanical ventilation,H2 receptor blockers,and broad-spectrum antibiotics all belong to the risk factors for the occurrence of atretic pneumonia in stroke patients(OR=1.733,1.104,1.235,3.188,2.464,4.035,3.706;P＜0.05).The maximum Youden index of the risk prediction model for atelectatic pneumonia was 0.771,the optimal cut-off value was 0.253,and the sensitivity and specificity were 92.03%and 93.15%respectively,showing high consistency.Conclusion:The risk prediction model of hypostatic pneumonia built based on the clinical characteristics and laboratory index detection results of stroke patients has high screening value and can help the development of treatment plans for patients with the disease,and has significant clinical value.