Analysis of hypoxemia in the patients of chronic obstructive pulmonary disease complicated with severe pneumonia based on Bayesian network model and the effect of interventional bronchoscopy

Liu Kaiyue; Xu Bo

doi:10.3969/j.issn.1002-1949.2024.03.008

Abstract：Objective To investigate the factors of hypoxemia in the patients of chronic obstructive pulmonary disease(COPD)complicated with severe pneumonia and the effect of intervention 85 bronchoscopy.Methods The clinical data of 183 patients of COPD complicated with admitted to the First Hospital of Fangshan District from October 2019 to October 2022 were retrospectively analyzed.The patients were randomly divided into training set(122 cases)and verification set(61 cases)at a ratio of 2∶1.The patients in the training set were divided into the hypoxemia group(37 cases)and the non-hypoxemia group(85 cases)according to whether hypoxemia occurred or not.The clinical data in the two groups were compared and the risk factors affecting the occurrence of hypoxemia were analyzed.Receiver operating characteristic(ROC)curve was used to evaluate the predictive value of various risk factors for the patients with hypoxemia.Bayesian network model was constructed for reasoning and prediction.The training set data were used to evaluate the discrimination and accuracy of the model,and the verification set data were used for external evaluation of the model.The therapeutic effect between bronchoscopic intervention combined with non-invasive positive pressure ventilation(NPPV)and conventional NPPV in COPD complicated with severe pneumonia was compared.Results Multivariate analysis showed that age,body mass index(BMI),smoking,white blood cell count,C-reactive protein value,oxygenation index,the lowest partial pressure of oxygen,and pulmonary function were all independent risk factors for the occurrence of hypoxemia(P<0.05).All 8 risk factors had high predictive value in the hypoxemia of COPD complicated with severe pneumonia patients(AUC =0.816,0.822,0.768,0.839,0.826,0.793,0.802,0.857),and the combined predictive results of multiple factors(AUC =0.899)were higher than the single index(Z =2.314,2.067,4.558,1.874,2.013,3.750,3.109,1.422,P<0.05).The results of training set and verification set showed that the Bayesian network prediction model has good discrimination and accuracy.Compared with conventional NPPV treatment,the combination of bronchoscopic intervention and NPPV treatment significantly reduced acid correction time,non-invasive ventilation time,and hospital stay;The incidence and mortality rate of invasive ventilation were significantly reduced,and the difference was statistically significant(all P<0.05).Conclusions Age,BMI,smoking,white blood cell count,C-reactive protein value,oxygenation index,the lowest partial pressure of oxygen,and pulmonary function are all risk factors for the hypoxemia of COPD complicated with severe pneumonia patients.The therapeutic effect of bronchoscopic intervention combined with NPPV is significantly better than conventional NPPV treatment.