作者：周大文，杨晓梅，赵文婷，刘文君，王璐

单位：

1.江苏省淮安市第二人民医院呼吸与危重症医学科2.江苏省淮安市第二人民医院肝胆外科

单位（英文）：

1.Department of Respiratory and Critical Care Medicine, the Second People's Hospital of Huai'an, Huaian 223001, China 2.Hepatobiliary Surgery, the Second People's Hospital of Huai'an, Huaian 223001, China

关键词：

肺疾病,慢性阻塞性; 呼吸衰竭; 通气机,机械; 影响因素分析; 列线图;

关键词（英文）：

Pulmonary disease, chronic obstructive; Respiratory failure; Ventilators, mechanical; Root cause analysis; Nomograms

中图分类号：

DOI：

10.12114/j.issn.1008-5971.2023.00.130

基金项目：

摘要：

目的 探讨慢性阻塞性肺疾病(COPD)合并呼吸衰竭患者无创呼吸机治疗失败的影响因素，构建其风险预测列线图模型并进行验证。方法 采用便利抽样法选取2020年5月至2022年5月于淮安市第二人民医院行无创呼吸机治疗的COPD合并呼吸衰竭患者为研究对象。纳入样本量为710例，将纳入患者按照7︰3分为建模组(497例)及验证组(213例)。收集患者的临床资料。COPD合并呼吸衰竭患者无创呼吸机治疗失败的影响因素分析采用多因素Logistic回归分析；采用R 4.1.0软件包及rms程序包建立COPD合并呼吸衰竭患者无创呼吸机治疗失败的风险预测列线图模型；采用Hosmer-Lemeshoe拟合优度检验评价该列线图模型的拟合程度；绘制校准曲线以评估该列线图模型预测COPD合并呼吸衰竭患者无创呼吸机治疗失败的效能；采用ROC曲线分析该列线图模型对COPD合并呼吸衰竭患者无创呼吸机治疗失败的预测价值。结果 建模组497例COPD合并呼吸衰竭患者中，治疗失败129例，归为治疗失败组；治疗成功368例，归为治疗成功组。治疗失败组和治疗成功组年龄、机械通气时间和治疗前动脉血氧分压(PaO2)、动脉血二氧化碳分压(PaCO2)、pH值、呼吸频率、血清白蛋白、C反应蛋白(CRP)及入院时格拉斯哥昏迷量表(GCS)评分、急性生理学与慢性健康状况评分系统Ⅱ(APACHEⅡ)评分比较，差异有统计学意义(P<0.05)。多因素Logistic回归分析结果显示，年龄、机械通气时间和治疗前PaO2、PaCO2、血清白蛋白、CRP及入院时APACHEⅡ评分是COPD合并呼吸衰竭患者无创呼吸机治疗失败的影响因素(P<0.05)。基于多因素Logistic回归分析结果，构建COPD合并呼吸衰竭患者无创呼吸机治疗失败的风险预测列线图模型。Hosmer-Lemeshoe拟合优度检验结果显示，建模组该列线图模型拟合较好(χ2=6.355,P=0.607)，验证组该列线图模型拟合较好(χ2=6.337,P=0.591)。校准曲线分析结果显示，该列线图模型预测建模组和验证组COPD合并呼吸衰竭患者无创呼吸机治疗失败的校准曲线贴近理想曲线。ROC曲线分析结果显示，该列线图模型预测建模组和验证组COPD合并呼吸衰竭患者无创呼吸机治疗失败的AUC分别为0.871[95%CI(0.848,0.915)]、0.872[95%CI(0.819,0.925)]。结论 年龄>60岁、机械通气时间延长、治疗前PaO2降低、治疗前PaCO2升高、治疗前血清白蛋白降低、治疗前CRP升高、入院时APACHEⅡ评分升高是COPD合并呼吸衰竭患者无创呼吸机治疗失败的危险因素，基于上述因素构建的列线图模型对COPD合并呼吸衰竭患者无创呼吸机治疗失败具有一定预测价值。

英文摘要：

Objective To analyze the influencing factors of noninvasive ventilator treatment failure in patients with chronic obstructive pulmonary disease (COPD) and respiratory failure, and to construct a nomogram model for predicting its risk and validate it. Methods Patients with COPD and respiratory failure who underwent noninvasive ventilator therapy at the Second People's Hospital of Huai'an from May 2020 to May 2022 were selected as research subjects using a convenience sampling method. The included sample size was 710, and the included patients were divided into modeling group (n=497) and validation group (n=213) according to the ration of 7∶3. Clinical data of patients were collected. The multivariate Logistic regression analysis was used to analyze the influencing factors of noninvasive ventilator treatment failure in patients with COPD and respiratory failure. The nomogram model for predicting the risk of noninvasive ventilator treatment failure in patients with COPD and respiratory failure was constructed by using the R 4.1.0 software package and rms package. Hosmer-Lemeshow goodness of fit test was used to evaluate the fitting degree of the nomogram model. Calibration curve was drawn to evaluate the effectiveness of the nomogram model for predicting the risk of noninvasive ventilator treatment failure in patients with COPD and respiratory failure, and the ROC curve was used to analyze the predictive value of the nomogram model for noninvasive ventilator treatment failure in patients with COPD and respiratory failure. Results Among the 497 patients with COPD and respiratory failure in the modeling group, 129 cases of treatment failure were classified as the treatment failure group; 368 cases of treatment success were classified as the treatment success group. There were significant differences in age, mechanical ventilation time, pre-treatment arterial partial pressure of oxygen (PaO2) , pre-treatment arterial partial pressure of carbon dioxide (PaCO2) , pre-treatment pH value, pre-treatment respiratory rate, pre-treatment serum albumin, pre-treatment C-reactive protein (CRP) , Glasgow Coma Scale (GCS) score at admission, Assessment of Acute Physiology and Chronic Health Status Ⅱ (APACHEⅡ) score at admission between the two groups (P < 0.05) . Multivariate Logistic regression analysis showed that age, mechanical ventilation time, pre-treatment PaO2, pre-treatment PaCO2, pre-treatment serum albumin, pre-treatment CRP, APACHEⅡ score at admission were the influencing factors of noninvasive ventilator treatment failure in patients with COPD and respiratory failure (P < 0.05) . The nomogram model for predicting noninvasive ventilator treatment failure in patients with COPD and respiratory failure was constructed based on the multivariate Logistic regression analysis results. The results of Hosmer-Lemeshow goodness of fit test showed that the nomogram model fit well in modeling group (χ2 =6.355, P=0.607) and validation group (χ2 =6.337, P=0.591) . The results of calibration curve analysis showed that the calibration curve of noninvasive ventilator treatment failure in patients with COPD and respiratory failure predicted by the nomogram model was close to the ideal curve. The results of ROC curve analysis showed that the AUC of the nomogram model for predicting the occurrence of noninvasive ventilator treatment failure in patients with COPD and respiratory failure in modeling group and validation group was 0.871 [95%CI (0.848, 0.915) ] , 0.872 [95%CI (0.819, 0.925) ] , respectively. Conclusion Age > 60, prolonged mechanical ventilation time, decreased pre-treatment PaO2, increased pre-treatment PaCO2, decreased pre-treatment serum albumin, increased pre-treatment CRP, increased APACHEⅡ score at admission are the risk factors of noninvasive ventilator treatment failure in patients with COPD and respiratory failure. The nomogram model constructed based on the above factors has a certain predictive value for the risk of noninvasive ventilator treatment failure in patients with COPD and respiratory failure.

参考文献：