Abstract:

Objective　To construct and validate a risk model for initial weaning failure in patients with sepsis and respiratory failure. Methods　This study included 120 patients with sepsis and respiratory failure admitted to the Emergency Center, East Hospital, the First Affiliated Hospital of Xi'an Jiaotong University from January 2019 to January 2024. According to the outcome of weaning, they were divided into a successful weaning group (102 cases) and a failed weaning group (18 cases). There were 53 males and 49 females in the successful weaning group, aged (61.77±7.15) years. In the failed weaning group, there were 11 males and 7 females, aged (62.12±7.21) years. The general data were compared between the two groups. A risk model for initial weaning failure was constructed based on the results of logistic regression analysis and a nomogram was developed. The receiver operating characteristic curve (ROC) was established with initial weaning failure as the positive sample, and the fit of the risk model was verified by Hosmer-Lemeshow. Statistical methods used were t test and χ² test. Results　The smoking ratio, duration of mechanical ventilation, sepsis related Sequential Organ Failure Assessment (SOFA) score, Acute Physiology and Chronic Health Evaluation Ⅱ (APACHEⅡ) score, and rapid shallow breathing index (RSBI) in the failed weaning group were higher than those in the successful weaning group [72.2% (13/18) vs. 46.5% (47/102), (9.04±1.14) d vs. (6.58±0.96) d, (7.47±1.09) min vs. (5.14±0.84) min, (19.71±1.80) min vs. (17.39±1.64) min, (79.31±8.42) breaths/min/L vs. (58.38±6.45) breaths/min/L], with statistically significant differences (χ²=4.18, t=9.74, 10.35, 5.11, and 10.04, all P<0.05). Logistic regression analysis showed that smoking [P=0.011, 95% confidence interval (CI) 1.679-55.648], duration of mechanical ventilation (P=0.024, 95%CI 1.087-3.189), SOFA score (P=0.009, 95%CI 1.281-5.874), APACHEⅡ score (P<0.001, 95%CI 1.890-7.384), and RSBI (P=0.003, 95%CI 1.041-1.215) were the influential factors for initial weaning failure in patients with sepsis and respiratory failure. The ROC results showed that the combined prediction of smoking, duration of mechanical ventilation, SOFA score, APACHEⅡ score, and RSBI [area under the curve (AUC) = 0.980, sensitivity=94.7%, and specificity=92.1%] was significantly better than that of smoking (AUC=0.609, sensitivity=68.4%, and specificity=53.5%), duration of mechanical ventilation (AUC=0.868, sensitivity=57.9%, and specificity=99.0%), SOFA score (AUC=0.876, sensitivity=89.5%, and specificity=68.3%), APACHEⅡ score (AUC=0.811, sensitivity=73.7%, and specificity=78.2%), and RSBI alone (AUC=0.927, sensitivity=94.7%, and specificity=79.2%), and the risk model showed good fit through Hosmer-Lemeshow test. Conclusions　Smoking, duration of mechanical ventilation, SOFA score, APACHEⅡ score, and RSBI are significantly associated with initial weaning failure in patients with sepsis and respiratory failure. The constructed risk model has a good fit.

Key words:

Sepsis, Respiratory failure, Weaning failure, Risk model

摘要：

目的　构建脓毒症合并呼吸衰竭患者初次撤机失败的风险模型并验证。方法　选取2019年1月至2024年1月西安交通大学第一附属医院东院区急诊中心纳入的120例脓毒症合并呼吸衰竭患者，根据撤机结局分为撤机成功组和撤机失败组。撤机成功组102例，男53例，女49例，年龄（61.77±7.15）岁；撤机失败组18例，男11例，女7例，年龄（62.12±7.21）岁。根据logistic回归分析结果构建初次撤机失败的风险模型并构建列线图，以初次撤机失败为阳性样本建立受试者操作特征曲线（ROC），并经Hosmer-Lemeshow验证该风险模型的拟合度。统计学方法采用t检验、χ²检验。结果　撤机失败组的抽烟比例、机械通气时间、脓毒症相关性器官功能衰竭评价（SOFA）评分、急性生理学和慢性健康状况评价Ⅱ（APACHEⅡ）评分、呼吸浅快指数（RSBI）均高于撤机成功组［72.2%（13/18）比46.5%（47/102）、（9.04±1.14）d比（6.58±0.96）d、（7.47±1.09）分比（5.14±0.84）分、（19.71±1.80）分比（17.39±1.64）分、（79.31±8.42）次/min/L比（58.38±6.45）次/min/L］，差异均有统计学意义（χ²=4.18，t=9.74、10.35、5.11、10.04，均P<0.05）。logistic回归分析表明，抽烟［P=0.011，95%置信区间（CI）1.679～55.648］、机械通气时间（P=0.024，95%CI 1.087～3.189）、SOFA评分（P=0.009，95%CI 1.281～5.874）、APACHEⅡ评分（P<0.001，95%CI 1.890～7.384）、RSBI（P=0.003，95%CI 1.041～1.215）是脓毒症合并呼吸衰竭患者初次撤机失败的影响因素。ROC结果表明抽烟、机械通气时间、SOFA评分、APACHEⅡ评分及RSBI联合的预测效能［曲线下面积（AUC）=0.980，灵敏度94.7%，特异度92.1%］优于抽烟（AUC=0.609，灵敏度68.4%，特异度53.5%）、机械通气时间（AUC=0.868，灵敏度57.9%，特异度99.0%）、SOFA评分（AUC=0.876，灵敏度89.5%，特异度68.3%）、APACHEⅡ评分（AUC=0.811，灵敏度73.7%，特异度78.2%）及RSBI（AUC=0.927，灵敏度94.7%，特异度79.2%）单独，且经Hosmer-Lemeshow验证该风险模型的拟合度良好。结论　抽烟、机械通气时间、SOFA评分、APACHEⅡ评分及RSBI与脓毒症合并呼吸衰竭患者初次撤机失败相关，构建的风险模型拟合度良好。

关键词:

脓毒症, 呼吸衰竭, 撤机失败, 风险模型