急性呼吸窘迫综合征患者早期风险预测模型的建立

Establishment of an early risk prediction model for patients with acute respiratory distress syndrome in ICU

目的建立适合ICU医护人员早期预测患者急性呼吸窘迫综合征(acute respiratory distress syndrome,ARDS)发生风险的模型。方法采用便利抽样的方法,收集2017年1月—2018年4月入住ICU的符合标准的患者223例,按照其是否发生ARDS分为ARDS组(n=69)和非ARDS组(n=154),并对两组资料进行对比,利用Logistic回归分析建立风险预测模型,应用ROC曲线下面积检验模型预测效果。结果ARDS早期风险预测模型纳入年龄(OR=1.136)、肺损伤预测评分(OR=2.328)、英国国家早期预警评分(OR=2.830)和肺炎严重指数(OR=1.033)4个预测因子。模型公式Z=0.128×年龄+0.845×肺损伤预测评分+1.040×英国国家早期预警评分+0.032×肺炎严重指数-20.059。其ROC曲线下面积为0.926,灵敏度为0.906,特异度为0.812,Youden指数为0.718。结论本模型可以有效预测ICU患者ARDS的发生,且所有指标在入住ICU早期即可获得,可为ICU医护人员早期及时对不同风险分层患者采取预防性护理干预提供借鉴。

Objective To establish a predictive model for ICU medical staff to early predict the risk of occurrence of acute respiratory distress syndrome(ARDS). Methods A total of 223 patients who were admitted to the ICU between January 2017 and April 2018 and met the criteria were enrolled in this case-control study. They were classified into an ARDS group(n=69) and a non-ARDS group(n=154). The clinical data between the groups were compared. A risk prediction model was established by binary logistic regression,and the effectiveness of the model was verified by the area under the ROC curve. Results Age(OR =1.136),lung injury prediction score(OR=2.328),UK national early warning score(OR=2.830) and pneumonia severity index(OR=1.033) were included in this ARDS early risk prediction model. The model formula was Z=0.128 × age+0.845 × lung injury prediction score +1.040 × national early warning score +0.032 × pneumonia severity index-20.059. The area under the ROC curve of this model was 0.926 with a sensitivity of 0.906,a specificity of 0.812 and the Youden index of 0.718.Conclusion The risk prediction model described here can effectively predict the occurrence of ARDS in ICU patients,and all 4 indicators can be obtained in the early stage of ICU,thus providing references for ICU medical staff to take preventive care interventions for patients with different risk stratifications in the early stage.