心脏外科术后患者胃肠道并发症风险预测模型的构建及效果验证

目的 构建心脏外科术后患者胃肠道并发症的风险预测模型,并验证预测效果。 方法 采用便利抽样法,选择2019年10月—2021年5月在郑州市某三级甲等医院行心脏外科手术的798例患者为研究对象,根据术后是否发生胃肠道并发症分为发生胃肠道并发症组（n=70）和未发生胃肠道并发症组（n=728）,对两组相关资料进行比较,应用Logistic回归分析建立风险预测模型,应用受试者操作特征曲线（receiver operating characteristic curve,ROC）下面积检验模型预测效果。选取2021年6月—8月符合标准的122例患者进行模型预测效果验证。 结果 心脏外科术后患者胃肠道并发症的风险预测模型共纳入6个危险因素：术前用药（OR=2.995）、术前日常生活活动能力（activities of daily living,ADL）得分（OR=0.989）、主动脉阻断时间（OR=1.737）、是否使用主动脉内球囊反搏（intra-aortic balloon pump,IABP）技术（OR=2.372）、首次经口进食时间（OR=1.795）、转出ICU时ADL得分（OR=0.986）。风险预测模型公式：Logit（P）=-2.224+1.097 × 术前是否用药+0.552 × 主动脉阻断时间+0.864 × 是否使用IABP技术+0.585 × 首次进食时间-0.014 × 转出ICU时ADL得分-0.011 × 术前ADL得分。该模型的ROC曲线下面积为0.788,灵敏度为0.857,特异度为0.594,Youden指数为0.451。模型验证结果显示,灵敏度为0.700,特异度为0.815,准确率为86.90%。 结论 该研究构建的心脏外科术后患者胃肠道并发症风险预测模型的预测效能良好,可为临床医护人员对心脏外科术后患者胃肠道并发症的评估提供参考及借鉴。

Construction of a risk prediction model for gastrointestinal complications in patients after cardiac surgery and its prediction effect

Objective To establish a risk prediction model for gastrointestinal complications in adult patients after cardiac surgery,and to verify the prediction effect. Methods A total of 798 patients after the cardiac surgery were selected as research subjects by convenience sampling method from October 2019 to May 2021 in Zhengzhou city,Henan province. The patients were divided into an occurrence group （n=70） and a nonoccurrence group （n=728） according to whether gastrointestinal complications occurred after the operation. Data of the 2 groups was compared to determine the potential risk factors,and Logistic regression analysis was performed to construct the prediction model. The area under the receiver operating characteristic curve（AUC） was used to test the prediction effect of the model. A total of 122 patients hospitalized in the same hospital from June 2021 to August 2021 were selected to verify the model. Results 6 risk factors were included in this study,namely preoperative medication （OR=2.995）,preoperative ADL score（OR=0.989）,aortic occlusion time（OR=1.737）,use of intra-aortic balloon pump（IABP）（OR=2.372）,the first time of eating（OR=1.795）,and ADL score of being transferred out of ICU（OR=0.986）. The model formula：logit（P）=-2.224+1.097 × preoperative medication+0.552 × aortic occlusion time+0.864 × use of IABP+0.585 × the first time of eating-0.014 × ADL scores when transferred out of ICU-0.011 × preoperative ADL scores. The ROC curve of this model was 0.788,with the sensitivity of 0.857,the specificity of 0.594,and the Youden index is 0.451.The model verification results showed that the sensitivity was 0.700;the specificity was 0.815;the accuracy was 86.90%. Conclusion The risk prediction model had satisfactory prediction,which can provide references for clinical staff to evaluate the gastrointestinal complications of patients after cardiac surgery.