Construction and Validation of Risk Prediction Model for Gastrointestinal Bleeding After Cardiopulmonary Bypass Heart Surgery
Objectives:To establish a risk prediction model for gastrointestinal bleeding after cardiopulmonary bypass heart surgery,and to verify the prediction efficacy.Methods:A total of 1 002 patients who underwent cardiopulmonary bypass heart surgery in the department of cardiac great vascular surgery of our hospital from January 2019 to November 2023 were collected by convenient sampling method.They were divided into gastrointestinal bleeding group(n=47)and non-gastrointestinal bleeding group(n=955).Logistic regression analysis was used to establish the risk prediction model,and the area under ROC curve test and Hosmer-Lemeshow χ2 test were used to compare the two groups of data Model prediction effect.Bootstrap method was used for internal validation.Results:The risk prediction model of gastrointestinal bleeding after cardiopulmonary bypass heart surgery included four predictors:time of aortic occlusion(OR=1.021,95%CI:1.012-1.030),history of digestive disease(OR=5.710,95%CI:1.697-19.212),use of intra-aortic balloon counterpulsation(OR=22.180,95%CI:5.870-83.808),and continuous kidney replacement therapy(OR=12.159,95%CI:5.066-29.181).Model formula:Logit(P)=-5.821+0.021×time of aortic occlusion+1.742×history of digestive disease+3.099×whether intra-aortic balloon counterpulsation was used+2.498×whether continuous renal replacement therapy was used.The area under ROC curve was 0.812(95%CI:0.746-0.877),sensitivity was 64.6%,specificity was 85.7%,and Youden index was 0.503.After internal validation by Bootstrap method,the consistency index after correction is 0.813.Conclusions:The risk prediction model constructed in this study cohort has a good auxiliary prediction performance for the occurrence of gastrointestinal bleeding after cardiopulmonary bypass surgery,which is helpful for risk stratification for gastrointestinal bleeding after cardiopulmonary bypass surgery and facilitate clinical decision-making in daily clinical work.
NETL
NSTL
万方数据
