Development Of A Risk Model For In-hospital Death Of Elderly Patients Undergoing Valve Replacement Surgery

Background:Valve replacement surgery is an effective treatment for patients with severe valvular heart disease.However,valve replacement surgery has higher risk of serious postoperative complications and higher mortality than other general surgery.Elderly patients with reduced organ function and poor tolerance to surgery have even higher mortality.Therefore,early identification of high-risk patients and good perioperative management are the key measures to improve the prognosis of elderly patients undergoing valve replacement surgery.However,the risk model for elderly patients undergoing valve replacement surgery is absent.This study is to develop and validate a simple scoring system to predict in-hospital death of elderly patients undergoing valve replacement surgery.Methods:Our study included a total of 4554 elderly patients(age≥60 years)undergoing valve replacement surgery at Guangdong Provincial People’s Hospital from January 2010 to April 2019.Patients were assigned randomly(3:1)to the development cohort(n=3418)and the validation cohort(n=1136).The independent risk factors and the logistic regression model(EVRS logistic model)for in-hospital death of elderly patients undergoing valve replacement surgery were identified by stepwise multivariable logistic regression according development cohort data.The risk score(EVRS score)was developed by the the coefficient β of logistic regression model.The calibration and discrimination power of EVRS score were evaluated by Hosmer-Lemeshow goodness-of-fit test and area under the receiver operating characteristic curve,which were compared with EuroSCORE Ⅱ.Results:The in-hospital mortality was 4.9%and 4.1%in the development and validation cohort,respectively.EVRS score has two parts with 8 variables:preoperative variables(age>66 years,NYHA III-IV,serum creatinine>200umol/L,severe tricuspid regurgitation)and intraoperative variables(vasoactive agent usage,RBC transfusion>5U,mechanical assist,CPB duration>190min).The calibration and discrimination power for in-hospital death was excellent in development cohort(area under the curve[AUC]=0.791;Hosmer-Lemeshow chi-square=2.37,p=0.306)and internal validation cohort(AUC=0.830;Hosmer-Lemeshow chi-square=0.35,p=0.839).The discrimination power of EVRS score for in-hospital death was higher than that of EuroSCORE Ⅱ(development cohort:AUC 0.791 vs 0.722,p=0.0003;validation cohort:AUC 0.830 vs0.739,P=0.0217).Conclusions:EVRS score with great calibration and discrimination power was a simple and robust risk-stratified tool for in-hospital death of elderly patients undergoing valve replacement surgery.