Analysis of predictive efficacy of EuroSCORE Ⅱ and SinoSCORE for patients undergoing coronary bypass surgery in single center / 天津医药

@#Objective TocomparethepredictiveefficacyofEuroSCOREⅡandSinoSCOREinthepostoperative mortalityofChinesepatientsunderwentcoronaryarterybypassgrafting(CABG). Methods Theclinicaldataof4507 patientswithCABGatourdepartmentinJanuary2011andApril2015wereretrospectivelyanalyzed.Cardiovascularrisk stratificationwasperformedonpatientsusingEuroSCOREⅡandSinoSCORE.PatientsweredividedintoⅠ,Ⅱ,ⅢandⅣ groupsaccordingtothepredictedfatalityrates.Themortalityrateswerepredictedinallgroupsofpatientsrespectively. Predictive effectiveness was analyzed by the analysis of discernment and calibration force. Results The in-hospital mortalityratewas1.35%inallpatients,whilethemeanmortalityratepredictedbyEuroSCOREⅡwas1.470%±1.215% (95%CI：1.43-1.50), and predicted by SinoSCORE was 2.860%±3.454% (95% CI：2.76-2.96). The AUC values of EuroSCOREⅡandSinoSCOREwere0.728and0.716.ItwasfoundthatthecalibrationdegreeofEuroSCOREⅡwaspoor andSinoSCOREwasacceptabledetectedbyHosmer-LemeshowTest.EuroSCOREⅡunderestimatedthemortalityratesof groupⅣ,butoverestimatedmortalityratesinothergroupsofpatients.SinoSCOREunderestimatedmortalityratesofpatients ingroupⅠandoverestimatedmortalityratesinothergroupsofpatients.EuroSCOREⅡonlyachievedgooddiscrimination forpatientsofgroupⅠ(AUC=0.707),andSinoSCOREachievedgooddiscriminationforpatientsofgroupⅡ(AUC=0.754). EuroSCOREⅡoverestimatedthemortalityrateintheisolatedCABGgroupandunderestimatedmortalityratesinpatients withothercardiacsurgeries.SinoSCOREoverestimatedmortalityratesingroupⅡ.TheAUCvaluesofEuroSCOREⅡand SinoSCOREwere0.694and0.687inisolatedCABGgroup.TheAUCvaluesofEuroSCOREⅡandSinoSCOREwere0.772 and0.669inCABGcombinedwithothercardiacsurgeries.Conclusion EuroSCOREⅡhasagoodpredictiveefficacyin theentiregroupofpatientsandⅠ,ⅡandⅢgroups,buthasapoorperformanceingroupⅣ.SinoSCOREoverestimates mortalityratesintheentiregroupandⅠ,ⅡandⅢgroups,anditunderestimatesmortalityratesinpatientsofgroupⅠ. Theapplicationandestablishmentofriskmodelsshouldfocusondifferentheartdiseasesanddifferentrisklevels,andthe modelingmethodofestablishedrisksystemsneedstobeimproved.