Evidence-based Evaluation Of Mortality Risk Prediction Models For Acute Coronary Syndrome

Aim at:Systematic evaluate the accuracy of prediction models for the risk of mortality in patients with acute coronary syndrome(ACS)In order to provide evidence for clinical nursing staff to choose right predictive model for patients with acute coronary syndrome.Aim to promote the application of risk prediction model in nursing,so as to inprove the nursing quality for ACS patients and reduce the mortality of ACS patients.Methods:We use evidence-based method to evaluate prediction models for the risk of mortality in patients with ACS.First searching literatures,retrieval method was computer retrieval combined with manual retrieval.We search:MEDLINE?EMBASE?CINAHL?Web of Science?Cochrane Library?CNIK etc.Search terms were: coronary disease?Coronary heart disease?myocardial infarction?myocardial infarct*?predict model?predict instrument?predict score?predict index?prognose model etc.Time limit is 2009 to 2019.The selected literatures were analyzed and data were extracted.We use Meta-analysis software to analyze the date and descriptive analysis was performed on studies that were unable to perform meta-analysis.Results:A total of 8,277 papers were included in this study,and 25 were finally included after screening.A total of 306,390 ACS patients were included,Among them,there were 158,080 male patients(51.6%)and 147,793 female patients(48.4%).The patients came from 11 countries The combined effect sizes of the models in the prediction model accuracy study showed that the GRACE model: sensitivity =0.78,specificity =0.76,and AUC =0.86;CAMI model :sensitivity=0.78,specificity =0.70,and AUC =0.85;TIMI model: sensitivity =0.51,specificity =0.81,and AUC =0.64;REMS model: sensitivity =0.78,specificity =0.46,and AUC =0.41;The other included 8 prediction models only included one literature so that could not be combined with effect size(EPICOR?CRUSADE?SAMI?GWTG?LNS?SYNTAXII?APACHEII)sensitivity between 0.77 with 0.95,specificity between 0.22 with 0.99,AUC between 0.71 with 0.92.In the comparative accuracy of prediction model,6 papers were finally included after screening,the Combined effect size is shown that GRACE model(AUC=0.79)better than TIMI model(AUC=0.59);GRACE model(AUC=0.73)better than REMS model(AUC=0.41)too;The others effect size cannot be combination's literatures showed APACHEII model(AUC=0.82)a little lower than GRACE model(AUC=0.86).Conclusion:The GRACE model and CAMI model showed good accuracy for the risk of mortality in ACS patients,TIMI model and the REMS model have average accuracy.The APACHEII model,SYNTAXII model and EPICOR model also show good accuracy for the mortality risk of ACS patients.However,evidence was not enough and further studies on these models are needed.In order to reduce the mortality of ACS patients and improve the quality of nursing,we recommend nurse use GRACE model to manage the risk of mortality in ACS patients.