Epidemiological Survey And Development Of Risk Predication Model For In-hospital Cardiac Arrest Of Acute Coronary Syndromes

Objectives1.To investigate the incidence of in-hospital cardiac arrest(IHCA),return of spontaneous circulation(ROSC)and survivor to discharge for Acute Coronary Syndromes(ACS)patients,and to explore the influencing factors of ROSC and survivor to discharge.2.To analyze the independent risk factors of cardiac arrest by multicenter case-control study.3.To develop a risk prediction model for cardiac arrest of ACS patients,and evaluate the efficiency of the model.Methods1.The incidence and outcomes of IHCA in ACS patients were evaluated in this retrospective study between the date of discharge from January 1,2012 to December 30,2016 of 3 hospitals in Fujian province.Univariate and multivariate logistic analysis were used to identify the influencing factors of ROSC and survivor to discharge.2.We conducted a retrospective case-control study of 88 patients experiencing CA,in the control group,we random selected the ACS patients who had no experienced CA residing in the same ward,with the same admission time and the same disease.Univariate and multivariate logistic regression analysis were used to analyze the independent risk factors of IHCA in ACS patients.3.We used random forest figure and classification and regression trees method to develop a decision tree to predict the risk of IHCA.Results1.The total number of ACS admissions across the 3 hospitals during this study period was 21 337,and 320 ACS patients experiencing CA,the IHCA incidence was 15.00‰.The mean age was 70.53 ± 11.81.Baseline ACS assessments yielded the following distribution: 37(11.6%)UA,162(50.6%)STEMI,and121(37.8%)NSTENMI,134(41.9%)ROSC and 186(58.1%)non-ROSC,68(21.2%)discharged to survivor and 252(78.8%)patients died in hospital.2.In multivariate logistic regression analysis,factors associated with ROSC included age <70,shockable rhythm,the time of ROSC ≤ 15 min and 16 ~ 30 min,PCI(P<0.05).Factors associated with survival included age <70,shockable rhythm,the time of ROSC ≤ 15 min and 16 ~ 30 min,Killip≤II,CCI≤2(P<0.05).3.MEWS,Vi EWS and NEWS were statistically significant when evaluated by ROC curve(P<0.001),and the AUC were 0.73、0.74、0.76,respectively.Among these three early warning scores,Vi EWS was the most accurate in detecting cardiac arrest.The AUC of Vi EWS in 30 min,1h,8h,16 h and 24 h prior to cardiac arrest was 0.82、0.78、0.75、0.66、0.67,respectively(P<0.05).Vi EWS score could predict the onset of cardiac arrest 8 hours before cardiac arrest with a Youden index of 0.39,at a specificity and sensitivity of 0.68,0.71,higher than other time,cutoff of 5 points.4.In multivariate logistic regression analysis,factors associated with cardiac arrest included Killip≥III,chest pain,Vi EWS≥ 5 points,SCr≥186μmol / L,BNP≥5000pg/ml,junctional arrhythmia and ventricular arrhythmia(P<0.05).5.The samples were randomly divided into training group(60%)and test group(40%).The C&RT algorithm was used to analyze the decision tree model of cardiac arrest in ACS patients.The decision tree model consisted of 4 strata,with10 nodes.Four explanatory variables were screened out from the prediction model,including arrhythmia,Vi EWS,BNP and platelet count.The risk statistic of misclassification probability of the model was 0.14,and AUC was 0.86 which had a statistically significant difference(P<0.01),suggesting that the classification tree model for predicting cardiac arrest fitted the actuality very well.Conclusions1.The incidence of IHCA,ROSC,and survival to discharge were higher in ACS patients than other diseases.Age <70,shockable rhythm,the time of ROSC ≤15min and 16~30min,PCI were the independence risk factors of ROSC(P<0.05).Factors associated with survival included age<70,shockable rhythm,the time of ROSC≤15min and16~30min,Killip≤II,CCI≤2 were the independence risk factors of survival(P<0.05).2.By 8h prior to cardiac arrest,the early warning score had significant change,Vi EWS was more accurate to detecte cardiac arrest than other early warning score.Killip≥III,chest pain,Vi EWS≥5points,SCr≥186μmol/L,BNP≥5000 pg/ml,junctional arrhythmia and ventricular arrhythmia were the independence predictors for cardiac arrest(P<0.05).3.According to the risk prediction model,the scale of cardiac arrest could be used to estimate the risk prediction of ACS patients in clinical practice.The evaluation of this model showed that the discrimination and the value of this model are preferable.