| BACKGROUNDSAcute chest pain is one of most common causes for visits to emergency department(ED)throughout the world.Furthermore,the cause of chest pain is extremely heterogeneous with a wide spectrum of conditions ranging from high-risk diseases(such as acute coronary syndrome(ACS)and aortic dissection)to low-risk problems(such as gastroesophageal reflux and intercostal neuralgia).The ED is an important battlefield for the treatment of patients with acute chest pain It is necessary to evaluate the patients correctly and select different triage directions according to risk stratification to ensure the timely treatment of high-risk patients and to avoid excessive examination of low-risk patients.In North America and Europe,the inappropriate discharge of patients with ACS from ED is estimated to be between 2%and 5.3%,and among these patients,mortality is nearly two times greater as compared with admitted patients.On the other hand,60%-80%acute chest pain is not due to ACS and 50%-67.2%have non-cardiac conditions,which do not require unnecessary diagnostic tests and prolonged observation.Therefore,although high-risk patients should not be missed,it is unnecessary to recognize all patients with chest pain as high-risk and prolong the period of observation or conduct excessive examination.Low-risk patients who are suitable for early discharged should be identified as many as possible to reduce emergency room congestion and resource wasteThe rate of missed diagnosis of high-risk chest pain reflects the safety of risk stratification,of which less than 1%is the internationally recognized threshold value.In other words,the sensitivity of risk assessment strategy should be more than 99%.Meanwhile,the proportion of patients identified as low-risk refers to the effectiveness of risk stratification.Among them,the security is more importantIn the initial risk assessment of acute chest pain,risk score is the most widely used objective tool in clinical practice for the easy accessibility,sample calculation and quantifiable performance.The common risk scores include the Global Registry of Acute Coronary Events(GRACE)score,the Thrombolysis in Myocardial Infarction(TIMI)score,the History,Electrocardiogram,Age,Risk factors,Troponin(HEART)score and the Emergency Department Assessment of Chest Score(ED ACS)score.The accelerated diagnostic protocols(ADPs),integrating these scores with initial electrocardiograph(ECG)and serial cardiac troponin(cTn)detections,could reach better safety and effectiveness.These risk stratification models were all derived from foreign patients with chest pain or ACS,and have showed excellent performance in the derivation cohort and multiple validation cohort.However,the safety and effectiveness of the above risk assessment models in Chinese chest pain patients have not been verified,so they cannot be truly applied to the emergency clinic in China.Furthermore,some models still have undefined problems.First,there are six different calculation models of GRACE score to predict different outcomes at different time periods in ACS patients,but it remains unknown which model is more suitable for risk stratification in patients with acute chest pain.Second,although the symptomatological variable in the HEART score has considered the characteristics of all aspects of comprehensive symptoms,they still failed to specify which characteristics are typical and which are non-specific manifestations of myocardial ischemia,and failed to prove which combination is more reasonable and effective.In addition,due to the disparities of disease composition,medical system and the variable access,it is hard for the existing prediction models to work well in any external cohort.Developing specific models for specific people is the best way to solve real problems.Therefore,to create a new risk assessment model based on information from Chinese patients with chest pain will be more conducive to optimize the risk stratification and scientific management of patients with acute chest pain.As a result,we established the regional representative cohort of Chinese patients with chest pain,to evaluate and compare the performance of existing models in risk stratification of Chinese patients with chest pain,to explore the better combination of symptomatological variables,and to develop a new assessment model according to the characteristics of Chinese patients.All of these efforts were aimed to improve the safety and effectiveness of patient triage and to optimize the prognosis of patients and medical resource allocation.METHODSStudy DesignThis study was a prospective cohort study,which was divided into two parts.The first one was for evaluating and comparing the risk stratification ability of the existing models in the chest pain patients in China.The data came from the Evaluation and Management of Patients with Acute ChesT pain in China(EMPACT)cohort,which was onducted in 21 representative public hospitals in Shandong province.Patients were enrolled from August 2015 to September 2017.The second part was to derive a new risk assessment model in the EMPACT cohort,and validate it in the external cohort from 6 grade III teaching hospitals outside Shandong province.This study has been approved by the ethics committee at these participating hospitals.Written informed consent was obtained from all participants.Inclusion and Exclusion CriteriaCohort inclusion criteria:?over 18 years of age;?presenting to the EDs;?acute chest pain or ACS related symptoms;?symptoms occurring within 24 hours;and?informed consent signed.Cohort exclusion criteria:?chest pain caused by trauma;?persistent or recurrent chest pain caused by rheumatism or cancer;?transfer from other grade II or III hospitals;and ?re-visit within 30 days of enrolment.Exclusion criteria for data analys:is:?missing cTn,ECG and vital signs;?lost to follow-up;?diagnosed as ST-segment elevation myocardial infarction(STEMI).Data Collection and Quality ControlSubjects with acute chest pain and ACS-related symptoms were directly identified by trained research assistants and enrolled when they present to the ED.Research assistants screened all the ED visits on a daily basis to capture the eligible patients consecutively where possible.Data collection was prospectively conducted on a standardized case report form(CRF),in which the variables have been developed by the steering committee in accordance with the international standards.Patients were followed up through telephone 30 days after enrollment.Through continuous registration quality control,on-site training,on-site verification,on-line and off-line central data monitoring,etc.,the progress and quality of data collection were under real-time quality control.Risk Stratification ModelsThe GRACE,TIMI,HEART and ED ACS scores were evaluated.The ADAPT-ADP,mADAPT-ADP,EDACS-ADP and HEART-Pathway were assessed based on one or two tests of cTn.Low-risk cutoff:GRACE<109,TIMI=0,HEART<3,EDACS<16;ADAPT-ADP:TIMI=0+cTn(-)+ECG(-);mADAPT-ADP:TIMI?1+cTn(-)+ECG(-);EDACS-ADP:EDACS<16+cTn(-)+ECG(-).Outcomes and AdjudicationThe primary outcome was the composite endpoint of MACEs within 30 days,including death from all causes,acute myocardial infarction(AMI),emergency revascularization,cardiac arrest and cardiogenic shock.Two senor cardiologists from the clinical events committee adjudicated the MACEs independently using all available clinical records,and discrepancies were evaluated by a third senior physician.If patients were lost to follow-up,a local death registry was used to supplement the survival status.Statistical AnalysisContinuous variables are presented as the mean(standard deviation),and categorical variables are presented as the number of cases(percentage).Baseline characteristics between groups were compared using t tests for continuous variables and chi-square(?2)tests for categorical variables.Pearson product-moment correlation was used as "r" to describe the direction and quantify the strength of the linear association between scores and the incidence of MACEs in individuals with chest pain.The calibration was evaluated using the Hosmer-Lemeshow goodness-of-fit test(HLT).Discrimination of scores was assessed by the area under the curve(AUC)of receiver operating characteristic(ROC)curves.Reclassification was performed to assess how well a risk score improved predictions compared with another one based on category-free net reclassification improvement(NRI)and absolute integrated discrimination improvement(IDI).Diagnostic accuracy with 95%confidence intervals(CIs)of the different scores was determined,including sensitivity,specificity,negative predictive value(NPV)and positive predictive value(PPV).Derivation of the new chest pain symptom score(nCPSS):through two-stage multivariable logistic regression,all aspects of the chest pain characteristics were effectively integrated into the nCPSS score.Through variable exploration and multivariable logistic regression analysis,the independent predictors of 30-day MACE were determined,and the weight of each variable was assigned according to ? value to form the EMPACT score.The stability of the model was verified internally by 10-fold cross validation.Sensitivity and proportion of patients identified as low-risk with different EMPACT cutoffs were calculated and compared with those of GRACE,TIMI and HEART scores.The performance of the EMPACT score was external validated in the validation cohort.A P value of less than 0.05(two-sided significance testing)was considered statistically significant in the analysis.All statistical analyses were performed using SAS V.9.4(SAS Institute Inc.,Cary,North Carolina,USA)or MedCalc V.18.11.3(MedCalc Software,Ostend,B elgium).RESULTSPatientsFrom August 2015 to September 2017,a total of 13918 patients with acute chest pain presented in the EDs of 21 participating hospitals in Shandong Province,of which 5569 patients did not meet the inclusion criteria and 8349 patients were enrolled in the derivation cohort.There were 1877 patients excluded for lack of information for calculating scores,including the result of cTn tests,ECG,vital signs or lost to follow-up.Additionally,1573 patients were excluded for diagnosis of STEMI.Finally,4899 patients entered the data analysis process.OutcomesThere were 1096(22.4%)chest pain patients with adjudicated MACEs in 30 days after presentation,including 103 patients(2.1%)who died from all causes,997(20.4%)with index AMI,27(0.6%)with subsequent AMI,83(1.7%)who underwent emergency revascularization,66(1.3%)who experienced cardiogenic shock and 62(1.3%)who experienced cardiac arrest.Overall Validation and Comparison of Current ModelsThe GRACE(IHDthMI),TIMI,HEART and EDACS scores exhibited very strong linear relationships with the actual MACE rates in patients with undifferentiated chest pain,(r 0.94?0.9 7,P<0.001).The model goodness-of-fit of the HEART score was better(P=0.676).The AUC of the GRACE(IHDthMI)for predicting 30-day MACE was 0.81(0.80,0.82),superior to other GRACE models(P<0.001).The AUCs of the HEART,TIMI and EDACS were 0.78(0.77,0.80),0.68(0.66,0.69)and 0.62(0.60,0.63),respectively.The differences of AUCs between GRACE(IHDthMI)and these models were statistically significant(P<0.001).The GRACE(IHDthMI)score had better risk reclassifications than other scores(positive NRI and IDI,P<0.001).Safety and Effectiveness of Current ModelsThe GRACE(IHDthMI)?109 identified 2064(42%)patients as low-risk with a sensitivity of 0.900(0.880,0.917);The TIMI=0 identified 535(11%)patients as low-risk with a sensitivity of 0.980(0.970,0.987);The HEART<3 identified 957(20%)patients as low-risk with a sensitivity of 0.965(0.953,0.975).A GRACE(IHDthMI)<56 or HEART<1 achieved the sensitivity of>99%,but the proportion of patients classified as low-risk decreased significantly(7%and 3%).The ADAPT-ADP,mADAPT-ADP,EDACS-ADP and HEART-Pathway based on the result of initial cTn test still failed to achieve sensitivity?99%.In patients with a series of cTn tests(n=1824),the sensitivity of each ADP based on 2 cTn results in identifying low-risk patients reached>99%,but the proportion of low-risk patients was very low(5%?11%).Development and Assessment of the nCPSSSeven aspects of the chest pain symptoms were effectively integrated into a nCPSS score after two stages of multivariable logistic regression analysis.Assignments of variables were as follows:Characters:squeezing/crushing/heaviness/burning/distending/aching(1),stabbing(-2);Location:substemal(1);Severity:heavy pain(2);Duration:persistent or>20 min(2);Associated symptoms:sweating(2),vomiting(2),dyspnea(1),dizziness(-3),palpitations(-1);Precipitating factors:mood(-2),deep inspiratory/cough(-6);Relieving factors:no relief(1).Significantly positive linear correlation was presented between nCPSS and MACE incidence(r=0.952,P<0.001).The AUC of nCPSS to predict MACE was 0.70(0.68,0.71),which was significantly better than the ED ACS score and the history variable of the HEART score(P<0.001).Development and Validation of the EMPACT scoreAfter multivariable logistic regression analysis,the variables of the EMPACT model,including age,sex,risk factors,nCPSS,ECG and cTn,were proven to be independent predictors for 30-day MACE.The weights of each variable were:age(1),male(2),nCPSS(1),number of risk factors(1),ECG new ischemia(3),and positive first cTn(10).In the derivation cohort,the EMPACT score showed strong positive linear correlation with MACE incidence(r=0.949,P<0.001)with an AUC of 0.88(0.87,0.89),which was significantly better than the GRACE(IHDthMI),HEART and TIMI scores(P<0.001).After removing the ECG and cTn variables,the EMPACT score still outperformed the EDACS score(P<0.001).An EMPACT ?4 identified 677(14%)patients as low-risk with a sensitivity of 0.994(0.987,0.997).The EMPACT<5+first ECG(-)+first cTn(-)identified 866(18%)patients as low-risk with a sensitivity of 0.991(0.983,0.996).At the same sensitivity level(99%,98%and 95%),specificity and the proportion of patients classified as low-risk by the EMPACT score were significantly higher than those of other models.In the validation cohort(n=758),the AUC of EMPACT was 0.87(0.85,0.89).An EMPACT<4 identified 101(13%)patients as low-risk with a sensitivity of 1.000(0.978,1.000).CONCLUSIONS In the Chinese chest pain cohort,the GRACE and HEART scores showed good overall discrimination between high-risk and low-risk patients,among which GRACE(IHDthMI)was the best.The widely accepted risk stratification scores or the ADPs could not achieve safe and effective identification of low-risk chest pain patients in Chinese population.The nCPSS,an organic combination of complex chest pain symptomatology,outperformed the EDACS score and the history variable of the HEART score.Moreover,the nCPSS was parallel to other variables(demography,risk factors,ECG and cTn)in the final risk assessment model and could fully reflect the independence and predictive advantages of symptomatology.Based on the Chinese chest pain patient cohort,the EMPACT score exhibited excellent overall discrimination of high vs low-risk acute chest pain,which was significantly better than the GRACE,HEART and TIMI scores.An EMPACT achieved a sensitivity of>99%,and the proportion of patients identified as low-risk was significantly higher than those of other models. |
PDF Full Text Request