Is It Possible to Simplify Risk Stratification Scores for Patients with ST-Segment Elevation Myocardial Infarction Undergoing Primary Angioplasty?

(1)

Revista

Portuguesa

de

Cardiologia

Portuguese

Journal

of

Cardiology

www.revportcardiol.org

ORIGINAL

ARTICLE

Is

it

possible

to

simplify

risk

stratiﬁcation

scores

for

patients

with

ST-segment

elevation

myocardial

infarction

undergoing

primary

angioplasty?

Ana

Teresa

Timóteo

∗

,

Ana

Luísa

Papoila,

João

Pedro

Lopes,

José

Alberto

Oliveira,

Maria

Lurdes

Ferreira,

Rui

Cruz

Ferreira

CardiologyDepartment,SantaMartaHospital,CentroHospitalardeLisboaCentral,EPE,Lisbon,Portugal

Received7May2012;accepted25June2013 Availableonline22November2013

KEYWORDS Riskscores; Acutecoronary syndromes; Prognosis Abstract

Introduction:ThereareseveralriskscoresforstratiﬁcationofpatientswithST-segment ele-vationmyocardialinfarction(STEMI),themostwidelyusedofwhicharetheTIMIandGRACE scores.However,thesearecomplexandrequireseveralvariables.Theaimofthisstudywasto obtainareducedmodelwithfewervariablesandsimilarpredictiveanddiscriminativeability.

Methods:Westudied607patients (age62years,SD=13;76%male)whowereadmittedwith STEMIandunderwentsuccessfulprimaryangioplasty.Ourendpointswereall-causein-hospital and30-daymortality.ConsideringallvariablesfromtheTIMIandGRACEriskscores,multivariate logisticregressionmodelswereﬁttedtothedatatoidentifythevariablesthatbestpredicted death.

Results:ComparedtotheTIMIscore,theGRACEscorehadbetterpredictiveanddiscriminative performanceforin-hospitalmortality,withsimilarresultsfor30-daymortality.Afterdata mod-eling,thevariableswithhighestpredictiveabilitywereage,serumcreatinine,heart failure andtheoccurrenceofcardiacarrest.ThenewpredictivemodelwascomparedwiththeGRACE riskscore,afterinternalvalidationusing10-foldcrossvalidation.Asimilardiscriminative per-formancewasobtainedandsomeimprovementwasachievedinestimatesofprobabilitiesof death(increasedforpatientswhodiedanddecreasedforthosewhodidnot).

Conclusion: ItispossibletosimplifyriskstratiﬁcationscoresforSTEMIandprimaryangioplasty usingonlyfourvariables(age,serumcreatinine,heartfailureandcardiacarrest).Thissimpliﬁed modelmaintainedagoodpredictiveanddiscriminativeperformanceforshort-termmortality. © 2012 SociedadePortuguesa de Cardiologia. Published by Elsevier España, S.L. All rights reserved.

∗_{Corresponding}_author.

E-mailaddress:anatimoteo@yahoo.com(A.T.Timóteo).

(2)

PALAVRAS-CHAVE Scoresderisco; Síndromascoronárias agudas;

Prognóstico

Serápossívelsimplificarosscoresdeestratificaçãoderiscoemdoentescomenfarte

agudodomiocárdiosubmetidosaangioplastiaprimária?

Resumo

Introdução:Existemváriosscoresparaestratificaçãoderiscoemdoentescomenfarteagudo miocárdicocomelevaçãosegmentoST(EAMCEST),sendoosmaisutilizadosoTIMIeoGRACE. Contudo,sãocomplexosenecessitamdeváriasvariáveisparaoseucálculo.Esteestudoteve como objetivo encontrar um modelo deprediçãode risco, com menos variáveise idêntica capacidadepreditiva/discriminativa.

Métodos: Estudaram-se607doentes(62anos,SD=13;76%sexomasculino),admitidospor EAM-CESTesubmetidosaangioplastiaprimáriacomsucesso.Consideraram-secomooutcomes,a mortalidadeintra-hospitalareaos30diasdeseguimento.Paraidentiﬁcarquaisdasvariáveis dosreferidosscoresserevelarammaisinﬂuentesnaprevisãodamortalidade,foramefetuadas análisesderegressãologística.

Resultados: Dos dois scores clássicos, o GRACE foi o que apresentou melhor capacidade preditiva/discriminativaparaamortalidadeintra-hospitalar,comresultadossemelhantespara amortalidadeaos30dias.Naconstruçãodomodeloreduzido,asvariáveisselecionadasforam a idade, a creatinina, a insuficiência cardíaca e a paragem cardíaca. As estimativas das probabilidadesdemorte intra-hospitalar,obtidas atravésdevalidaçãocruzada, foram com-paradas com as do modelo originaldo score GRACE. As capacidades discriminativas foram idênticastendoaindasidoobtidaalgumamelhorianasestimativasdasprobabilidadesdemorte (aumento/diminuiçãoparaosdoentesquemorreram/nãomorreram).

Conclusões:ÉpossívelumasimplificaçãodosscoresdeestratificaçãoderiscoparaEAMCESTe angioplastiaprimáriacomapenasasvariáveisidade,creatinina,insuficiênciacardíacae para-gemcardíaca.Omodelosimplificadomanteveumbomdesempenhopreditivo/discriminativo paraamortalidadeacurto-prazo.

Listofabbreviations

ACEI angiotensin-convertingenzymeinhibitor ACS acutecoronarysyndrome

AUC areaunderthecurve

CABG coronaryarterybypassgrafting CI conﬁdenceinterval

GRACE GlobalRegistryofAcuteCoronaryEvents HL Hosmer-Lemeshow

ICU intensivecareunit

IDI integrateddiscriminationimprovement NRI netreclassiﬁcationimprovement OR oddsratio

PCI percutaneouscoronaryintervention ROC receiveroperatingcharacteristic SD standarddeviation

STEMI ST-elevationmyocardialinfarction TIMI ThrombolysisInMyocardialInfarction

Introduction

Patients with ST-segment elevation myocardial infarction (STEMI)areatincreasedriskofcardiovascularevents, par-ticularlydeath, in both short-andlong-term follow-up.1,2

In these patients, early risk stratiﬁcation plays a central role, as the beneﬁts of newer and more aggressive and costlytreatment strategiesseemtobeproportionaltothe

risk of adverse clinical events. Different scores are now availablebasedoninitialclinicalhistory,electrocardiogram, andlaboratory tests,whichenableearlyriskstratiﬁcation on admission. The Thrombolysis In Myocardial Infarction (TIMI) scorewas developed using thedatabase of alarge clinical trial (Intravenous nPA for Treatment of Infarcting Myocardium EarlyII).3 _The _more_recent_Global _Registry_of

AcuteCoronaryEvents(GRACE)scorewasbasedonthe reg-istryofthesamename,inapopulationofpatientsacrossthe entirespectrumofacutecoronarysyndromes(ACS).4_Both

scoresweredevelopedforshort-termprognosis.

Although percutaneous coronary intervention (PCI) has signiﬁcantlyimproved theoutcomeofpatientswithSTEMI comparedtoﬁbrinolytictreatment, high-risk patientsstill have considerablemortalityand morbidityand implemen-tationofnewtreatmentmodalitiesishighlydesirable.1,5,6

Thereis noagreementonhowtodeﬁnehigh-risk patients with STEMI; different studies used different clinical def-initions and scores and there is no unanimity on which deﬁnitionorscoreshouldbeusedtoidentifyapatient’srisk category.1,7---10

Theaimofthisstudy wastocomparetheperformance oftheTIMIandGRACEscoresforriskstratiﬁcationofSTEMI patients,andtoassessthefeasibilityofdevelopingasimpler modelwithsimilarpredictiveperformance.

Methods

Thisisaretrospectivestudyofconsecutivepatients admit-tedtoourintensivecareunit(ICU)withadiagnosisofSTEMI

(3)

between January 2005 andDecember 2009. Although this wasaretrospectivestudy,datawerecollectedprospectively andrecordedinacomputerdatabaseofACSpatients admit-tedtoourinstitution’sICU(single-centerACSregistry).The inclusion criteria were a history of chest pain at rest or othersymptomssuggestiveofanACS,withthemostrecent episodeoccurringwithin12hoursofadmission,associated with at least 1-mm ST-segmentelevation in twoor more contiguousleadsoratleast2-mmelevationinleadsV1-V4 orneworpresumablynewleftbundlebranchblockonthe electrocardiogramandserialincreasesinserumbiochemical markersofcardiacnecrosis.Theregistryenrolledpatients treated withrescueand primaryPCI, but for thepurpose of this study we included only patients who underwent successfulprimaryangioplasty(post-proceduralTIMIgrade >2 flow and<10% diameter stenosis in the infarct-related artery)(98%ofallSTEMIpatientsincludedinourregistry). Thebiomarker usedwascardiactroponinI,witha thresh-old for positivity of 0.06 ng/ml. We evaluated patients’ demographiccharacteristics,riskfactorsforcoronaryartery disease,previouscardiachistory,laboratorydata,vitalsigns onadmissionandin-hospitaltreatment.Hypertension, dia-betesandhyperlipidemiaweredefinedaseitherpreviously knownorunderspecifictherapy.Foreachpatient,a numeri-calclassificationaccordingtotheTIMIandGRACEriskscores wascalculatedfromtheinitialclinicalhistory, electrocar-diogramandlaboratoryvaluescollectedonadmission.3,4

The study endpoint wasall-cause mortality in-hospital andat30days.Follow-upwasobtainedinallpatientswho survived todischarge by telephonecontact 30 days after admissionbyadedicatedfollow-upteam.Vitalstatuswas availablefor99.6%ofpatients.

The study complies with the ethical guidelines of the Declaration of Helsinki andwritten informed consentwas obtainedfromallsubjects.

Statisticalanalysis

An exploratory analysis was carried out for all variables. Categorical variables were presented as percentages and continuousvariablesasmeanandstandarddeviation(SD).

AllvariablesusedintheTIMIandGRACEriskscoreswere consideredinmultivariateanalysisandanewreducedmodel wasobtainedwiththevariables thatattainedp<0.05in a backwardstepwiselogistic regressionanalysis.Inorderto compare the classical GRACE risk score with the reduced model,GRACEscoreestimatesofprobabilitiesofdeathwere calculatedusingGranger’smodelforin-hospitalmortality4

anda10-foldcrossvalidationwasusedtoestimatethe prob-abilitiesofdeathbythereducedmodel.

Predictive anddiscriminativeabilitieswereassessedby theHosmer-Lemeshow(HL)testandbytheareaunderthe receiveroperatingcharacteristiccurve(AUC),respectively. ThemethoddescribedbyDeLongetal.wasusedtocompare AUCsfromeachofthesemodels.11

Continuous net reclassiﬁcation improvement (NRI) and integrateddiscriminationimprovement(IDI)werealso cal-culated. Thenet proportionof patientswithevents, with higherprobabilitiesofdeath(NRIevents)andofpatients

with-outevents,withlowerprobabilitiesofdeath(NRInonevents),

were calculated using both models. The NRI is the sum

of NRIevents and NRInonevents and quantiﬁes the

correct-nessofupwardanddownwardreclassiﬁcationofpredicted probabilities.12 _The _IDI _is _a _measure _of _improvement

in prediction and may be viewed as the difference

between improvement in average sensitivity and in aver-age1-speciﬁcity.12 _To _complement_this_comparative_study,

predictiveness curves were also calculated.13 _The

inter-pretation of this curve is straightforward: a marker (or a model)thatis uselessassigns equalrisk toallindividuals, andhence,thecorrespondingpredictivenesscurveisa hori-zontalline atthe prevalenceofthe disease;ontheother hand, a marker (or a model) that is highly informative aboutrisk yieldsa predictivecurvethatis closetoa step function.

Two-tailed tests of significance are reported. For all comparisons, a p value of <0.05 was considered statisti-callysignificant.Whenappropriate,confidenceintervals(CI) werecalculated,witha95%confidencelevel.

The statistical analysiswas carried out usingIBM SPSS Statistics,version19.0(IBMCorp.,NorthCastle,NewYork, USA)andRsoftware.14

Results

Atotal of 607 consecutive patients were included in this study,mainlymale (76%),with amean age of 62(SD=13) years. The baseline characteristics and hospital manage-ment are presented in Table 1. During hospital stay, 33 patients died (5.4%). At 30-day follow-up, there were 38 deaths(6.3%).

BothTIMIandGRACEscoresshowedgooddiscriminatory ability regarding mortality in short-term follow-up. How-ever, the performance of the TIMI score wasnot asgood astheGRACEscore:itsgoodnessof ﬁt(predictiveability) wasinferior,asdemonstratedbythepvaluesoftheHLtest

(Table2).Concerning30-dayfollow-up,thetwoscoreswere

verysimilarintheirpredictiveperformance,butagainthe GRACEscorewasbetteratpredictingmortality.

Inastepwisemultivariatelogisticregressionanalysis,the mostimportant independent predictors of mortalitywere age,Killipclassonadmission,renalfunctionandthe occur-renceofcardiac arrest.The selectedvariables werethen usedtobuilt a newpredictive model for both in-hospital and30-daymortality(Table3).

Compared tothe GRACE risk score,after 10-foldcross validation,the reduced model showed the same discrim-inatory ability (AUC=0.92 for both, p=0.916) (Figure 1) andbetter predictive ability evaluatedby goodness of ﬁt (p=0.138andp=0.802,respectively)andbycalibrationplots

(Figure2).

Regarding the other metrics that were used to study modelimprovement(continuousNRIandIDI),reducingthe numberofvariablesusedtocalculatetheGRACEriskscore increasedthepredictedriskofdeathin15%ofpatientswho diedandreducedthepredictedriskofdeathin49%ofthose who did not. This reclassiﬁcation resulted in an overall continuous NRI of 64%. However,the size of the changes wasonly signiﬁcant for thosewithevents (IDI=0.11),with amoderateoverallIDIof0.103.Theseresultsaredetailed

in Table 4 and show a slightly better performance for

(4)

Table 1 Baseline characteristics and hospital management. Age(SD)(years) 62(13) Male(%) 76.1 Riskfactors(%) Hypertension 61.8 Smoking 45.0 Hyperlipidemia 51.1 Diabetes 21.9 Previoushistory(%) MI 10.7 PCI 6.8 CABG 0.8 Stroke 4.8 Onadmission HR(SD)(bpm) 80(22) SystolicBP(SD)(mmHg) 130(29) Serumcreatinine(SD)(mg/dl) 1.01(0.38) Killipclass>1(%) 8.2 Medicaltreatment(%) Aspirin 98.7 Clopidogrel 98.0 ACEI 88.1 Beta-blocker 84.2 Statin 94.4 Outcome(%) In-hospitalmortality 5.4 30-daymortality 6.3

ACEI:angiotensin-convertingenzymeinhibitor;BP:blood pres-sure;CABG:coronaryarterybypassgrafting;HR:heartrate;MI: myocardialinfarction;PCI:percutaneouscoronaryintervention.

the construction of a predictiveness plot. Analysis of the twocurvesrevealedthat thereduced modelhas abetter performance,asseeninFigure3:thereducedmodelassigns lowerrisktopatientswithlowerriskpercentiles(although very slightly) and higher risk to patients with higher risk percentiles(inthiscasemoremarkedly).

Table2 Discriminativeability andgoodness ofﬁtofthe TIMIandGRACEriskscores.

TIMI GRACE In-hospitalmortality AUC(95%CI) 0.84(0.77-0.92) 0.92(0.87-0.96) HLtest(p) 0.007 0.556 30-daymortality AUC(95%CI) 0.83(0.76-0.90) 0.88(0.82-0.95) HLtest(p) 0.016 0.142

AUC:areaunderthereceiveroperatingcharacteristiccurve;CI: conﬁdenceinterval;HL:Hosmer-Lemeshow.

ROC plot 1.0 1.0 0.8 0.8 1 - Specificity Grace Reduced model 0.6 0.6 0.4 0.4 Sensitivity 0.2 0.2 0.0 0.0

Figure 1 Receiver operating characteristic curve analysis comparingtheGRACEriskscoreandthesimpliﬁedprediction model(p=0.916).

Discussion

Effectiveriskstratiﬁcation isessentialinthemanagement of patients with ACS. Even in patients with STEMI, for

Calibration plot Calibration plot

Predicted risk Predicted risk

Obser v e d r isk Obser v e d r isk 1.0 0.8 0.6 0.8 1.0 0.6 0.4 0.4 0.2 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0

Simplified model GRACE risk score

Figure2 Calibrationplotsforthesimpliﬁed modelandtheGRACEriskscore.Thediagonal lineindicatesperfectcalibration (predictedprobabilitiesofdeathequalestimatedprobabilitiesofdeath).

(5)

Table3 Multivariatelogisticregressionanalysis.

OR 95%CI p

In-hospitalmortality

Age(per10-yearincrement) 2.69 1.74-4.18 <0.001

Serumcreatinine(permg/dlincrement) 4.33 1.94-9.69 <0.001

Killipclass>1 3.44 1.28-9.23 0.014

Cardiacarrest 17.15 6.38-46.09 <0.001

30-daymortality

Age(per10-yearincrement) 2.13 1.49-3.04 <0.001

Serumcreatinine(permg/dlincrement) 3.42 1.67-7.00 0.001

Killipclass>1 3.56 1.44-8.79 0.006

Cardiacarrest 10.58 4.36-25.63 <0.001

CI:conﬁdenceinterval;OR:oddsratio.

whominitialtherapeuticoptionsarewell-deﬁned,patient risk characteristics have an impact on early therapeu-ticdecision-making,enablinginformeddecisionsregarding additional therapeutic interventions. After reperfusion treatment it is important toidentifypatients at high risk of further events, and hopefully tointervene in order to preventthoseevents. Careful attention topivotal factors that increase the risk of early mortality might illuminate theroleofsecond-tierinterventionsoradjunctive pharma-cotherapythatwouldfurtherlowerthefatalityrateofacute MI.Inaddition,riskstratiﬁcationmayoffertheopportunity toselectlow-riskpatientsfor alessaggressivestrategyas wellasforearlyhospitaldischarge.Toolsthatenhancethe clinician’sability torapidlyand accurately assessrisk are thusofsubstantialinterest.

Table 4 Statistics for model improvement (for hospital mortality)withthesimpliﬁedmodelafter10-foldcross vali-dationvs.theGRACEriskscore.

Events(n) 33 Nonevents(n) 574 ContinuousNRI(%) NRIevents 15.2(-18.0-49.3) NRInonevents 48.8(40.6-57.0) NRI 63.9(28.8-99.0) IDIstatistics IDIevents 0.1090(0.0231-0.1949) IDInonevents -0.0059(-0.0120-0.0003) IDI 0.1031(0.0170-0.1893) AUC

GRACEriskscore 0.92(0.87-0.96) Simpliﬁedmodel 0.92(0.88-0.95) pvalueforthedifference 0.916

Goodnessofﬁt(GRACErisk score)a

0.138

Goodnessofﬁt(simpliﬁed model)a

0.802

95% confidence intervals are shown in brackets. a Hosmer-Lemeshowgoodnessoffittest(pvalue).AUC:areaunderthe receiveroperatingcharacteristiccurve;IDI:integrated discrim-inationimprovement;NRI:netreclassificationimprovement.

The wide heterogeneity among clinical scores in the assessmentofpatients’riskmayberelevanttothedesign ofclinicaltrialstargetinghigh-riskpatients.Regarding exist-ingscores,ithasbeenshownthatonlyafewvariablesare neededtostratifypatientswithrespectto30-daymortality. However,mostscoresstillincludemorethanﬁvevariables, somewithfurthercomplexity.

Theidealscoreshouldhaveagoodbalancebetween com-plexityand utility.To beuseful in clinical practice,a risk stratiﬁcation tool should be simple and easily applied at thebedsideandshouldmake useof clinicaldatathat are routinelyavailableathospitalpresentationandoffer inde-pendentprognosticinformation.Thecomplexityofascoreis essentiallydeterminedbyfactorsrelatedtodatacollection, ratherthanthemethodologyinvolvedinthecalculation.In thisregard,theGRACEriskscorehasadvantages,asallits

1.0 1.0 0.8 0.8 0.6 0.6 Risk percentile 0.4 0.4 0.2 0.2 0.0 0.0 Predicted r isks Reduced model Predictiveness curve Grace

Figure 3 Predictiveness curves comparing the GRACE risk scoreandthereducedprediction model.The reducedmodel correctly assigns lower risk to patients with lower risk per-centilesandhigherrisktopatientswithhigherriskpercentiles. The horizontal line represents the prevalence of in-hospital death(5.4%).

(6)

variablesareobjectivedata.However,whenmorevariables arerequired,thisaddscomplexitytothescore.

This single-centerstudy, basedona consecutiveSTEMI cohort, conﬁrmed that the GRACE risk score is in some respects superior to the TIMI score in the estimation of short-term prognosis. This may be explained by changes in the treatment of STEMI patients over time. Modest increases in the use of aspirin, clopidogrel, glycoprotein IIb/IIIainhibitorsandlowmolecularweightheparin(instead ofunfractionatedheparin)havebeenseeninrecentyears, particularlyin STEMIpatients, likelyinpartaccounted for by an increase in the use of primary PCI instead of ﬁbri-nolysis as reperfusion therapy.15 _There _has _also _been _a

downwardtrend in the frequency of bleeding16_: _although

thefrequencyofmajorbleedingwouldbeexpectedtohave increasedover timeas a resultof more potent combina-tionsofantithrombotictherapiesandgreateruseofcardiac catheterization and PCI, the opposite is found. This may bedue to changesin contemporary clinical practice,and may have implications for mortality. Nevertheless, when majorbleedingoccurs, it isassociatedwithagreater risk of death. The TIMI score for STEMI was developed for risk stratiﬁcation in the era of thrombolytic reperfusion therapy3 _and _not _in _the _context _of _contemporary

treat-ment,inwhichPCI(mechanicalreperfusion)isthemainstay oftreatmentandnewantithromboticagentsareavailable. This may be why the TIMI score showedlower predictive accuracy.

Ontheotherhand,althoughtheGRACEdatabaseismore recent,only 74.2%of itsSTEMIpatients presentingtothe hospitalwithin12hoursofsymptomonsetreceived reper-fusiontherapyand,ofthese,only30.2%underwentprimary PCI.15 _A _third _of _patients _who _appeared _to _be _otherwise

eligiblereceived no reperfusion therapy and only a third receivedsuch therapy withinthe guideline-recommended targetsof30and90minutesfor thrombolysisandprimary PCI, respectively.15 _The _GRACE _registry _also _showed _that

patientsadmittedtoahospitalwithcatheterization facil-itiesweremorelikelytoundergointervention,whichmay explainthe higherrisk ofdeath withinsix months of dis-chargeobservedin thesepatients;thismaybepartlydue tothebias ofreferral of patientsin worse clinical condi-tiontotertiarycenters,butalsotohazardsrelatedtothe intervention.15,17

The relative performance of the different risk scores canalso beexplainedby their composition.The TIMI risk score includes age as a categorical variable, unlike the GRACEscore.Thisapproachmaymissthecontinuous pro-gnostic value observed over the entire spectrum of age. The TIMI risk scorealso does not take into accountrenal function,whichisoneofthemostimportantpredictorsof outcome.

Inourpopulation, multivariatelogisticregression

anal-ysis identiﬁed age, as a continuous variable, heart

failure on admission and baseline serum creatinine as the most signiﬁcant predictors of prognosis, together

with cardiac arrest. Serum creatinine has only more

recently been identiﬁed as a powerful risk variable, as it was not generally included in ACS databases before 2000.18

Afteraninternal 10-foldcrossvalidation,itwasshown thatitispossibletoobtainasimpliﬁedreducedprediction

model with only four variables that maintains the same discriminative abilityandwitha slightlybetter predictive performance.

Inthepresentstudy,othertraditionalriskmarkerswere notsigniﬁcantpredictorsofoutcome.This mayberelated

to the more aggressive invasive management adopted

in these high-risk patients, since all patients underwent successful primary angioplasty in the ﬁrst 12 hours of STEMI.Bycontrast,studiesthatdemonstratedthe progno-stic value of other variables such as cardiac biomarkers used peak instead of baseline levels as analyzed in our study.

We chose to study this speciﬁc patient population

because of the considerable heterogeneity found in ACS populations, which differ in type of medical treatment, use of PCI or ﬁbrinolysis, and mortality. These differ-ences make it difﬁcult to determine the applicability of risk scores to these differing populations. STEMI patients are a high-risk cohort of ACS patients, with the high-est mortality. Weanalyzed patients undergoingsuccessful primary PCI to obtain a more homogeneous population, since other patients have different baseline

characteris-tics and different treatment management and a worse

prognosis. For this reason, our study evaluated only pri-mary PCIpatients, reﬂectingthecontemporarytreatment

of STEMI in a more homogeneous population. In these

patients, risk stratiﬁcation is more important, particu-larly in the selection of new therapies to improve their prognosis, which is ominous, with in-hospital mortality of>5%.15

Limitations

This is arelatively smallsample, andthe factthatit is a single-centerstudymaylimitourconclusions.Moreover,the results onlyapplytoSTEMIpatientsundergoingsuccessful primaryPCI.

BecausetheGRACEriskscorewasdevelopedfor predic-tionofhospitalmortality,nofurthercomparisonsweremade regarding30-daymortality.

Itisalsoimportanttodeterminewhetherthevalidityof thisnewmodelseenforshort-termfollow-upismaintained inmedium-termfollow-up(oneyear).

Conclusions

Oftheclassicalscoresavailableforriskstratificationafter STEMIandinthecontemporaryeraoftreatment,theGRACE riskscorehasthehighestpredictiveaccuracy.Itis,however, possibletosimplifycurrentscoresafterSTEMIandprimary angioplastywithonlyfourvariables(age,serumcreatinine onadmission,heartfailureonadmissionandtheoccurrence of cardiac arrest). This simplified model maintainedgood predictiveabilityforshort-termmortalitycomparedtomore complexriskscores.Duetoourrelativelysmallsample, it wasnotpossibletoconstructanewscorebasedonthe pro-posed simplified model. However,this study supports the possibilitythatsuchasimplifiedscorecouldbeconstructed ifamuchlargersample(suchasfromalargeregistry)were used.

(7)

Ethical

disclosures

Protection of human and animal subjects.The authors declarethatnoexperimentswereperformedonhumansor animalsforthisstudy.

Conﬁdentialityofdata.Theauthorsdeclarethattheyhave followedtheprotocolsoftheirworkcenteronthe publica-tionofpatientdataandthatallthepatientsincludedinthe studyreceivedsufﬁcientinformationandgavetheirwritten informedconsenttoparticipateinthestudy.

Righttoprivacyandinformedconsent.The authorshave obtained the written informedconsent ofthe patients or subjectsmentionedinthearticle.Thecorrespondingauthor isinpossessionofthisdocument.

Conﬂicts

of

interests

Theauthorshavenoconﬂictsofinteresttodeclare.

References

1.AntoniucciD,MiglioriniA,ParodiG,etal.Abciximab-supported infarct artery stent implantation for acute myocardial infarction and long-term survival: a prospective, multi-center, randomized trial comparing infarct artery stenting plus abciximab with stenting alone. Circulation. 2004;109: 1704---6.

2.Steg PG, Goldberg RJ, Gore JM, et al. Baseline character-istics, management practices, and in-hospital outcomes of patients hospitalized withacute coronary syndromes in the GlobalRegistryofAcuteCoronaryEvents(GRACE).AmJCardiol. 2002;90:358---63.

3.MorrowDA,AntmanEM,CharlesworthA,etal.TIMIriskscore forST-elevationmyocardialinfarction:aconvenient,bedside, clinicalscoreforriskassessmentatpresentation.Circulation. 2000;102:2031---7.

4.GrangerCB,GoldbergRJ,DabbousO,etal.Predictorsof hos-pitalmortalityintheGlobalRegistryofAcuteCoronaryEvents. ArchInternMed.2003;163:2345---53.

5.Andersen HR, Nielsen TT, Rasmussen K, et al. A com-parison of coronary angioplasty with ﬁbrinolytic therapy in acute myocardial infarction. N Engl J Med. 2003;349: 733---42.

6.GrinesCL,BrowneKF,MarcoJ,etal.Acomparisonof immedi-ateangioplastywiththrombolytictherapyforacutemyocardial infarction. The Primary Angioplasty in Myocardial Infarction StudyGroup.NEnglJMed.1993;328:673---9.

7.BrodieBR,HansenC,StuckeyTD,etal.Door-to-balloontime with primary percutaneous coronary intervention for acute myocardialinfarction impactslate cardiacmortalityin high-riskpatientsand patientspresenting earlyaftertheonsetof symptoms.JAmCollCardiol.2006;47:289---95.

8.McNamaraRL,WangY,HerrinJ,etal.Effectofdoor-to-balloon timeonmortalityinpatientswithSTsegmentelevation myocar-dialinfarction.JAmCollCardiol.2006;47:2180---6.

9.WivviottSD,MorrowDA,FrederickPD,etal.Performanceofthe thrombolysisinmyocardialinfarctionriskindexintheNational RegistryofMyocardialInfarction-3and-4:asimpleindexthat predictsmortalityinST-segmentelevationmyocardial infarc-tion.JAmCollCardiol.2004;44:783---9.

10.Addala S,Grines CL,DixonSR,et al.Predictingmortalityin patientswithST-elevationmyocardialinfarction treatedwith primarypercutaneouscoronaryintervention(PAMIriskscore). AmJCardiol.2004;93:629---32.

11.DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areasundertwoormorecorrelatedreceiveroperating char-acteristic curves: a nonparametric approach. Biometrics. 1988;44:837---45.

12.PencinaMJ,D’AgostinoRB,SteyerbergEW.Extensions ofnet reclassiﬁcation improvement calculations tomeasure useful-nessofnewbiomarkers.StatMed.2011;30:11---21.

13.HuangY,PepeMS,FengZ.Evaluatingthepredictivenessofa continuousmarker.Biometrics.2007;63:1181---8.

14.R Development Core Team. R: A language and environ-mentforstatisticalcomputing.Vienna,Austria:Rfoundation for Statistical Computing; 2008. ISBN 3-900051-07-0, URL http://www.R-project.org

15.Goodman SG, Huang W, YanAT, et al. Theexpanded Global Registry of Acute Coronary Events: baseline characteristics, managementpractices,andhospitaloutcomesofpatientswith acutecoronarysyndromes.AmHeartJ.2009;158:193---201. 16.Fox KA, Carruthers K, Steg G, et al. Has the frequency of

bleeding changed over time for patients presenting withan acutecoronarysyndrome?TheGlobalRegistryofAcute Coro-naryEvents.EurHeartJ.2010;31:667---75.

17.VandeWerfF,GoreJM,AvezumA.Accesstocatheterisation facilitiesinpatientsadmittedwithacutecoronarysyndrome: multinationalregistrystudy.BMJ.2005;330:441---6.

18.AlSuwaidiJ,ReddanDN,WilliamsK,etal.Prognostic implica-tionsofabnormalitiesinrenalfunctioninpatientswithacute coronarysyndromes.Circulation.2002;106:974---80.