Discordant Cardiac Biomarkers: Frequency and Outcomes in Emergency Department Patients With Chest Pain

CARDIOLOGY AND VASCULAR/ORIGINAL RESEARCH

Discordant Cardiac Biomarkers: Frequency and Outcomes inEmergency Department Patients With Chest Pain

Alan B. Storrow, MDChristopher J. Lindsell, PhDJin H. Han, MDCorey M. Slovis, MDKaren F. Miller, RNW. Brian Gibler, MDJames W. Hoekstra, MDW. Franklin Peacock, MDJudd E. Hollander, MDCharles V. Pollack, Jr, MD, MAon behalf of the EMCREG-

i*trACS Investigators

From the Department of Emergency Medicine, Vanderbilt University Medical Center, Nashville, TN(Storrow, Han, Slovis, Miller); the Department of Emergency Medicine, University of Cincinnati,Cincinnati, OH (Lindsell, Gibler); the Department of Emergency Medicine, Wake Forest University,Winston-Salem, NC (Hoekstra); the Department of Emergency Medicine, The Cleveland ClinicFoundation, Cleveland, OH (Peacock); and the Department of Emergency Medicine, University ofPennsylvania, Philadelphia, PA (Hollander, Pollack).

Study objective: We evaluate associations between pairs of discordant cardiac biomarkers (positiveMB band of creatine kinase [CKMB] with negative creatine kinase, positive CKMB with negativecardiac troponin, and positive troponin with negative CKMB) and the presence of acute coronarysyndromes in emergency department (ED) chest pain patients.

Methods: This was a secondary analysis of a prospective registry. Data were obtained from themulticenter Internet Tracking Registry of Acute Coronary Syndromes, which included 17,713 ED visitsfor possible acute coronary syndrome between June 1999 and August 2001. First visits and first EDcardiac biomarker results from the 9 sites, 8 in the United States and 1 in Singapore, were included.Subjects were excluded for incomplete information or an initial ECG consistent with ST-segmentelevation myocardial infarction. Acute coronary syndrome was defined by diagnosis-related groupcode indicating myocardial infarction, positive invasive or noninvasive diagnostic testing,revascularization, or death during hospitalization or within 30 days.

Results: Of 8,769 eligible patients, 1,614 (18.4%) had acute coronary syndrome. The CKMB andcardiac troponin results were discordant in 7% of patients (CKMB�/cardiac troponin–, 4.9%,CKMB–/cardiac troponin� 2.1%), whereas increased CKMB with normal creatine kinase levelsoccurred in 239 (3.1%) patients. The unadjusted odds ratios with 95% confidence intervals for acutecoronary syndrome in patients with and without discordant markers were: CKMB�/CK� 5.7 (4.4-7.4), CKMB�/CK� 4.4 (3.6-5.2), CKMB�/cTn� 4.8 (3.4-6.8), CKMB�/cTn� 2.2 (1.7-2.8),CKMB�/cTn� 26.6 (18.0-39.3). For the group with cardiac troponin, the reference category wasnegative troponin and negative CKMB; for the group with creatine kinase, the reference category wasnegative CKMB but either a positive or negative creatine kinase.

Conclusion: Among the spectrum of ED patients with chest pain, an increased CKMB level with a normalcreatine kinase level identifies patients at increased risk for acute coronary syndrome. Similarly, anincreased troponin level regardless of CKMB level and an increased CKMB level regardless of troponinlevel identify patients at higher risk for acute coronary syndrome than those with uniformly normal cardiacbiomarker levels. Our data suggest that discordant cardiac biomarkers may identify patients at increasedrisk for acute coronary syndrome. [Ann Emerg Med. 2006;48:660-665.]

0196-0644/$-see front matterCopyright © 2006 by the American College of Emergency Physicians.doi:10.1016/j.annemergmed.2006.05.016

660 Annals of Emergency Medicine Volume , . : December

Storrow et al Discordant Cardiac Biomarkers

INTRODUCTIONBackground

In the emergency department (ED) evaluation of possibleacute coronary syndromes, it is routine to assay the serum levelsof cardiac biomarkers: creatine kinase, the MB band of creatinekinase (CKMB), and the cardiac troponins I or T. Thefrequency and implications of various combinations of results ofthese assays, specifically increased CKMB with normal creatinekinase and discordant CKMB and cardiac troponin in acutecoronary syndrome, have recently been reported in patients athigh risk for acute coronary syndrome.1,2 In the former, anincreased CKMB, regardless of the level of total creatine kinase,was a strong predictor of 180-day death or myocardialinfarction in nearly 20,000 patients from 4 large acute coronarysyndrome trials (The Global Use of Strategies to OpenOccluded Coronary Arteries IIb [GUSTO IIb],3 Platelet IIb/IIIa Antagonism for the Reduction of Acute CoronarySyndrome Events in a Global Organization Network A[PARAGON A],4 Platelet IIb/IIIa Antagonist for the Reductionof Acute Coronary Syndrome Events in a Global OrganizationNetwork B [PARAGON B],5 and Platelet Glycoprotein IIb/IIIa

Editor’s Capsule Summary

What is already known on this topicCardiac biomarkers are important in diagnosis and riskstratification of patients with suspected acute coronarysyndrome. How to interpret discordant biomarkers isunclear.

What question this study addressedAmong patients with suspected non–ST-segmentelevation acute coronary syndrome, how shouldemergency department (ED) clinicians interpret thefollowing pairs of discordant cardiac biomarkers:increased CKMB�normal creatine kinase levels,increased troponin�normal CKMB levels, and normaltroponin�increased CKMB levels?

What this study adds to our knowledgeIn an ED population with suspected non–ST-segmentelevation acute coronary syndrome, increase of eithercardiac troponin or CKMB levels appears to increase theodds of acute coronary syndrome compared with that ofpatients with normal concordant cardiac biomarkers.Because of the constraints of incorporation andevaluation bias, these conclusions should be viewed astentative at this time.

How this might change clinical practiceUntil more information becomes available, it appearswise to regard increase of either cardiac troponin orCKMB as a true positive result and proceed accordingly.

in Unstable Angina: Receptor Suppression Using Integrilin

Volume , . : December

Therapy [PURSUIT]6). In the latter, an increased cardiactroponin level, regardless of CKMB level, was associated withincreased mortality in nearly 30,000 patients analyzed from themore contemporary CRUSADE (Can Rapid Risk Stratificationof Unstable Angina Patients Suppress Adverse Outcomes withEarly Implementation of the ACC/AHA Guidelines) registry.

ImportanceThe conclusion from the above trials that patients with

increased CKMB and normal total creatine kinase levels orincreased cardiac troponin levels discordant with CKMB shouldreceive aggressive management commensurate with theirincreased risk has significant implications for ED care.However, both of these studies included only high-risk patients(ie, patients eligible for inclusion had objective signs ofnon–ST-segment elevation acute coronary syndrome and wereinpatients). A similar analysis across the full spectrum of chestpain syndrome presentations in the ED would help substantiateor refute these conclusions in a more generalizable set ofpatients for the practicing emergency physician.

Goals of This InvestigationWe evaluated the association between 3 discordant pairs of

cardiac biomarkers (CKMB�/creatine kinase–, CKMB�/cardiactroponin–, and cardiac troponin�/CKMB–) and the risk of acutecoronary syndrome in ED chest pain patients.

MATERIALS AND METHODSStudy Design

The Internet Tracking Registry of Acute CoronarySyndromes, detailed elsewhere,7 was a multicenter registry ofpatients with possible acute coronary syndrome using directobservation and medical record review. Institutional reviewboards or ethics committees approved patient enrollmentwithout informed consent at 8 of the 9 centers; at one center,verbal informed consent was required from subjects beforeenrollment.

SettingSeven academic and 2 community hospitals participated; ED

census varied between 10,000 and 160,000 visits during thecourse of the study, and acute coronary syndrome rates variedbetween 1% and 10%. Eight sites were located in the UnitedStates and 1 was in Singapore.

Selection of ParticipantsProspective enrollment occurred from June 1999 through

August 2001. Inclusion criteria were age 18 years or older andan ECG obtained for possible acute coronary syndromeevaluation. Subjects were excluded if they were transferredfrom another institution or if the ECG was obtained forroutine purposes before a non–cardiac-related procedure. Aconvenience sample was enrolled at each center; efforts were
made to capture the potential acute coronary syndrome
Annals of Emergency Medicine 661

Discordant Cardiac Biomarkers Storrow et al

populations as completely as possible. For this analysis, data forthe index visit of patients with cardiac biomarkers drawn in theED and with complete demographic and ECG data wereincluded.

Data Collection and ProcessingEach site used standardized data collection forms;

prospective ED data were collected directly from physicians,through direct patient interview, ED record review, and reviewof computerized laboratory results. Hospital course wasascertained by record review or daily follow-up of inpatients.The subject’s treating ED physician provided the ECGinterpretation. Timing of cardiac markers was documentedusing the time results available to the treating physician.Patients were contacted at 30 days for determination ofoutcomes. Primary follow-up was by telephone. Telephonefollow-up was attempted for up to 90 days. Events occurringduring the 30-day period after the ED presentation werecaptured. Medical record review and death registry review wereused to follow up patients who could not be contacted by

Table 1. Participating institutions, associated platforms, andupper limit of normal for creatine kinase and CKMB.

Institution Platform Marker ULN

Site 1 Roche Integra* CK 173 U/LDade Stratus CS† CKMB 3.5 ng/mL

Site 2 Hitachi Modular P* CK 220 U/LAbbott IMX‡ CKMB 5 ng/mL

Site 3 Hitachi 917* CK M�250 U/LF�160 U/L

Elecsys 2010* CKMB 5.8 ng/mLSite 4 Vitros 950§ CK M�170 U/L

F�135 U/LAbbott AxSYM‡ CKMB 7.5 ng/mL

Site 5 Beckman Coulter LX 20� CK M�249 U/LF�189 U/L

Bayer Immuno 1 (6/99-8/00)¶/ CKMB 7.0 ng/mLCentaur (8/00-2001)¶ 5.0 ng/mL

Site 6 Enzymatic, Roche Modular* CK M�210 U/LF�164 U/L

Immunoassay, Roche Modular* CKMB 5.0 �g/LSite 7 Hitachi 917* CK M�250 U/L

F�160 U/LElecsys 2010* CKMB 5.8 ng/mL

Site 8 Beckman Synchron System� CK 250 U/LAbbott AxSYM‡ CKMB 8 �g/L

Site 9 Vitros 950§ CK M�315 U/LF�150 U/L

Abbott AxSYM‡ CKMB 5 �g/L

CK, Creatine kinase; F, female; M, male; ULN, upper limit of normal.*Roche Diagnostics Corporation, Indianapolis, IN.†Dade Behring Inc, Deerfield, IL.‡Abbott Laboratories, Abbott Park, IL.§Ortho-Clinical Diagnostics, Johnson&Johnson, Raritan, NJ.�Beckman Coulter, Inc., Brea, CA.¶Bayer Corporation, Tarrytown, NY.

telephone.

662 Annals of Emergency Medicine

Patients were categorized into one of 5 groups:● Increased CKMB�normal creatine kinase● Increased CKMB�increased creatine kinase● Normal CKMB�increased cardiac troponin● Increased CKMB�normal cardiac troponin● Increased CKMB�increased cardiac troponin

Creatine kinase, CKMB, and cardiac troponin (either I or T)values less than or equal to the upper limit of normal for eachlocal laboratory were classified as normal or negative (Table 1and Table 2).

Outcome MeasuresAcute coronary syndrome was defined according to events,

positive test results, or revascularization either during thehospital visit or within 30 days of the ED presentation. Positivefindings were greater than 70% stenosis in any vessel atcatheterization, percutaneous coronary intervention, coronaryartery bypass grafting, a documented myocardial infarction,all-cause death, a positive noninvasive provocative test forischemia, positive myocardial perfusion imaging, or a DRGcode indicating revascularization (percutaneous coronaryintervention, coronary artery stent placement, or coronary arterybypass grafting), acute myocardial infarction or non–Q-wavemyocardial infarction, or unstable angina with catheterization.

Primary Data AnalysisData are described using means and SDs or frequencies and

percentages as appropriate. The odds of acute coronarysyndrome were compared between groups using univariablelogistic regression. We used this method, rather than the moretraditional cross-tabulations, to calculate simple, unadjustedodds ratios because our statistical package would only provideconfidence intervals (CIs) if we used univariable logistic

Table 2. Participating institutions, associated platforms, andupper limit of normal for cardiac troponin.

Institution Platform Marker ULN

Site 1 Dade Stratus CS* cTnI 0.06 ng/mLSite 2 ES300 Boehringer-Mannheim† cTnT 0.2 �g/LSite 3 Elecsys 2010‡ cTnT 0.1 ng/mLSite 4 Abbott AxSYM§ cTnI 2.0 ng/mLSite 5 Bayer Immuno 1 (6/99-8/00)�/

Bayer Centaur (8/00-2001)�cTnI 0.2 ng/mL/

1.5 ng/mLSite 6 Immunoassay, Roche Modular‡ cTnT 0.10 �g/LSite 7 Elecsys 2010‡ cTnT 0.1 ng/mLSite 8 Abbott AxSYM§ cTnI 2.0 ng/mLSite 9 Abbott AxSYM§ cTnI 2.0 ng/mL

cTn, Cardiac troponin.*Dade Behring Inc. Indianapolis, IN.†Boehringer-Mannheim Corporation, Indianapolis, IN.‡Roche Diagnostics Corporation, Indianapolis, IN.§Abbott Laboratories, Abbott Park, IL.�Bayer Corporation, Tarrytown, NY

regression. Data were managed in Microsoft Access (Microsoft



Corporation, Redmond, WA) and analyzed using SPSS v 13.0(SPSS, Inc., Chicago, IL).

RESULTSCharacteristics of Study Subjects

Of 17,713 patient visits included in the registry, 8,769 metinclusion criteria. For our analysis, subjects were excluded ifthey were repeated visits (N�2,105), had no ECG diagnosticcategory recorded (N�897), had an ECG result consistent withST-segment elevation myocardial infarction (N�653), hadmissing demographic data (N�148), no cardiac markers wereobtained (N�2,712), only 1 cardiac marker was obtained(N�1,042), or neither a creatine kinase total nor troponin wasavailable within 1 hour of the initial CKMB test (N�1,387).Characteristics of included and excluded patients are shown inTable 3. There were 4.4% of patients included who were lost tofollow-up; follow-up has been considered outcome negative.

Main ResultsTables 4 and 5 describe the outcomes for patients with

concordant and discordant cardiac biomarkers. There were 7,740

Table 3. Characteristics of subjects included and excludedfrom the analysis.*

CharacteristicIncluded Patients

(N�8769)Excluded Patients

(N�8944)

Age, y (mean, SD) 55.5 (15.5) 53.3 (16.0)Sex

Female 4,278 (48.8) 4,808 (53.8)Male 4,491 (51.2) 4,120 (46.1)

RaceWhite 3,171 (37.7) 2,807 (34.3)Black 2,697 (32.0) 3,720 (45.5)Hispanic 198 (2.4) 270 (3.3)Asian 2,343 (27.8) 1,365 (16.7)American Indian 13 (0.2) 17 (0.2)

Initial diagnostic impressionAcute MI 97 (1.1) 473 (5.3)Unstable angina/non–Q-

wave MI818 (9.4) 714 (8.0)

High-risk chest pain 2,583 (29.6) 2,110 (23.6)Low-risk chest pain 3,810 (43.6) 3,137 (35.1)Noncardiac chest pain 1,422 (16.3) 2,269 (25.4)

Cocaine user 127 (1.4) 155 (1.7)Amphetamine user 24 (0.3) 24 (0.3)Current smoker 2,364 (27.0) 2,699 (30.2)Diabetes 1,871 (21.3) 1,961 (21.9)Hypertension 4,334 (49.4) 4,445 (49.7)Hypercholesterolemia 2,204 (25.1) 2,005 (22.4)Angina 722 (8.2) 991 (11.1)Coronary artery disease 1,941 (22.1) 2,192 (24.5)Congestive heart failure 564 (6.4) 781 (8.7)Family history of heart

disease2,899 (33.1) 2,845 (31.8)

MI, Myocardial infarction.*Data are given as frequencies and percentages unless otherwise indicated.Initial diagnostic impression is the treating emergency physician’s ED diagnosis.

visits with both a total creatine kinase and a CKMB level available


(Table 4) and 6,675 visits with both a CKMB and a troponin levelavailable (Table 5); 5,646 patient visits are included in both groups.Table 6 shows the odds ratios for acute coronary syndrome inpatients with and without discordant markers. For the group withcardiac troponin, the reference category was negative troponin andnegative CKMB; for the group with creatine kinase, the referencecategory was a negative CKMB result but either positive or negativecreatine kinase result.

LIMITATIONSSeveral limitations inherent with the use of registry data apply to

this study. This was a convenience sample of subjects; although anattempt was made to enroll patients across the spectrum of diseaseseverity, selection bias may have occurred. Patients who wereacutely ill may have been inadvertently excluded because of theirrapid disposition. Patients with atypical symptoms or absence ofrisk factors may not have undergone evaluation for acute coronarysyndrome. There are inherent limitations to the use of DRGs fordiagnosis. Our inclusion criteria resulted in exclusion of the lowest-risk patients who did not have cardiac markers measured in theED. Although our methodology likely excluded the very highest-and lowest-risk populations, we included the majority of patientsfor whom discordant cardiac biomarkers may represent a diagnosticdilemma.

In addition to potential inclusion bias, marker positivity wasdetermined using the site’s own laboratory-reported upper limitof normal; a core laboratory was not used. Although there isunavoidable variation in the individual assays across multiplesites, we believe this accurately represents emergency medicalpractice and the use of biomarkers in clinical decisionmaking.

The use of record review and death registry review might biastoward the underdetection of endpoints, whereas the use of all-cause mortality might bias toward overdetection. There is also thepossibility of incorporation bias because the diagnosis of acutecoronary syndrome often incorporates the presence of a positivecardiac biomarker. This, as well as evaluation bias, could lead toinflated and possibly erroneous associations between biomarkerdiscordance and risk of acute coronary syndrome.

Finally, we did not control for renal function in our analysis.It has been argued that the predictive value of CKMB and, to alesser extent, cardiac troponin is deteriorated in the presence ofrenal insufficiency. However, this is controversial, and ourfindings are robust, suggesting that in a real-worldundifferentiated ED population, increase of any marker suggestsincreased acute coronary syndrome risk, even without a renalfailure adjustment.

DISCUSSIONCardiac biomarker testing is a primary ED evaluation tool

and is fundamental to the evaluation of most patients with chestpain and suspected acute coronary syndrome. Despite therecognition and recommendation that cardiac troponin is thestandard for myocardial infarction diagnosis,8 many cliniciansroutinely add creatine kinase, CKMB, or both. Such dualtesting may result in increased CKMB level in the absence of
increase in creatine kinase level or discordance between CKMB

Discordant Cardiac Biomarkers Storrow et al

and cardiac troponin levels, resulting in uncertainty in thediagnosis of myocardial infarction. We retrospectively analyzedthe frequency of these occurrences and their relation to outcomein a large multicenter cohort of ED patients with suspectedacute coronary syndrome.

Increases in CKMB without concomitant increase in creatinekinase has been previously reported,9-16 although these reportswere for admitted patients, with only 1 exception. The inpatientstudies reported a prevalence of increased CKMB level withnormal creatine kinase level between 5.3% and 44%, thehighest from a study of high-risk patients and defining “MBleak” as an increased CKMB relative index with a normalcreatine kinase level during the first 24 hours. These studiesconcluded that increases of CKMB level with normal creatinekinase level identified patients at increased risk,11-14 thatpersistently normal creatine kinase levels do not exclude acute

Table 4. Frequency and percentage of acute coronary syndromekinase and CKMB results.

EventCKMB(�)/CK(�/�)

N�6,955 (89.9%)

MI 183 (2.6)Positive noninvasive test result 242 (3.5)CABG 128 (1.8)Death 48 (0.7)Abnormal MPI 74 (1.1)Acute coronary syndrome 1107 (15.9)

MI, Myocardial infarction; CABG, coronary artery bypass graft; Abnormal MPI, nuc

Table 5. Frequency and percentage of acute coronary syndrometroponin and CKMB results.

EventCKMB(�)/cTn(�)N�5,945 (89.1%)

CKMN�

MI 76 (1.3)Positive noninvasive test 207 (3.5)CABG 73 (1.2)Death 33 (0.6)Abnormal MPI 79 (1.3)Acute coronary syndrome 757 (12.7)

Table 6. Unadjusted odds of acute coronary syndrome forconcordant and discordant cardiac biomarkers.*

Biomarkers Odds Ratio95% CI

(Odds Ratio)

CKMB(�)/CK(�) 5.70 (4.38–7.40)CKMB(�)/CK(�) 4.36 (3.64–5.23)CKMB(�)/cTn(�) 4.79 (3.40–6.76)CKMB(�)/cTn(�) 2.17 (1.72–2.75)CKMB(�)/cTn(�) 26.58 (18.00–39.30)

*For the group with cardiac troponin, the reference category had negative tropo-nin and negative CKMB results. For the group with creatine kinase, the refer-ence category had a negative CKMB but either positive or negative creatine ki-nase result.

myocardial infarction,9 that CKMB should be measured in all

664 Annals of Emergency Medicine

patients with suspected acute myocardial infarction regardless ofcreatine kinase level,12 and that total creatine kinase level shouldbe abandoned in favor of CKMB level.15 Of note, these werereported before cardiac troponin’s emergence as the primarybiomarker for acute coronary syndrome.

The single ED study13 evaluated a composite outcome in698 ED patients with suspected acute coronary syndrome at asingle center. Seventeen (2.4%) patients had a normal initialcreatine kinase level and increased initial CKMB level. Acuteand 6-month death, Q-wave myocardial infarction, orrevascularization was the same for all patients with increasedCKMB level, regardless of the creatine kinase level. As did we,these authors concluded that the adverse event rate for patientswith suspected acute coronary syndromes and an increasedCKMB level is the same whether or not the total creatine kinaselevel is increased, which is at variance with a recent report thatan isolated positive CKMB result has limited prognostic valueamong high-risk acute coronary syndrome patients in theCRUSADE initiative.2

Although initial reports on the prognostic significance ofdiscordant CKMB and troponin conflicted,17,18 a large analysisof 29,357 high-risk registry patients concluded an increasedtroponin level identifies patients at increased acute risk,regardless of CKMB, and an isolated positive CKMB result haslimited prognostic value.2 A low-risk subgroup with isolatedtroponin-level increase in an earlier study had a remarkablysimilar odds ratio (4.8; 95% CI 1.4 to 16.0) for 30-day death ormyocardial infarction as in our report.17 However, thispopulation was composed of patients admitted to chest pain

in groups of patients with concordant and discordant creatine

(�)/CK(�)39 (3.1%)

CKMB(�)/CK(�)N�546 (7.1%)

CKMB(�)/CK(�/�)N�785 (10.1%)

44 (18.4) 126 (23.1) 170 (21.7)19 (7.9) 36 (6.6) 55 (7.0)13 (5.4) 21 (3.8) 34 (4.3)7 (2.9) 12 (2.2) 19 (2.4)4 (1.7) 11 (2.0) 15 (1.9)

24 (51.9) 247 (45.2) 371 (47.3)

yocardial perfusion imaging consistent with ischemia.

in groups of patients with concordant and discordant cardiac

)/cTn(�)(2.1%)

CKMB(�)/cTn(�)N�428 (4.9%)

CKMB(�)/cTn(�)N�161 (2.4%)

.1) 24 (5.6) 78 (48.4)0.6) 21 (4.9) 21 (13.0).5) 12 (2.8) 13 (8.1).5) 2 (0.5) 12 (7.5).8) 11 (2.6) 5 (3.1)1.1) 103 (24.1) 128 (79.5)

with

CKMBN�2

1

lear m

with

B(�141

10 (715 (15 (35 (34 (2

58 (4

centers19 and is arguably higher risk than our cohort.



When compared to those with normal biomarker levels,increased CKMB level with normal creatine kinase level anddiscordant CKMB and cardiac troponin levels have beensuggested to predict worse prognosis in high-risk patients withnon–ST-segment acute coronary syndrome. However, suchan analysis has not been performed for the undifferentiatedpopulation of chest pain syndrome in the ED. We suggest thatamong this wide spectrum of emergency patients, an increasedCKMB level with a normal creatine kinase level or discordancein either direction between CKMB and cardiac troponinidentifies patients at higher risk for acute coronary syndrome.

Supervising editor: E. John Gallagher, MD

Author contributions: The study was conceived by ABS, JHH,and CMS. ABS, CJL, JHH, and CMS designed the trial. TheEmergency Medicine Cardiac Research and Education Groupwas the committee for the iTrACS registry. ABS, CJL, and JHHsupervised the conduct of the trial and data collection.Recruitment of participating centers and data wasaccomplished by everyone except CMS. Data managementand quality control were accomplished by ABS, CJL, JHH, andKFM. Statistical advice on the study design was provided byCJL, and the study statistical consultant was CJL. ABS, CJL,JHH, and CMS analyzed the data. Data oversight wasconducted by WBG, with JEH, WFP, JWH and CVP. Themanuscript was drafted by ABS and KFM. All of the authorscontributed to the revisions and the final submission. ABStakes responsibility for the paper as a whole.

Funding and support: The i*trACS project was funded in partby Millennium Pharmaceuticals, Inc., Schering-PloughPharmaceuticals.

Publication dates: Received for publication March 21, 2006.Revisions received April 27, 2006; May 2, 2006; and May 5,2006. Accepted for publication May 12, 2006. Availableonline August 14, 2006.

Reprints not available from the authors.

Address for correspondence: Alan B. Storrow, MD, VanderbiltUniversity Medical Center, Department of EmergencyMedicine, 703 Oxford House, Nashville, TN 37232-4700; 615-936-5934, fax 615-936-1316; [email protected].

REFERENCES1. Galla JM, Mahaffey KW, Sapp SK, et al. Elevated creatine kinase-

MB with normal creatine kinase predicts worse outcomes inpatients with acute coronary syndromes: results from 4 largeclinical trials. Am Heart J. 2006;151:16-24.

2. Newby LK, Roe MT, Chen AY, et al. Frequency and clinicalimplications of discordant creatine kinase-MB and troponinmeasurements in acute coronary syndromes. J Am Coll Cardiol.2006;47:312-318.

3. GUSTO IIb Investigators. A comparison of recombinant hirudinwith heparin for the treatment of acute coronary syndromes: theGlobal Use of Strategies to Open Occluded Coronary Arteries.
N Engl J Med. 1996;335:775-782.

4. PARAGON Investigators. International, randomized, controlledtrial of lamifiban (a platelet glycoprotein IIb/IIIa inhibitor),heparin, or both in unstable angina: the PARAGONInvestigators: Platelet IIb/IIIa Antagonism for the Reduction ofAcute coronary syndrome events in a Global OrganizationNetwork. Circulation. 1998;97:2386-2395.

5. Mukherjee D, Mahaffey KW, Moliterno DJ, et al. Promise of combinedlow-molecular-weight heparin and platelet glycoprotein IIb/IIIa inhibition:results from Platelet IIb/IIIa Antagonist for the Reduction of Acutecoronary syndrome events in a Global Organization Network B(PARAGON B). Am Heart J. 2002;144:995-1002.

6. PURSUIT Trial Investigators. Inhibition of platelet glycoprotein IIb/IIIa with eptifibatide in patients with acute coronary syndromes:the PURSUIT Trial Investigators: Platelet Glycoprotein IIb/IIIa inUnstable Angina: Receptor Suppression Using Integrilin Therapy.N Engl J Med. 1998;339:436-443.

7. Lindsell CJ, Pollack CV, Anantharaman V, et al. The InternetTracking Registry of Acute Coronary Syndromes (i*trACS): a multi-center registry of patients with suspicion of ACS reported usingthe standardized reporting guidelines for emergency departmentchest pain studies. Ann Emerg Med. 2006;48:666-677.

8. Alpert JS, Thygesen K, Antman E, et al. Myocardial infarction redefined:a consensus document of the Joint European Society of Cardiology/American College of Cardiology Committee for the redefinition ofmyocardial infarction. J Am Coll Cardiol. 2000;36:959-969.

9. Dillon MC, Calbreath DF, Dixon AM, et al. Diagnostic problem in acutemyocardial infarction: CK-MB in the absence of abnormally elevated totalcreatine kinase levels. Arch Intern Med. 1982;142:33-38.

10. Dorogy ME, Hooks GS, Cameron RW, et al. Clinical andangiographic correlates of normal creatine kinase with increasedMB isoenzymes in possible acute myocardial infarction. AmHeart J. 1995;130:211-217.

11. Hong RA, Licht JD, Wei JY, et al. Elevated CK-MB with normaltotal creatine kinase in suspected myocardial infarction:associated clinical findings and early prognosis. Am Heart J.1986;111:1041-1047.

12. Lloyd-Jones DM, Camargo CA Jr, Giugliano RP, et al.Characteristics and prognosis of patients with suspected acutemyocardial infarction and elevated MB relative index but normaltotal creatine kinase. Am J Cardiol. 1999;84:957-962.

13. Peacock WF, Emerman CL, McErlean ES, et al. Normal CK,elevated MB predicts complications in acute coronary syndromes.J Emerg Med. 2001;20:385-390.

14. Pettersson T, Ohlsson O, Tryding N. Increased CKMB (massconcentration) in patients without traditional evidence of acutemyocardial infarction: a risk indicator of coronary death. EurHeart J. 1992;13:1387-1392.

15. Vladutiu AO, Schachner A. Increased creatine kinase (CK) MBisoenzyme in patients with “normal” total CK activity suspectedof acute myocardial infarction. J Med. 1989;20:73-81.

16. Yusuf S, Collins R, Lin L, et al. Significance of elevated MBisoenzyme with normal creatine kinase in acute myocardialinfarction. Am J Cardiol. 1987;59:245-250.

17. Rao SV, Ohman EM, Granger CB, et al. Prognostic value ofisolated troponin elevation across the spectrum of chest painsyndromes. Am J Cardiol. 2003;91:936-940.

18. Yee KC, Mukherjee D, Smith DE, et al. Prognostic significance ofan elevated creatine kinase in the absence of an elevatedtroponin I during an acute coronary syndrome. Am J Cardiol.2003;92:1442-1444.

19. Newby LK, Storrow AB, Gibler WB, et al. Bedside multimarkertesting for risk stratification in chest pain units: the Chest PainEvaluation by Creatine Kinase-MB, myoglobin, and Troponin I
(CHECKMATE) Study. Circulation. 2001;103:1832-1837.

Documents

Discordant Cardiac Biomarkers: Frequency and Outcomes in Emergency Department Patients With Chest Pain