Journal of Critical Care (2010) 25, 322–328

Etiology of troponin elevation in critically ill patientsWendy Lim MD, MSca,b,⁎, Richard Whitlock MD, PhDc,d, Vikas Khera MDa,b,Philip J. Devereaux MD, PhDa,d,e, Andrea Tkaczyk MNb, Diane Heels-Ansdell MScd,e,Michael Jacka MD, MScf, Deborah Cook MD, MSca,d,e

aDepartment of Medicine, McMaster University, Hamilton, Ontario, Canada L8N 3Z5bSt Joseph's Healthcare, Hamilton, Ontario, Canada L8N 4A6cDepartment of Surgery, McMaster University, Hamilton, Ontario, Canada L8N 3Z5dHamilton Health Sciences, Hamilton, Ontario, Canada L8N 3Z5eDepartment of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada L8N 3Z5fDepartment of Critical Care, University of Alberta, Edmonton, Alberta, Canada T6G 2B7

C

rwv(Pad

0d

Keywords:Troponin;Critically ill;Intensive care unit;Etiology

AbstractPurpose: The aim of this study was to assess the etiology of cardiac troponin elevation among patientsadmitted to the intensive care unit (ICU) and to examinewhether etiology affectsmortality and length of stay.Methods: All patients admitted over 2 months underwent screening with troponin measurements and wereincluded if 1 or more measurements were elevated. Two adjudicators retrospectively reviewed patient chartsto determine the likely cause of troponin elevation.Results: Of 103 patient admissions, 52 (50.5%) had 1 or more elevated troponin measurements, and 49(94.2%) had medical charts available for review. Troponin elevation was adjudicated as myocardialinfarction (MI) in 53.1% of patients, sepsis in 18.4%, renal failure in 12.2%, and other causes in 16.3%.Overall ICU mortality was 16.0%; 2.0% for patients with no troponin elevation, 23.1% in patients with MI,and 39.1% in patients with troponin elevation not due toMI. Having an elevated troponin level not due toMIwas significantly associated with increased hospital mortality compared with having no troponin elevation.Conclusions: The most common cause of troponin elevation among critically ill patients was MI. Patientswith elevated troponin hadworse outcomes compared with patients without troponin elevation, and troponinelevation not due to MI was predictive of increased hospital mortality.© 2010 Elsevier Inc. All rights reserved.

⁎ Corresponding author. St Joseph's Healthcare, Hamilton, Ontario,anada L8N 4A6.E-mail addresses: limwp@mcmaster.ca (W. Lim),hitlock@mountaincable.net (R. Whitlock),

ikas.khera@learnlink.mcmaster.ca (V. Khera), philipj@mcmaster.ca.J. Devereaux), tkaczyaj@mcmaster.ca (A. Tkaczyk),nsdell@mcmaster.ca (D. Heels-Ansdell), mjacka@ualberta.ca (M. Jacka),ebcook@mcmaster.ca (D. Cook).

883-9441/$ – see front matter © 2010 Elsevier Inc. All rights reserved.oi:10.1016/j.jcrc.2009.07.002

1. Introduction

Cardiac troponin I and T are regulatory proteins thatcontrol the interaction of actin and myosin, the proteinsinvolved in myocardial cell contraction. Injury to myocardialcells results in release of troponin into the blood, which canbe measured by sensitive and specific assays for thediagnosis of myocardial infarction (MI). As per the Joint

323Etiology of troponin elevation in critically ill patients

European Society of Cardiology/American College ofCardiology Foundation/American Heart Association/WorldHeart Federation (ESC/ACCF/AHA/WHF) Task Force forthe Redefinition of Myocardial Infarction, MI is diagnosedby a typical rise or fall in troponin and 1 or more of thefollowing: ischemic symptoms, electrocardiographic (ECG)changes indicative of ischemia, or imaging evidence of MI[1]. Troponin elevation is thus a very useful marker for MI,but it is important to recognize that an elevated troponin levelreflects myocardial damage that may occur through severalmechanisms. Plaque rupture with superimposed thrombosiswithin a coronary artery resulting in myocardial cell death isthe most common mechanism for MI among patientspresenting to a hospital. However, pulmonary embolism,renal failure, sepsis, hypotension, and other causes are allrecognized to result in troponin elevation [2,3]. In patientswith pulmonary embolism, strain in the right side of the heartand myocardial necrosis in the right ventricle are believed toresult in troponin elevation. In sepsis, systemic hypotensionresulting in suboptimal coronary artery blood flow andcytokine release are hypothesized to result in troponinleakage from myocardial cells [4,5].

It is currently unclear if the outcome of patients with MIdue to plaque rupture is similar to that of patients with MIdue to other etiologies. Nonetheless, recognition of thedifferent etiologies resulting in MI has led the Joint ESC/ACCF/AHA/WHF to divide MI into the following clinicalclassifications: type 1, MI related to ischemia due to plaquerupture; type 2, MI secondary to supply-demand imbalance;type 3, sudden unexpected cardiac death; type 4a, MIassociated with percutaneous coronary intervention; type 4b,MI associated with stent thrombosis; and type 5, MIassociated with coronary artery bypass grafting [1].

Elevated troponin levels are observed in up to 50% ofcritically ill patients in the intensive care unit (ICU) [6,7],although many of these patients do not have evidence ofclinically important atherosclerotic coronary artery disease[8,9]. This suggests that noncritical coronary artery occlu-sions are clinically important in the ICU setting or thatetiologies other than plaque rupture may be more common incritically ill patients. Despite having different etiologies fortroponin elevation, a systematic review of observationalstudies of critically ill patients with elevated troponinindicates that these patients have an adjusted odds ratio(OR) for death of 2.5 (95% confidence interval [CI], 1.9-3.4)compared with patients without troponin elevation [10].Thus, troponin elevation among critically ill patients iscommon and seems to be an independent predictor for deathin the ICU.

Identifying the cause for troponin elevation is animportant first step in determining the optimal managementfor these patients. It is unknown whether the traditionaltherapies for MI, including antiplatelet and antithromboticagents, β-blockers, and afterload reducing agents, have anyeffect (either beneficial or harmful) in patients with MI due tosupply-demand imbalance.

The primary objective of this study was to assess thefrequency of the various etiologies of troponin elevationamong critically ill patients through an adjudication process.The secondary objective was to examine whether patientswith different etiologies of troponin elevation had similarICU and hospital mortality and length of stay.

2. Methods

2.1. Patient identification

All consecutive patients admitted to the ICU at StJoseph's Hospital, Hamilton, Canada, from January toMarch 2006 underwent systematic screening of cardiactroponin T and 12-lead ECG. In brief, screening troponin Tand ECG investigations were performed at ICU admission,then daily for the first week in the ICU, followed byalternate days for up to 1 month, and then weekly thereafteruntil ICU death or discharge, for a maximum of 2 months.Results of this screening study have been previouslypublished [11]. We used a deferred consent procedure toensure timely acquisition of troponin samples and ECGs.Patient data that were already available in the medical chartwere collected, but no additional screening troponinmeasurements or ECGs were performed if consent wasnot obtained. The St Joseph's Hospital Research EthicsBoard approved this study.

Only patients who had 1 or more elevated troponinmeasurements were included in this study of etiologies andoutcomes. An elevated cardiac troponin T was defined as avalue greater than or equal to 0.04 μg/L, which representsthe assay coefficient of variation of 10%. The analyticallower detection limit of the assay was 0.01 μg/L.

2.2. Setting

The ICU at St Joseph's Hospital is a 15-bed, university-affiliated medical-surgical ICU. Although the hospital has acoronary care unit for patients with primary cardiacdiagnoses or requiring telemetry, such patients alsorequiring mechanical ventilation and those receiving ino-tropes and/or vasopressors are admitted to the ICU. TheICU is a closed teaching unit staffed by intensivists andphysicians-in-training.

2.3. Data collection

On ICU admission, we prospectively collected patientdemographics and baseline data. Daily, we collectedlaboratory results, cardiac medications and anticoagulants,new clinical events and complications that the ICU teamdetected (development of arrhythmias, pulmonary edema,nonfatal cardiac arrest, and cardiogenic shock), need foradvanced life support (mechanical ventilation, inotropes and/

324 W. Lim et al.

or vasopressors, and hemodialysis), and ischemic cardiacsymptoms. We also collected information on whether theICU team made a diagnosis of MI during the patient's ICUstay. We prospectively followed all patients throughout theirICU stay and recorded their vital status and duration of stayin ICU and hospital. Data were collected from the patient'smedical chart, the ICU computerized clinical informationsystem, and the hospital laboratory system. Data wererecorded on paper case report forms and entered into anExcel program for analysis.

To adjudicate the likely etiology of troponin elevation, thecharts of the patients with 1 or more measurements of anelevated troponin level were reviewed in duplicate andindependently by 2 adjudicators. The adjudicators weresenior critical care fellows who had at least 5 years ofpostgraduate training and have worked independently in theSt Joseph's Hospital ICU under the direction of staffintensivists. Adjudication results were collected on astandardized data collection form on which possibleetiologies of troponin elevation were listed (Appendix A).We piloted the data collection form in duplicate on 10 patientcharts before full data collection to refine the categories andetiologic definitions. Each patient's medical history, dailyphysician and nursing notes, hourly physiologic data,laboratory and radiology reports, the progression of the MI,and other relevant investigations were used to adjudicate theetiology of troponin elevation and type of MI. Adjudicatorswere asked to indicate the most likely etiology based on theforegoing information and their clinical judgment. Ifdiscordant etiologies were identified between the 2 adjudi-cators as the most likely cause of troponin elevation, theadjudicators discussed the case to reach agreement. Adjudi-cators also independently classified all patients with MI into1 of 2 of the Joint ESC/ACCF/AHA/WHF categories: type 1(MI related to ischemia due to plaque rupture) or type 2 (MIsecondary to supply-demand imbalance) [1]. As per the JointESC/ACCF/AHA/WHF Guidelines [1], evidence for classi-fying as type 1 MI included (1) coronary angiographicevidence of acute vessel occlusion and (2) history ofcoronary artery disease with typical ECG evidence ofprogression of infarction (development of new Q waves).Because coronary angiography is rarely performed in theICU, we accepted the second criterion alone as evidence oftype 1 MI. Evidence for classifying as type 2 included nohistory of coronary artery disease or risk factors thereof andthe presence of a condition recognized as associated withtroponin elevation (eg, hypotension, hypovolemia, andsupraventricular tachycardia).

2.4. Statistical analysis

We report continuous data as mean and standard deviationor median and interquartile range, as appropriate. Percen-tages are reported for binary data. Continuous variables arecompared using unpaired t tests or the Wilcoxon 2-sample

rank sum test and dichotomous variables using Fisher exacttest. We report the etiology of troponin elevation asproportions and calculate raw agreement and chance-corrected agreement (κ statistic) between the 2 adjudicatorsregarding the etiology of troponin elevation and theagreement on the assessment of type 1 and 2 MI. Wecompared the outcomes (mortality and length of stay) ofpatient subgroup by etiology of troponin elevation. As anexploratory analysis, we performed multivariable logisticregression analyses, adjusting for factors known to affectmortality, including Acute Physiology and Chronic HealthEvaluation (APACHE) II score and advanced life support(mechanical ventilation, inotropes or vasopressors, andhemodialysis), reported using ORs and 95% CIs.

3. Results

Over the 2-month study period, there were 112 ICUadmissions representing 103 unique patients; 9 patients wereadmitted twice. The clinical characteristics of the 103enrolled patients have been previously published [11]. Inbrief, most admissions were medical (63.1%), and theadmitting APACHE II diagnoses included cardiovascular(13.8%), respiratory (21.6%), gastrointestinal (12.3%),neurologic (9.3%), sepsis (13.8%), metabolic (16.9%), andhematologic (12.3%). Only 1 of the 9 patients admitted witha cardiovascular diagnosis had an admission diagnosis ofacute MI. The remaining 36.9% were surgical patients, with27 (71.1%) undergoing elective surgery and 11 (28.9%)undergoing emergent surgery.

Of these 103 patients, 52 (50.5%) had 1 or more elevatedmeasurements of troponin based on screening measurementsdone during their ICU admission. Of these patients, 49(94.2%) had charts available for review in the medicalrecords department; these were the patients included in thisstudy of etiology and outcomes.

Table 1 shows the frequencies of each etiology beingidentified as the most likely cause of troponin elevation foreach adjudicator, in addition to their consensus. The 2adjudicators reached consensus in all cases for whichdiscordant etiologies were initially identified as the mostlikely cause of troponin elevation. The most commonlyidentified cause was MI, occurring in 26 (53.1%) patients.Raw agreement for the most likely cause was 67.3%, and κwas 0.55. Of patients with MI, 16 (61.5%) patients wereadjudicated as having type 1MI and 10 (38.5%) patients wereadjudicated as having as type 2 MI. Because the number ofpatients with MI was small, we did not examine differencesbetween type 1 and type 2 MI in subsequent analyses.

Table 2 shows the baseline patient characteristicsseparated by etiology of troponin elevation. Patients withMI were more likely to be mechanically ventilated onadmission compared with patients with troponin elevationdue to other etiologies.

Table 2 Baseline patient characteristics according to etiology

MI(n = 26)

Other etiology(n = 23)

No elevatedtroponin (n

Age, mean (SD) 69.8 (15.2) 66.7 (19.2) 59.3 (16.9)Female sex, no. (%) 11 (42.3) 8 (34.8) 24 (47.1)APACHE II score,mean (SD)

26.4 (9.1) 26.5 (10.5) 16.2 (8.3)

Medical, no. (%) 21 (80.8) 19 (82.6) 23 (45.1)Prior coronary arterydisease, b no. (%)

10 (38.5) 5 (21.7) 9 (17.6)

Baseline life supportinterventions, c no. (%)Invasive MV 21 (80.8) 10 (43.5) 25 (49.0)Noninvasive MV 0 (0) 2 (8.7) 2 (3.9)Baseline inotropes andvasopressors, c no. (%)Epinephrine 0 (0) 3 (13.0) 0 (0)Dopamine d 2 (7.7) 3 (13.0) 1 (2.0)Norepinephrine 3 (11.5) 8 (34.8) 3 (5.9)Dobutamine 0 (0) 2 (8.7) 0 (0)Phenylephrine 2 (7.7) 1 (4.3) 0 (0)Vasopressin 0 (0) 1 (4.3) 0 (0)Any of the above 5 (19.2) 10 (43.5) 4 (7.8)Hemodialysis, c no. (%)Intermittent 3 (11.5) 4 (17.4) 2 (3.9)CRRT 0 (0) 1 (4.3) 0 (0)

CRRT indicates continuous renal replacement therapy; MV, mechanical ventilaa P values for continuous data based on unpaired t test; and for binary datab Coronary artery disease = angina, MI.c Study day 1.d N3 μg kg−1 min−1.

Table 1 Frequency of each etiology adjudicated as being themost likely cause for troponin elevation in critically ill patients

Etiology for elevatedtroponin

Adjudicator1 (n = 49)

Adjudicator2 (n = 49)

Consensus(n = 49)

MI 21 (42.9) 24 (49.0) 26 (53.1)Sepsis 7 (14.3) 9 (18.4) 9 (18.4)Renal failure 8 (16.3) 7 (14.3) 6 (12.2)Cardiac contusion/cardiopulmonaryresuscitation

2 (4.1) 2 (4.1) 3 (6.1)

Chronic obstructivepulmonary disease

2 (4.1) 2 (4.1) 3 (6.1)

Congestive heart failure 3 (6.1) 1 (2.0) 2 (4.1)Pulmonaryhypertension

1 (2.0) 1 (2.0) 0 (0)

Hypotension 1 (2.0) 1 (2.0) 0 (0)Left ventricularhypertrophy/strain

0 (0) 1 (2.0) 0 (0)

Other a 4 (8.2) 1 (2.0) 0 (0)

Values are presented as no. (%).a Eclamptic seizure, malignancy, loss of airway, cocaine-induced

vasospasm.

325Etiology of troponin elevation in critically ill patients

Table 3 shows the morbidity and mortality outcomes ofthe patients with elevated troponin based on their underlyingidentified etiology. Patients with elevated troponin due toreasons other than MI had the highest ICU and hospitalmortality rates (39.1% and 47.8%, respectively) comparedwith patients with MI (23.1% and 30.8%, respectively) andpatients without troponin elevation (2.0% and 2.0%,respectively). The comparison of ICU and hospital mortalityin patients with elevated troponin (due to MI and reasonsother than MI) compared with patients with no troponinelevation was statistically significant (P b .01). Patients withtype 1 and type 2 MI did not have a significantly differentICU mortality (18.8% and 30.0%, respectively; P = .64) orhospital mortality (25.0% and 40.0%, respectively; P = .66).

Table 4 shows the results of multivariable regressionanalysis by etiology of troponin elevation. Patients withelevated troponin due to reasons other than MI had anincreased risk of hospital death (OR, 86.2; 95% CI, 3.7-1985.8) in comparison with patients with no elevatedtroponin. The odds of dying in the hospital among patientswith troponin elevation due to MI compared with patientswith no elevated troponin was 14.3 (95% CI, 0.98-207.2).

= 51)P a

MI vs otheretiology

MI vs noelevated troponin

Other etiology vsno elevated troponin

.53 .01 .10

.77 .81 .45

.97 b.001 b.001

1.0 .003 .003.23 .06 .75

.009 .008 .80

.22 .55 .58

.10 – .03

.66 .26 .09

.09 .40 .003

.22 – .091.0 .11 .31.47 – .31.12 .16 b.001

.69 .33 .07

.47 – .31

tion., on Fisher exact test.

Table 3 Frequency of morbidity and mortality outcomes based on etiology of troponin elevation

No. ofpatients

Duration ofmechanical ventilation,median (IQR), d

Duration of ICUstay, median(IQR), d

Duration ofhospital stay,median (IQR), d

ICUmortality,no. (%)

Hospitalmortality,no. (%)

Total 100 2 (0-4.5) 4 (2-7) 12 (6-23) 16 (16.0) 20 (20.0)MI 26 4 (2-8) 5.5 (4-9) 15 (6-34) 6 (23.1) 8 (30.8)Other etiology 23 3 (0-11) 6 (3-11) 15 (10-30) 9 (39.1) 11 (47.8)Sepsis 9 7 (1-20) 9 (4-20) 20 (11-38) 5 (55.6) 5 (55.6)Renal failure 6 0 (0-1) 3 (2-5) 20 (13-20) 0 (0) 0 (0)Cardiac contusion/CPR 3 8 (3-11) 10 (3-11) 10 (3-12) 3 (100.0) 3 (100.0)COPD 3 4 (0-56) 7 (5-65) 19 (6-65) 1 (33.3) 3 (100.0)CHF 2 7.5 (0-15) 11.5 (8-15) 17.5 (13-22) 0 (0) 0 (0)No elevated troponin 51 1 (0-2) 3 (2-5) 9 (5-20) 1 (2.0) 1 (2.0)P value for MI vsother etiology a

– .62 .49 .95 .35 .25

P value for ACS/MI vsno elevated troponin a

– b.001 .01 .18 .005 b.001

P value for other etiologyvs no elevated troponin a

– .02 .002 .08 b.001 b.001

ACS indicates acute coronary syndrome; CPR, cardiopulmonary resuscitation; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease;IQR, interquartile range.

a P values for continuous data (durations) based on Wilcoxon rank sum test and for binary data based on Fisher exact test.

326 W. Lim et al.

4. Discussion

Of 103 consecutive patients admitted to our single-centermedical-surgical ICU, approximately 50% of patients had atleast 1 elevated troponin measurement under screeningconditions [11]. Approximately half of these patients withtroponin elevation were adjudicated in duplicate to have MI,and of the patients with MI, more than one third wereconsidered to have a supply-demand imbalance in thecoronary circulation as the etiology for troponin elevation.Troponin elevation in the absence of MI was an independentpredictor of hospital mortality. Troponin elevation in thepresence of MI also appeared strongly associated withhospital mortality, although this was not significant due tothe relatively small sample size. Therefore, an elevatedtroponin measurement in critically ill patients is common andappears to confer a worse prognosis; however, our studyraises the question of whether the prognostic relevance of

Table 4 ICU and hospital mortality

ICU mortality, OR (95% CI

Univariable Mu

APACHE II (10-point increment) 5.7 (2.6-12.5) 4.Mechanical ventilation a – –Inotropes or vasopressors a 81.9 (9.9-674.8) 26.Hemodialysis a 2.0 (0.6-7.2) 0.Etiology for elevated troponin b

MI 15.0 (1.7-132.6) 7.Other etiology 32.1 (3.7-275.7) 13.

a Ever/never throughout ICU stay in study. All patients who died in the ICUmortality.

b As compared with no elevated troponin.

this biomarker varies depending on the etiology of thetroponin elevation.

The finding that elevated troponin is an independentpredictor of adverse outcomes has been previously docu-mented. In a systematic review of critically ill patients whohad 1 or more measurements of an elevated troponin, weobserved an increased risk of death (OR, 2.5; 95% CI, 1.9-3.4) compared with patients without troponin elevation [10].In unadjusted analyses, patients with troponin elevation hadan increased length of ICU stay of 3 days and an increasedlength of hospital stay of 2.2 days. These results wereconfirmed in a recent retrospective study of 1657 medicalICU patients [12]. In another retrospective study, 240critically ill patients without other evidence of acute coronarysyndromes and who had normal (b0.10 ng/mL) andintermediate (0.10-1.49 ng/mL) levels of troponin I wereevaluated. Patients with an intermediate troponin level had a28% in-hospital mortality compared with 5% in the group

) Hospital mortality, OR (95% CI)

ltivariable Univariable Multivariable

8 (1.5-15.2) 4.9 (2.4-9.7) 5.3 (1.6-17.9)16.3 (2.1-128.1) 16.8 (0.96-292.3)

8 (2.7-266.6) 22.7 (6.5-79.6) 4.6 (0.9-23.2)3 (0.03-3.4) 1.4 (0.4-5.0) 0.10 (0.007-1.3)

0 (0.4-115.9) 22.2 (2.6-190.3) 14.3 (0.98-207.2)8 (0.9-219.9) 45.8 (5.4-390.3) 86.2 (3.7-1985.8)

received mechanical ventilation; thus, no ORs are presented for the ICU

327Etiology of troponin elevation in critically ill patients

with normal troponin level; furthermore, elevated troponinwas an independent predictor of hospital mortality andincreased length of stay [13]. In a prospective study of 990consecutive troponin-positive patients admitted to a single-center emergency department, 9% of patients had troponinelevations due to reasons other than MI, and these patientshad a higher in-hospital mortality (27% compared with 13%;P = .002) compared with patients with troponin elevation dueto MI or other cardiac causes [14].

Weaknesses of this study include the retrospectiveadjudication, which could introduce random or systematicerror in interpretations due to missing, misleading, orincomplete data in medical records. We attempted to addressthese limitations by using duplicate, independent assessmentof the possible etiologies, then conducting a calibration andpilot exercise to standardize adjudication. As in clinicalpractice, there was no reference standard of coronaryangiography performed for all patients at the time of theirtroponin elevation. The limited number of events andpatients precluded robust multivariable analyses, and thesingle-center design attenuates the generalizability of theseresults. Finally, because our coronary care unit does notadmit patients requiring ventilatory or inotropic support, it ispossible that the number of patients with MI requiring theseinterventions would be increased in our ICU compared withother ICUs. However, we examined the patients withcardiovascular diagnoses admitted to the ICU and foundthat very few actually had MI as an admitting diagnosis.

Understanding the etiology of troponin elevation incritically ill patients is important to better target appropriatetherapies for MI in this population. It is currently unknownwhether traditional anti-ischemic therapies such as β-blockers and angiotensin-converting enzyme inhibitors,which have been shown to be life-saving in patients withMI outside of the ICU, are of benefit in critically ill patientswith an elevated troponin in the setting of another diagnosis,such as septic shock. Prospective studies evaluating theetiology of troponin elevation are required to betterunderstand how to interpret troponin elevation in the ICU,the clinical significance of this finding, and ultimately, theappropriate management of such patients.

5. Conclusions

Elevated troponin levels are present in approximately50% of critically ill patients when assessed by screening.The most common etiology as adjudicated through thisduplicate independent process was MI, diagnosed in half thepatients. Of these patients, more than one third wereconsidered to have a supply-demand mismatch rather thanclassic plaque rupture. Patients with troponin elevation dueto reasons other than MI have at least as high a risk ofhospital death as do patients with MI compared with patientswithout troponin elevation. Thus, different etiologies for

troponin elevation may be associated with different rates ofmorbidity and mortality, although larger studies are neededto confirm or refute these findings. Further prospectiveevaluation is required to better determine the etiology andprognostic relevance of elevated troponin measurementsduring critical illness.

Acknowledgments

We thank the critically ill patients and families whoparticipated in this study. We are grateful for the assistanceof Josephine Baldwin and Dr Matthew McQueen regardingthe troponin measurements. We appreciate discussions aboutthis study with Drs David Bennett and David Treacher.

Dr Lim is the recipient of the Canadian Institutes forHealth Research Randomized Controlled Trials MentoringProgram Award and the EJ Moran Campbell Internal CareerResearch Award (McMaster University). Dr Devereaux issupported by a Canadian Institutes for Health Research NewInvestigator Award. Dr Cook holds a Canada Research Chairof the Canadian Institutes for Health Research and aCanadian Institutes for Health Research RandomizedControlled Trials Mentoring Program Award.

Appendix A. Causes of troponin elevation usedin the daily data collection form

Type 1 MI (plaque rupture) a

Type 2 MI (supply-demand imbalance) b due to:HypotensionHypovolemiaSupraventricular tachycardiaSevere anemiaVasospasmSepsisLeft ventricular hypertrophy/strainIntracranial hemorrhage/strokeCardiac contusion/cardiopulmonary resuscitationCardiac infiltrative disorders (eg, amyloid)ChemotherapyMyocarditisPericarditisCardiac surgeryCongestive heart failureCardiomyopathyPulmonary embolism/pulmonary hypertensionChronic obstructive pulmonary diseaseRenal failure

a Requires (1) coronary angiographic evidence of acute vesselocclusion or (2) history of coronary artery disease with typical ECGevidence of progression of infarction (development of new Q waves).

b Requires no history of coronary artery disease or risk factorsthereof and the presence of a condition recognized as associatedwith troponin elevation.

328 W. Lim et al.

References

[1] Thygesen K, Alpert JS, White HD, et al. Universal definition ofmyocardial infarction. Circulation 2007;116:2634-53.

[2] Jeremias A, Gibson CM. Narrative review: alternative causes forelevated cardiac troponin levels with acute coronary syndromes areexcluded. Ann Intern Med 2005;142:786-91.

[3] Roongsritong C, Warraich I, Bradley C. Common causes oftroponin elevations in the absence of acute myocardialinfarction: incidence and clinical significance. Chest 2004;125:1877-84.

[4] Bouhemad B, Nicolas-Robin A, Arbelot C, et al. Isolated andreversible impairment of ventricular relaxation in patients with septicshock. Crit Care Med 2008;36:766-74.

[5] Favory R, Neviere R. Significance and interpretation of elevatedtroponin in septic patients. Crit Care 2006;10:224.

[6] Guest TM, Ramanathan AV, Tuteur PG, et al. Myocardial injury incritically ill patients. JAMA 1995;273:1945-9.

[7] Noble JS, Reid AM, Jordan LVM, et al. Troponin I and myocardialinjury in the ICU. Br J Anaesth 1999;82:41-6.

[8] Ammann P, Maggiorini M, Bertel O, et al. Troponin as a risk factor formortality in critically ill patients without acute coronary syndromes.J Am Coll Cardiol 2003;41:2004-9.

[9] Quenot JP, Le Teuff G, Quantin C, et al. Myocardial injury in criticallyill patients: relation to increased cardiac troponin I and hospitalmortality. Chest 2005;128:2758-64.

[10] Lim W, Qushmaq I, Devereaux PJ, et al. Elevated cardiac troponinmeasurements in critically ill patients. Arch InternMed 2006;166:2446-54.

[11] Lim W, Holinski P, Devereaux PJ, et al. Detecting myocardialinfarction in critical illness using screening troponin measurements andECG recordings. Crit Care 2008;12:R36.

[12] Babuin L, Vasile VC, Rio Perez JA, et al. Elevated cardiac troponin isan independent risk factor for short- and long-term mortality inmedical intensive care unit patients. Crit Care Med 2008;36:759-65.

[13] Stein R, Gupta B, Agarwal S, et al. Prognostic implications of normal(b0.10 ng/ml) and borderline (0.10 to 1.49 ng/ml) troponin elevationlevels in critically ill patients without acute coronary syndrome. Am JCardiol 2008;102:509-12.

[14] Ilva TJ, Eskola MJ, Nikus KC, et al. The etiology and prognosticsignificance of all-cause troponin I positivity in emergency departmentpatients. J Emerg Med 2008 [epub ahead of print].