This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
CARDIOLOGY/ORIGINAL RESEARCH
Development of a Clinical Prediction Rule for 30-Day CardiacEvents in Emergency Department Patients With Chest Pain and
Possible Acute Coronary SyndromeErik P. Hess, MD, MSc, Robert J. Brison, MD, Jeffrey J. Perry, MD, MSc, Lisa A. Calder, MD, MSc,Venkatesh Thiruganasambandamoorthy, MD, MSc, Dipti Agarwal, MBBS, Annie T. Sadosty, MD,
Marco L. A. Silvilotti, MD, MSc, Allan S. Jaffe, MD, Victor M. Montori, MD, MSc,George A. Wells, PhD, Ian G. Stiell, MD, MSc
From the Department of Emergency Medicine, Division of Emergency Medicine Research (Hess, Agarwal, Sadosty) and the Department of InternalMedicine, Division of Cardiology (Jaffe) and Division of Endocrinology and Nutrition (Montori), Mayo Clinic College of Medicine, Rochester, MN; theKnowledge and Evaluation Research Unit, Rochester, MN (Hess, Montori); the Department of Emergency Medicine, Queens University, Kingston,
Ontario, Canada (Brison, Silvilotti); and the Department of Emergency Medicine (Perry, Calder, Thiruganasambandamoorthy, Stiell) andDepartment of Epidemiology and Community Medicine (Perry, Calder, Thiruganasambandamoorthy, Wells, Stiell), University of Ottawa, Ottawa,
Ontario, Canada.
Study objective: Evaluation of emergency department (ED) patients with chest pain who are at low risk for acutecoronary syndrome is resource intensive and may lead to false-positive test results and unnecessarydownstream procedures. We seek to identify patients at low short-term risk for a cardiac event for whomadditional ED investigations might be unnecessary.
Methods: We prospectively enrolled patients older than 24 years and with a primary complaint of chest pain from 3academic EDs. Physicians completed standardized data collection forms before diagnostic testing. The primaryadjudicated outcome was acute myocardial infarction, revascularization, or death of cardiac or unknown cause within30 days. We used recursive partitioning to derive the rule and validated the model with 5,000 bootstrap replications.
Results: Of 2,718 patients enrolled, 336 (12%) experienced a cardiac event within 30 days (6% acute myocardialinfarction, 10% revascularization, 0.2% death). We developed a rule consisting of the absence of 5 predictors:ischemic ECG changes not known to be old, history of coronary artery disease, pain typical for acute coronarysyndrome, initial or 6-hour troponin level greater than the 99th percentile, and age greater than 50 years. Patientsaged 40 years or younger required only a single troponin evaluation. The rule was 100% sensitive (95% confidenceinterval 97.2% to 100.0%) and 20.9% specific (95% confidence interval 16.9% to 24.9%) for a cardiac event within30 days.
Conclusion: This clinical prediction rule identifies ED chest pain patients at very low risk for a cardiac event whomay be suitable for discharge. A prospective multicenter study is needed to validate the rule and determine itseffect on practice. [Ann Emerg Med. 2012;59:115-125.]
Please see page 116 for the Editor’s Capsule Summary of this article.
Provide feedback on this article at the journal’s Web site, www.annemergmed.com.A podcast for this article is available at www.annemergmed.com.
Chest pain is the second most common presenting complaintamong adults presenting to emergency departments (EDs)across the United States, accounting for more than 6 millionpatient visits annually.1 Contemporary data suggest that 2.1%of patients with acute myocardial infarction receive
misdiagnoses, with a reported 30-day mortality of 9.1%.2 As a i
Volume , . : February
esult, physicians initiate formal diagnostic testing for acuteoronary syndrome at a very low risk threshold.3 Large numbersf low-risk patients undergo prolonged evaluations that result inalse-positive test results, unnecessary downstream procedures,nd increased health care costs.
mportanceA number of clinical prediction rules have been developed that
dentify patients at low risk for acute coronary syndrome.4-11 Of
A Prediction Rule for 30-Day Cardiac Events Hess et al
the published clinical prediction rules, none have adhered to themethodological standards for clinical prediction rules in theirdevelopment and validation,12-14 have performed well whenvalidated in other clinical settings,15 or have used contemporarycardiac troponin assays as advocated by the latest consensusdefinition.16 In the absence of a highly sensitive and reliableclinical prediction rule, a recent American Heart AssociationScientific Statement recommends that hemodynamicallystable patients with possible acute coronary syndrome and noobjective evidence of ischemia (normal or near-normal ECGresult and negative baseline cardiac injury markers) beadmitted to a chest pain unit for further evaluation by anaccelerated diagnostic protocol, including cardiac stresstesting.17 In describing the role of risk scores such as theThrombolysis in Myocardial Infarction (TIMI) or GlobalRegistry of Acute Coronary Events (GRACE), it mentionsthat favorable (low-risk) scores facilitate identification ofpatients for whom chest pain unit management might beappropriate. The current investigation seeks to augmentthese recommendations by developing a clinical predictionrule that identifies ED patients at very low risk for 30-daycardiac events who might be safe for discharge without chestpain unit admission, provided outpatient follow-up can be
Editor’s Capsule Summary
What is already known on this topicEmergency physicians have a low threshold foradmitting patients with potential acute coronarysyndromes.
What question this study addressedThis study developed a clinical decision rule foridentification of low-risk patients.
What this study adds to our knowledgePatients younger than 50 years and with atypicalchest pain, a nonischemic ECG result, and 2negative troponin test results during 6 hourswithout known coronary disease are at low risk fordeath, myocardial infarction, or revascularizationwithin 30 days (sensitivity 100% [95% confidenceinterval 97.2% to 100%] and specificity 20.9%[95% confidence interval 16.9% to 24.9%]).
How this is relevant to clinical practiceIt is unlikely this rule, even if validated, wouldchange clinical practice because it formally codifieswhat has probably been common practice for a longtime.
arranged within 30 days of the ED visit. s
116 Annals of Emergency Medicine
oals of This InvestigationThis study’s goal was to derive and internally validate a
linical prediction rule to identify adults with chest pain and atery low risk for a cardiac event, for whom additional EDnvestigations might be unnecessary.
ATERIALS AND METHODStudy Design
The study had a prospective observational cohort design.atients were enrolled from July 2007 to February 2010.
ettingThe study was conducted in 3 academic EDs located in
anada and the United States. The Civic Campus of thettawa hospital, located in Ottawa, Ontario, Canada, has
0,000 annual ED visits. The Kingston General Hospital isocated in Kingston, Ontario, Canada, and has 53,000 annualD visits. The Saint Marys hospital is located in Rochester,N, and has 73,000 annual ED visits.
election of ParticipantsWe enrolled consecutive adults older than 24 years and with
primary complaint of anterior chest pain, for whom cardiacroponin testing was ordered by the treating physician as aomponent of an assessment for acute coronary syndrome. Weefined anterior chest pain as any discomfort located on thenterior thorax between the suprasternal notch and the xiphoidrocess, using the posterior axillary line as the border betweenhe anterior and posterior thorax. Patients who had acute ST-egment elevation on the initial ECG in at least 2 contiguouseads, hemodynamic instability (pulse rate persistently greaterhan 100 or less than 50 beats/min or systolic blood pressureersistently below 90 mm Hg), an unreliable clinical history asetermined by the treating physician, a clear traumatic cause ofhest pain, a clinical history of cocaine use or positive test resultor cocaine, a terminal noncardiac illness, pregnancy, previousnrollment within the past 30 days, or inability to receiveollow-up by telephone were excluded. A study nurse assessedligibility in collaboration with the treating physician andbtained informed consent. The institutional review board atach site approved all study procedures.
In the Ottawa and Kingston, Ontario, sites, cardiologyonsultation is obtained at the discretion of the attendingmergency physician for patients considered to be at moderater high risk for acute coronary syndrome. The consultingardiologist determines whether admission to an inpatientbservation unit, hospital ward, or cardiac ICU is indicated on aase-by-case basis. Patients considered to be at low risk for acuteoronary syndrome are discharged for outpatient follow-up oncehe ED evaluation for acute myocardial infarction is complete.n the Rochester, MN, site, patients at high risk for acuteoronary syndrome are admitted to the hospital ward or cardiacCU, and patients at low to moderate risk for acute coronary
yndrome are admitted to an ED-based observation unit, where
Volume , . : February
i(
IlvRtr1boes
O
mco
tittaloRpa
baa1ac
trfschaptf
tmt
Hess et al A Prediction Rule for 30-Day Cardiac Events
serial cardiac troponin levels, serial ECGs, and cardiac stresstesting are obtained according to protocol.18 Patients who ruleout for acute myocardial infarction and have no evidence ofischemia on serial ECGs and cardiac stress testing are dischargedfrom the observation unit for outpatient follow-up.
Data Collection and ProcessingThe 64 variables selected for assessment in the study were
chosen according to review of the literature, input from theinvestigators, and feedback solicited from physician assessors. Acomplete list of the variables selected can be found elsewhere.19
Variable definitions were determined according to standardizedreporting guidelines.20 The study was designed before thepublication of the reporting guidelines for acute coronarysyndrome research in EDs in Australasia.21
On-duty attending or supervised emergency medicineresident physicians performed patient assessments andcompleted a standardized data collection form before obtainingresults of diagnostic tests (other than the initial ECG).Interobserver assessments were conducted when feasible by anindependent attending physician blinded to the first physician’sdata collection form. After enrollment, a study nurse reviewedthe hospital record to collect data on the cardiovascular history,cardiac risk factors, and all investigations related to the visit.Data were recorded on predesigned case record forms. Each siteinvestigator monitored local compliance with patientenrollment procedures throughout the duration of the study.
We defined the time of symptom onset as the time of onsetof the episode of pain that prompted the patient’s visit to theED. Clinicians categorized the period from the onset ofsymptoms to ED presentation into one of 3 categories: today(�24 hours), yesterday (24 to 48 hours), and greater than orequal to 2 days ago. If the pain developed within the past 24hours, the specific time of onset was recorded. Known coronaryartery disease was defined as at least one 50% stenosis oncoronary angiography; demonstrated ECG changes, perfusiondefects, or wall motion abnormalities on exercise,pharmacologic, or rest imaging studies; or previousdocumentation of acute myocardial infarction. Patients whoreported a history of coronary artery disease during their EDvisit that was not confirmed by objective criteria in the medicalrecord were classified as having a self-reported history ofcoronary artery disease.20 We instructed clinicians to classify thecause of the chest pain syndrome with the following method: if,after taking into account all the factors from the history andphysical examination, the clinician considered the painsyndrome to be of cardiac cause, it was classified as typical. Ifthe clinician’s overall assessment was that the pain syndromewas of noncardiac cause, it was classified as atypical.
Blinded to predictor variables and patient outcome, 2 board-certified emergency physicians (E.P.H., R.J.B.) classified allECGs with a structured standardized format20 previouslydemonstrated to correlate with 30-day outcomes in ED patientswith chest pain.22 A subset of 80 randomly selected ECGs (40
interpreted as acute ischemia and 40 not interpreted as acute t
Volume , . : February
schemia) was independently interpreted by a third investigatorA.S.J.), and interrater reliability was assessed.
Cardiac troponin T (Elecsys; Roche Diagnostics,ndianapolis, IN) (99th percentile reference limit �0.01 �g/L;ower limit of detection 0.01 ng/mL; 10% coefficient ofariation 0.035 ng/mL) was the cardiac biomarker used at theochester and Ottawa sites. At the Kingston site, cardiac
roponin I (Beckman Coulter, Brea, CA) (99th percentileeference limit 0.04 �g/L; lower limit of detection 0.01 �g/L;0% coefficient of variation 0.06 �g/L23) was the cardiaciomarker in routine clinical use. Serum for cardiac troponin Tr I was to be obtained at patient arrival and greater than orqual to 6 hours from pain onset, with at least 3 hours betweenamples.16,24
utcome MeasuresWe defined the primary composite outcome to be acute
yocardial infarction, coronary revascularization, or death ofardiac or unknown cause within 30 days, whether the outcomeccurred in the ED, in the hospital, or after discharge.
Acute myocardial infarction was defined as a cardiacroponin measurement above the 99th percentile, with anncreasing or decreasing pattern on serial measurements (forroponin I, defined as a �20% change between samples24; forroponin T, significant change defined according to anlgorithm developed in the Mayo Clinic cardiovascularaboratory), development of pathologic Q waves on the ECG,r ECG evolution consistent with acute myocardial infarction.16
evascularization was defined as reestablishment of coronaryatency by percutaneous coronary intervention or coronaryrtery bypass graft surgery.
The primary outcome was ascertained by investigatorslinded to the predictor variables. If a diagnosis could not bessigned, a cardiologist (A.S.J.) reviewed all the clinical data andssigned an adjudicated outcome diagnosis. All positive and0% of randomly selected negative outcomes were confirmed bysecond coinvestigator blinded to the standardized data
ollection forms. Disagreements were resolved by consensus.All management decisions were left to the discretion of the
reating physician. Because not all patients with chest painoutinely undergo coronary angiography, structured telephoneollow-up was conducted for all subjects by a study nurse,tarting at 30 days. Patients were asked whether they hadonsulted another physician, had experienced a heart attack, orad undergone any cardiac testing, including coronaryngiography with or without invasive intervention. If theatient answered in the affirmative or was unsure of the answero any of these questions, additional documentation was soughtrom the facility in which any procedures were performed.
If the study nurse was unable to contact the patient byelephone, outcomes were ascertained by review of the hospitaledical record. For those subjects unable to be contacted by
elephone and with no subsequent hospital visits, we reviewed
he provincial coroner’s database (Ottawa and Kingston sites) or
Annals of Emergency Medicine 117
ahEdhw
R
r(KO23crttdtpo1iruaaDepaau
Fs
A Prediction Rule for 30-Day Cardiac Events Hess et al
the Social Security Death Index25 (Rochester site) to ascertainvital status.
Primary Data AnalysisWe measured interobserver agreement for each variable by
calculating the � coefficient for dichotomous or categoricalvariables and the intraclass correlation coefficient26 forcontinuous variables. Variables with � values greater than orequal to 0.6 were considered to represent substantial agreementand were considered for inclusion in the rule. � Values were notcalculated for variables collected from the medical record (eg,age, increased troponin level, history of coronary artery disease).We further explored variables of potential interest (either on astatistical or clinical basis) by using cut points to ensure that thefinal model did not contain any continuous variables so thatclinicians could categorize patients as high or low risk withoutperforming any calculations.
To aid in selection of the best variables for multivariableanalyses, we used univariable logistic regression to determine thestrength of association between each variable and the compositeprimary outcome. Variables found to be both reliable (��0.6)and associated with the primary outcome (��.20) wereevaluated with recursive partitioning. The objective was toidentify the best combination of predictor variables that washighly sensitive for the primary outcome, with the maximumpossible specificity and clinical sensibility. Recursive partitioningcreates a branching decision tree by dividing the patientpopulation into subgroups with and without the outcome ofinterest, according to contents of predictor variables in thesubgroup. Because our previous experience suggested thatrecursive partitioning may be more suitable than logisticregression when the intent is to classify one outcome at theexpense of the other (when high sensitivity is preferred to overallaccuracy), we performed recursive partitioning with AnswerTree software (version 3.1; IBM SPSS, Chicago, IL). Thevariables chosen by the best model constituted the decision rule.
We cross-validated the derived decision rule by comparingthe classification of each patient to the actual status for theprimary outcome, allowing us to estimate the sensitivity andspecificity of the rule. Bootstrapping was used to calculate 95%confidence intervals (CIs) for both the sensitivity and specificityestimates.27 For bootstrapping, 5,000 samples of 500observations were generated, and the mean CIs for bothestimates were reported. Because the sensitivity estimate wasclose to 1, a continuity correction was calculated for the CI, anda 1-sided test was used. To determine the 95% CI forproportions with correction for continuity, we used the methoddescribed by Newcombe.28
We estimated the sample size according to the precision ofthe sensitivity of the derived decision rule. As with previousdecision rule studies, we prespecified 120 outcome events toderive a rule that is 100% sensitive, with a lower 95%confidence limit of 97.0%.29,30 To have the greatest utility forpracticing emergency physicians, we aimed to include at least
120 outcome events occurring outside the ED (inhospital or c
118 Annals of Emergency Medicine
fter ED discharge). Review of quality data from the Ottawaospital indicated that 10% of patients who presented to theD with chest pain would meet outcome criteria within 30ays. We estimated that half of these events would occur afterospital admission or ED discharge. The a priori sample sizeas estimated to be 2,400 patients.
ESULTSWe enrolled 2,718 patients from 3 academic EDs (Figure 1),
epresenting 82% of eligible patients. There were 1,01737.4%) patients enrolled in Ottawa, Ontario, 496 (18.2%) iningston, Ontario, and 1,205 (44.3%) in Rochester, MN.utcome status was ascertained at 30 days for all patients;
,629 (97%) were able to be contacted by telephone for the0-day structured follow-up, 58 (2%) had their vital statusonfirmed by review of subsequent visits in the hospital medicalecord, and the vital status of 31 (1%) was assessed by searchinghe provincial coroner’s database (Ottawa and Kingston sites) orhe Social Security Death Index (Rochester site). There were 12isagreements in outcome ascertainment between investigatorshat required adjudication by a third investigator. Of allatients, 336 (12%) met at least 1 component of the compositeutcome within 30 days of the ED visit (Table 1). There were67 (6%) patients who met criteria for acute myocardialnfarction, 271 (10%) who underwent coronaryevascularization, and 6 (0.2%) who died of a cardiac ornknown cause. The frequency of acute myocardial infarctionnd death was similar between the 3 study sites (Table E1,vailable online at http://www.annemergmed.com).emographic and cardiovascular history characteristics of the
nrolled patients were similar to those of the eligible missedatients (Table E2, available online at http://www.nnemergmed.com). Table E3 (available online at http://www.nnemergmed.com) details the proportion of patients whonderwent cardiac diagnostic investigations to evaluate for acute
igure 1. Flow diagram of multicenter prospective cohorttudy of ED patients with chest pain.
Hess et al A Prediction Rule for 30-Day Cardiac Events
Volume , . : February
Table 2 shows the association between the predictor variablesnd the composite outcome, as determined by univariablenalyses and the interobserver agreement for predictor variablesbtained from the chest pain history and ECG. Of the 250nterobserver cases, the pain began in less than 24 hours in 124ases. The intraclass correlation coefficient for time of onset was.916, indicating excellent interrater agreement. In general,atients with subsequent events had more previous cardiacvents, had more risk factors by history, and were more apt toave abnormal ECG findings.
The model developed with recursive partitioning analysis isisplayed in Figure 2. The variables in this final statistical modelere then combined into a simple, clinically sensible list toresent the decision rule, titled the North American Chest Painule (Box). The classification performance of the rule wasccurate across 5,000 bootstrapped test sets and was consistentith the estimation from the original data set. Table 3 shows
he classification performance of the model for predicting acuteyocardial infarction, revascularization, or death within 30
ays, using the bootstrapped 95% CI estimates. Figure 3Ahows the performance of the prediction rule at an age cutoff ofged 50 years or younger. In this model, patients aged 40 yearsr younger and who otherwise have none of the risk factors cane safely discharged from the ED if the initial cardiac troponinesult is negative. This group was represented in 7% of theohort. Patients aged 41 to 50 years require a second troponinest, obtained at least 6 hours from pain onset, and can beischarged if the troponin measurement is negative and nother criteria are met. This group was represented in 11% of theohort.
If the age cutoff is increased to 60 years or younger, thepecificity of the rule increases to 34% and 4 positive outcomesere missed. Each of these cases was considered to be an
typical presentation by the attending physician, had no newschemic changes on ECG, and had negative serial cardiacroponin results. One patient developed ST-segment elevationn the ECG component of a dobutamine stress echo, had a
able 1. Continued.
haracteristicNo. (%) of Patients
(n�2,718)*
Acute myocardial infarction 8 (0.3)Revascularization 171 (6.3)Death of cardiac/unknowncause
4 (0.2)
QR, Interquartile range.Unless otherwise indicated.Among the 892 patients with known coronary artery disease, 666 (74.7%) wereonfirmed with objective criteria in the medical record and 226 (25.3%) werescertained by patient self-report.Only 1 site (Rochester, MN) had an observation unit.In at least 1 major epicardial coronary artery.Acute myocardial infarction, revascularization, or death of cardiac/unknownause within 30 days of the ED visit. Some patients experienced more than oneype of cardiac event.
Table 1. Baseline patient characteristics, ECG findings, and30-day outcomes for 2,718 ED patients with chest pain.
CharacteristicNo. (%) of Patients
(n�2,718)*
DemographicsMean age, y (SD) 60.0 (14.9)Age range, y 25–99Male patient 1,439 (52.9)Arrival by ambulance 661 (24.3)Median length of stay in
department, h (IQR)3.0 (3.1)
Range 0.4–16.3Cardiac risk factors
Hypertension 1,592 (58.6)Diabetes mellitus 559 (20.6)Hypercholesterolemia 1,458 (53.6)Family history of cardiac disease 784 (28.8)Smoking history 1,393 (51.3)Current 483 (17.8)Former (�1 y ago) 155 (5.7)Former (�1 y ago) 755 (27.8)No cardiac risk factors 274 (10.1)
Peripheral vascular disease 133 (4.9)No cardiovascular history 1,395 (51.3)
Initial ECG interpretationNormal (%) 1,169 (43.0)Nonspecific ST-segment changes 599 (22.0)Abnormal, not diagnostic 578 (21.3)Ischemia known to be old 241 (8.9)Ischemia not known to be old 131 (4.8)
ManagementAdmitted to the hospital 865 (31.8)Admitted to the observation
unit‡
(n�1,205)577 (47.9)
Cardiac stress testing performed 896 (33.0)Coronary angiography performed 510 (18.8)Coronary stenosis �70%
§329 (12.1)
OutcomesCardiac event
�336 (12.4)
Acute myocardial infarction 167 (6.1)Revascularization 271 (10.0)Death of cardiac/unknown
cause6 (0.2)
Cardiac event occurring after EDdisposition (inhospital or out-
175 (6.4)
0% lesion in the right posterior descending artery identified on
Annals of Emergency Medicine 119
A Prediction Rule for 30-Day Cardiac Events Hess et al
Table 2. Univariate association between demographic, clinical, and ECG predictor variables and the primary outcome in 2,718 EDpatients with chest pain.
Variable
Group, No. (%) of Patients*InterobserverAgreement,
� Value (n�250)Positive for PrimaryOutcome (n�336)
Associated symptomsNausea or vomiting 71 (21.1) 550 (23.0) 0.89 (0.68–1.18) 0.54Shortness of breath 158 (47.0) 923 (38.8) 1.40 (1.12–1.77 0.42Diaphoresis 95 (28.3) 496 (20.8) 1.50 (1.16–1.94) 0.61Worse with movement 50 (14.9) 481 (20.2) 0.69 (0.50–0.95) 0.43Typical for acute coronary syndrome
‡250 (74.4) 778 (32.6) 5.99 (4.62–7.77) 0.75
Atypical for acute coronary syndrome‡
79 (23.5) 1,527 (64.1) 0.17 (0.13–0.22) 0.75Initial interpretation of the ECG
§(n for ��80) 0.85
Normal (%) 84 (25.0) 1,076 (45.2) 0.40 (0.31–0.51) NA
120 Annals of Emergency Medicine Volume , . : February
cther
bvstticodvmctcaddstnmcas
D
Hess et al A Prediction Rule for 30-Day Cardiac Events
coronary angiography that received a stent, and subsequentlyexperienced a postrevascularization non–ST-segment acutemyocardial infarction. The other 3 patients underwentpercutaneous coronary intervention and did not experiencemyocardial infarction or death.
LIMITATIONSOur study has several potential limitations. We enrolled only
patients who presented with a chest pain syndrome. Patients atrisk for acute coronary syndrome who presented with non–chestpain syndromes such as shortness of breath, nausea, orgeneralized fatigue were not included, which limits theapplication of our findings to those who have chest pain as apresenting symptom. Risk stratification of patients with possibleacute coronary syndrome who present with non–chest painsyndromes should proceed as otherwise indicated in standardpractice. Not all (82%) eligible patients were enrolled, likelybecause they were missed when the ED was busy. We did,however, collect demographic and cardiovascular historycharacteristics of eligible missed patients, and they were similarto those of eligible enrolled patients.
Investigators, blinded to predictor variables and outcomes,interpreted ECGs. If the prediction rule were to be applied inclinical practice, the treating physician would interpret ECGs inthe course of clinical care. However, the study investigators whointerpreted the ECGs were blinded to both the predictorvariables and outcome at the ECG interpretation and are board-certified, practicing emergency physicians with training similarto that of those who would apply the rule in clinical practice.
We did not include patients with potential acute coronarysyndrome who were medically managed as a component of theprimary outcome. As such, the prediction rule may not identifysome patients who may benefit from pharmacologic
Table 2. Continued.
VariablePositive for PrimOutcome (n�336
Nonspecific ST-segment changes (%) 67 (19.9)Abnormal but not diagnostic of ischemia (%) 82 (24.4)Infarction or ischemia known to be old (%)
�36 (10.7)
Infarction or ischemia not known to be old (%) 67 (19.9)Other ECG findings
*Unless otherwise indicated.†The primary outcome was acute myocardial infarction, coronary revascularization‡Attending physicians were instructed to record whether the chest pain syndromecoronary syndrome) in cause.§� Was dichotomously assessed according to whether ECGs were classified as in�Infarction or ischemia on ECG was defined as T-wave inversion greater than or eqcontiguous leads.
management. All enrolled patients did not undergo definitive r
Volume , . : February
ardiac testing such as coronary angiography or cardiac stressesting introducing the potential for evaluation bias. We did,owever, conduct structured telephone follow-up for allnrolled patients as a proxy outcome measure to attenuate theisk of evaluation bias.
In deriving the rule, we assessed univariate associationsetween all potentially important predictors and includedariables that were both associated with the outcome and ofufficient interrater reliability in multivariate analyses. Althoughhis approach to prediction rule development involves multipleesting and introduces the possibility of overfitting, we thoughtt important that all potentially important predictors thatlinicians might consider important be evaluated in the processf deriving the rule. In addition, clinical prediction ruleevelopment involves several phases, including derivation,alidation, and implementation, and the robustness of theodel is tested in each phase. Some may question including
oronary revascularization as an outcome because the decisiono treat a coronary stenosis may vary between interventionalardiologists and between regions of the country. However, thebsence of acute ischemic ECG changes and troponin increasesoes not necessarily indicate the absence of ischemic heartisease. Cardiac ischemia may not be severe enough to causeufficient myocardial damage to release detectable quantities ofroponin,31 and the initial ECG result may be normal oronspecific, even in cases that ultimately meet criteria for acuteyocardial infarction.32 Despite the user-dependent nature of
oronary revascularization, excluding coronary revascularizations an outcome criterion may miss cases of acute coronaryyndrome.
ISCUSSIONWe successfully developed a highly sensitive clinical decision
roup, No. (%) of Patients*InterobserverAgreement,
� Value (n�250)Negative for PrimaryOutcome (n�2,382)
100 (4.2) 1.29 (0.77–2.16) NA208 (8.7) 1.37 (0.96–1.97) NA
eath within 30 days of the ED visit.cardiac (typical for acute coronary syndrome) or noncardiac (atypical for acute
n or ischemia not known to be old (acute ischemia) or “other.”0.2 mm or ST-segment depression greater than or equal to 1 mm in at least 2
G
ary)*
, or dwas
farctioual to
ule to identify patients at very low risk for acute myocardial
Annals of Emergency Medicine 121
pp
ETwhpstrrcte
cisi
A Prediction Rule for 30-Day Cardiac Events Hess et al
infarction, revascularization, or death within 30 days of the EDvisit. This rule would allow emergency physicians to determinewhich patients require additional ED investigations and whichcan be safely discharged for further outpatient evaluation.Patients who have none of the risk factors in the decision ruleand are aged 40 years or younger can be safely discharged if theinitial cardiac troponin level is less than the 99th percentilereference limit; patients aged 41 to 50 years can be discharged ifa second cardiac troponin value less than the 99th percentile isobtained at least 6 hours from pain onset. Patients aged 60 yearsor younger have a very low event rate and can be considered fordischarge in practice settings in which a less than 2% miss rate isacceptable, depending on local resources. To the best of theauthors’ knowledge, this is the largest clinical decision rule studyin ED patients with chest pain that has been conducted strictly
Figure 2. Derivation of clinical de
according to the methodological standards for clinical w
122 Annals of Emergency Medicine
rediction rules. Nevertheless, the prediction rule requiresrospective multicenter validation before use in clinical practice.
Several prediction models have been developed to identifyD chest pain patients who are at low risk for a cardiac event.he TIMI risk score, which was initially developed in patientsith unstable angina or non–ST-segment myocardial infarction,as been broadly validated in several ED populations with chestain.33 The pooled sensitivity of the TIMI risk score in a recentystematic review was 97%, the pooled specificity was 25%, andhe miss rate of patients with a score of zero was 2%. This missate is likely too high for use in the US setting. Than et al34
ecently reported the performance of the TIMI risk scoreombined with a point-of-care multibiomarker panel usingroponin, creatine kinase MB, and myoglobin. Their protocolfficiently identified 418 of 421 major adverse cardiac events
on rule by recursive partitioning.
ithin 2 hours of presentation (sensitivity 99.3%; specificity
Volume , . : February
ca
tEnnwccishHid
F
eofcp
tm
ahd
ups
Hess et al A Prediction Rule for 30-Day Cardiac Events
11.0%) and identified 10% of the cohort with a TIMI risk scoreof zero who could be safely discharged from the ED. Ourclinical decision rule identified 18% of a cohort who could besafely discharged from the ED, using only contemporary cardiactroponin assays, and as more sensitive cardiac troponin assaysare developed this proportion is likely to increase.35
The History, ECG, Age, Risk factors and Troponin(HEART) score has been retrospectively derived36 andvalidated37 in ED patients with chest pain. The score uses datafrom the history, ECG, age, cardiac risk factors, and troponinlevel to predict the short-term risk of acute myocardialinfarction, revascularization, and death. In the derivation phase,2.5% of patients classified as low risk (HEART score 0 to 3) hada cardiac event within 3 months. In the validation phase, thesensitivity of a HEART score of 0 to 3, however, was 98.1%and the specificity was 15.5%. These are less optimal prognosticaccuracy estimates compared with those observed in the currentinvestigation.
The Vancouver Chest Pain Rule, which was developedaccording to the methodological standards for clinicalprediction rules, also identified patients younger than 40 yearsas a low-risk group appropriate for discharge from the ED.8 Forpatients older than 40 years, these investigators found thatpatients without certain chest pain characteristics (pain radiatingto neck, arm, or jaw and pain does not increase with deepbreath or palpation) could also be classified as low risk andconsidered for ED discharge. In our study, however, whichincluded 250 cases in which the interobserver reliability ofpredictor variables was assessed, only radiation of pain to theback or abdomen had a � of at least 0.6; radiation of pain toother areas was insufficiently reliable to be considered for
Table 3. Prognostic accuracy of clinical decision rule for 30-day cardiac events in 2,718 patients with chest pain andpossible acute coronary syndrome.
A, Age cutoff 50 years or younger
Decision RuleCardiac Event
Within 30 DaysNo Cardiac EventWithin 30 Days
Yes 336 1,885No 0 497
Sensitivity 100.0% (95% CI 97.2% to 100.0%); specificity 20.9% (95% CI 16.9%to 24.9%); positive predictive value 15.1% (95% CI 13.7% to 16.7%); negativepredictive value 100.0% (95% CI 99.0% to 100.0%); stress testing proportion81.7% (95% CI 80.2% to 83.1%).
B, Age cutoff 60 years or younger
Decision RuleCardiac Event
Within 30 DaysNo Cardiac EventWithin 30 Days
Yes 332 1,581No 4 801
Sensitivity 98.8% (95% CI 95.4% to 100.0%); specificity 33.6% (95% CI 28.9%to 38.1%); positive predictive value 17.4% (95% CI 15.7% to 19.1%); negativepredictive value 99.5% (95% CI 98.6% to 99.8%); stress testing proportion70.4% (95% CI 68.6% to 72.1%).
inclusion in the rule. We also used a more sensitive marker of n
Volume , . : February
ardiac injury, cardiac troponin, rather than creatine kinase MB,s a predictor variable in our study.
Marsan et al10 reported findings similar to those observed inhe current investigation. These investigators observed that forD chest pain patients younger than 40 years and with aormal ECG result, an initial normal cardiac biomarker result,o cardiac risk factors, and no abnormal cardiac history, thereere no cardiac events at 30 days. Collin et al38 reported no
ardiac events at 1 year for a cohort of 609 patients with similarharacteristics. Overall, the observations in the currentnvestigation add to the growing literature that urgent cardiactress testing in ED chest pain patients younger than 40 yearsas limited utility and is likely an inefficient use of resources.owever, in our population we also observed that a physician
mpression of atypical or noncardiac chest pain had a highegree of interobserver reliability (��0.75), had a strong
igure 3. North American Chest Pain Rule.**Does not apply to patients with acute ST-segment
levation, hemodynamic instability, a clear traumatic causef chest pain, a history of cocaine use or tested positiveor cocaine, a terminal noncardiac illness, or who wereonsidered to have an unreliable history by the treatinghysician.
†Defined as ST-segment deviation greater than or equalo 1 mm or T-wave inversion greater than or equal to 0.2m in at least 2 contiguous leads.‡As determined by the attending emergency physician.§Samples for contemporary cardiac troponin assay
nalysis were obtained at presentation and at least 6ours from symptom onset in this cohort. Positive wasefined as greater than 99th percentile reference limit.
�In addition, an age cutoff of 60 years or younger can besed with a less than 2% miss rate and lower proportion ofatients who require urgent stress testing in practiceettings in which this miss rate is acceptable.
egative association with 30-day cardiac events (odds ratio 0.14;
Annals of Emergency Medicine 123
sO
PRAS
Et
Ah
R
1
1
1
1
1
1
A Prediction Rule for 30-Day Cardiac Events Hess et al
95% CI 0.13 to 0.21), and, when combined with findingsroutinely available from the history and physical examination,ECG, and troponin level, assists in identifying patients aged 50years or younger who are at very low risk for a cardiac eventwithin 30 days.
Our prediction rule may differentially affect resource use,depending on local practice patterns. In the Ottawa andKingston sites, the hospital admission rates were 18% and 21%,respectively, and neither of these sites have an ED observationunit. Use of the prediction rule in these sites would likelyincrease resource use. Using the age cutoff of 60 years oryounger may attenuate the effect on resource use and could beanticipated to have a miss rate of less than 2%. In the Rochestersite, however, the hospital admission rate was 48% and the EDobservation unit admission rate was 48% (combined admissionrate of 96%). At the age cutoff of 50 years or younger, theestimated testing rate was 82%. Use of the rule at the Rochestersite would likely decrease resource use.
The safety and potential utility of our prediction ruledepends on the availability of outpatient follow-up within 30days of the ED visit. Some patients may not have ready access toa primary care provider. Additional ED investigations may beappropriate for patients with unreliable access to follow-up or inhealth care settings in which outpatient follow-up is not readilyavailable.
In summary, we derived and internally validated a clinicalprediction rule that identified adults with chest pain at very lowrisk for a cardiac event for whom additional ED investigationsmay be unnecessary. Application of the rule has potential toincrease the efficiency of diagnostic investigation; patients whodo not meet any of the high-risk criteria could be safelydischarged from the ED, and patients with any of the riskfactors in the rule could be considered for additionalinvestigation or observation unit admission. Prospectivemulticenter validation is needed before use in clinical practice.
Supervising editor: Judd E. Hollander, MD
Author contributions: All authors contributed significantly tothe study and the preparation of the article. EPH, JJP, ASJ,GAW, and IGS were responsible for the conception and designof the study. LAC, VT, DA, and ATS assisted with dataacquisition. EPH and RJB were responsible for dataacquisition, data entry, and preliminary analysis. EPH andGAW performed all statistical analyses. All authorsparticipated in the drafting and multiple revisions of the articleand gave final approval of the submitted work. EPH takesresponsibility for the paper as a whole.
Funding and support: By Annals policy, all authors are requiredto disclose any and all commercial, financial, and otherrelationships in any way related to the subject of this articleas per ICMJE conflict of interest guidelines (seewww.icmje.org). The study was funded by a Fellow-to-FacultyTransition award from the American Heart Association, theSociety for Academic Emergency Medicine, and the Emergency
Medicine Foundation (award 0775030N). The study was also
124 Annals of Emergency Medicine
upported by the Ontario Innovation Fund and the University ofttawa Department of Emergency Medicine.
ublication dates: Received for publication April 29, 2011.evisions received June 27, 2011, and July 5, 2011.ccepted for publication July 13, 2011. Available onlineeptember 1, 2011.
arn CME Credit: Continuing Medical Education is available forhis article at http://www.ACEP-EMedHome.com.
ddress for correspondence: Erik P. Hess, MD, MSc, [email protected].
EFERENCES1. Niska RW, Bhuiya F, Xu J. National Hospital Ambulatory Medical
Care Survey: 2007 emergency department summary. Natl HealthStat Rep. 2010;26:1-32.
2. Schull MJ, Vermeulen MJ, Stukel TA. The risk of missed diagnosisof acute myocardial infarction associated with emergencydepartment volume. Ann Emerg Med. 2006;48:647-655.
3. Pines JM, Szyld D. Risk tolerance for the exclusion of potentiallylife-threatening diseases in the ED. Am J Emerg Med. 2007;25:540-544.
4. Baxt WG, Skora J. Prospective validation of artificial neuralnetwork trained to identify acute myocardial infarction. Lancet.1996;347:12-15.
5. Goldman L, Cook EF, Brand DA, et al. A computer protocol topredict myocardial infarction in emergency department patientswith chest pain. N Engl J Med. 1988;318:797-803.
6. Goldman L, Weinberg M, Weisberg M, et al. A computer-derivedprotocol to aid in the diagnosis of emergency room patients withacute chest pain. N Engl J Med. 1982;307:588-596.
7. Selker HP, Beshansky JR, Griffith JL, et al. Use of the acutecardiac ischemia time-insensitive predictive instrument (ACI-TIPI)to assist with triage of patients with chest pain or othersymptoms suggestive of acute cardiac ischemia. A multicenter,controlled clinical trial. Ann Intern Med. 1998;129:845-855.
8. Christenson J, Innes G, McKnight D, et al. A clinical predictionrule for early discharge of patients with chest pain. Ann EmergMed. 2006;47:1-10.
9. Boersma E, Pieper KS, Steyerberg EW, et al. Predictors ofoutcome in patients with acute coronary syndromes withoutpersistent ST-segment elevation. Results from an internationaltrial of 9461 patients. The PURSUIT Investigators. Circulation.2000;101:2557-2567.
0. Marsan RJ Jr, Shaver KJ, Sease KL, et al. Evaluation of a clinicaldecision rule for young adult patients with chest pain. AcadEmerg Med. 2005;12:26-31.
1. Lyon R, Morris AC, Caesar D, et al. Chest pain presenting to theemergency department—to stratify risk with GRACE or TIMI?Resuscitation. 2007;74:90-93.
2. McGinn TG, Guyatt GH, Wyer PC, et al. Users’ guides to themedical literature, XXII: how to use articles about clinical decisionrules. Evidence-Based Medicine Working Group. JAMA. 2000;284:79-84.
3. Laupacis A, Sekar N, Stiell IG. Clinical prediction rules. A reviewand suggested modifications of methodological standards. JAMA.1997;277:488-494.
4. Stiell IG, Wells GA. Methodologic standards for the developmentof clinical decision rules in emergency medicine. Ann Emerg Med.1999;33:437-447.
5. Hess EP, Thiruganasambandamoorthy V, Wells GA, et al.
Diagnostic accuracy of clinical prediction rules to exclude acute
Hess et al A Prediction Rule for 30-Day Cardiac Events
coronary syndrome in the emergency department setting: asystematic review. CJEM. 2008;10:373-382.
16. Thygesen K, Alpert JS, White HD, et al. Universal definition ofmyocardial infarction. Circulation. 2007;116:2634-2653.
17. Amsterdam EA, Kirk JD, Bluemke DA, et al. Testing of low-riskpatients presenting to the emergency department with chest pain:a scientific statement from the American Heart Association.Circulation. 2010;122:1756-1776.
18. Cullen MW, Reeder GS, Farkouh ME, et al. Outcomes in patientswith chest pain evaluated in a chest pain unit: the Chest PainEvaluation in the Emergency Room study cohort. Am Heart J.2011;161:871-877.
19. Hess EP, Wells GA, Jaffe A, et al. A study to derive a clinicaldecision rule for triage of emergency department patients withchest pain: design and methodology. BMC Emerg Med. 2008;8:3.
20. Hollander JE, Blomkalns AL, Brogan GX, et al. Standardizedreporting guidelines for studies evaluating risk stratification of EDpatients with potential acute coronary syndromes. Acad EmergMed. 2004;11:1331-1340.
21. Cullen L, Than M, Brown AF, et al. Comprehensive standardizeddata definitions for acute coronary syndrome research inemergency departments in Australasia. Emerg Med Australas.2010;22:35-55.
22. Forest RS, Shofer FS, Sease KL, et al. Assessment of thestandardized reporting guidelines ECG classification system: thepresenting ECG predicts 30-day outcomes. Ann Emerg Med.2004;44:206-212.
23. Morrow DA, Rifai N, Sabatine MS, et al. Evaluation of the AccuTnIcardiac troponin I assay for risk assessment in acute coronarysyndromes. Clin Chem. 2003;49:1396-1398.
24. Macrae AR, Kavsak PA, Lustig V, et al. Assessing the requirementfor the 6-hour interval between specimens in the American HeartAssociation classification of myocardial infarction in epidemiologyand clinical research studies. Clin Chem. 2006;52:812-818.
25. Quinn J, Kramer N, McDermott D. Validation of the Social SecurityDeath Index (SSDI): an important readily-available outcomesdatabase for researchers. West J Emerg Med. 2008;9:6-8.
27. Efron B. The Jacknife, the Bootstrap and Other Resampling Plans.Philadelphia, PA: Society for Industrial & Applied Mathematics;
1982.
Volume , . : February
8. Newcombe RG. Two-sided confidence intervals for the singleproportion: comparison of seven methods. Stat Med. 1998;17:857-872.
9. Stiell IG, Wells GA, Vandemheen KL, et al. The Canadian C-SpineRule for radiography in alert and stable trauma patients. JAMA.2001;286:1841-1848.
0. Perry JJ, Stiell IG, Sivilotti MLA, et al. High risk clinicalcharacteristics for subarachnoid haemorrhage in patients withacute headache: prospective cohort study. BMJ. 2010;341:c5204.
1. Anderson JL, Adams CD, Antman EM, et al. ACC/AHA 2007guidelines for the management of patients with unstable angina/non-ST-elevation myocardial infarction: a report of the AmericanCollege of Cardiology/American Heart Association Task Force onPractice Guidelines (Writing Committee to Revise the 2002Guidelines for the Management of Patients With UnstableAngina/Non-ST-Elevation Myocardial Infarction) developed incollaboration with the American College of Emergency Physicians,the Society for Cardiovascular Angiography and Interventions, andthe Society of Thoracic Surgeons; endorsed by the AmericanAssociation of Cardiovascular and Pulmonary Rehabilitation andthe Society for Academic Emergency Medicine. J Am Coll Cardiol.2007;50:e1-157.
2. Welch RD, Zalenski RJ, Frederick PD, et al. Prognostic value of anormal or nonspecific initial electrocardiogram in acutemyocardial infarction. JAMA. 2001;286:1977-1984.
3. Hess EP, Agarwal D, Chandra S, et al. Diagnostic accuracy of theTIMI risk score in patients with chest pain in the emergencydepartment: a meta-analysis. CMAJ. 2010;182:1039-1044.
4. Than M, Cullen L, Reid CM, et al. A 2-h diagnostic protocol toassess patients with chest pain symptoms in the Asia-Pacificregion (ASPECT): a prospective observational validation study.Lancet. 2011.
5. Thygesen K, Mair J, Katus H, et al. Recommendations for the useof cardiac troponin measurement in acute cardiac care. EurHeart J. 2010;31:2197-2204.
6. Six AJ, Backus BE, Kelder JC. Chest pain in the emergency room:value of the HEART score. Neth Heart J. 2008;16:191-196.
7. Backus BE, Six AJ, Kelder JC, et al. Chest pain in the emergencyroom: a multicenter validation of the HEART score. Crit PathwCardiol. 2010;9:164-169.
8. Collin MJ, Weisenthal B, Walsh KM, et al. Young patients withchest pain: 1-year outcomes. Am J Emerg Med. 2011;29:265-
270.
Annals of Emergency Medicine 125
Tm
C
MRMP
K
CAtrial fibrillation (%) 232 (8.5) 42 (7.0)
A Prediction Rule for 30-Day Cardiac Events Hess et al
Table E1. Frequency of acute myocardial infarction and deathof cardiac or unknown cause at the 3 study sites.