Article, Cardiology

Missed myocardial infarctions in ED patients prospectively categorized as low risk by established risk scores

a b s t r a c t

Study objectives: Few studies have prospectively compared multiple cardiac risk prediction scores. We compared the rate of missed acute myocardial infarction (AMI) in chest pain patients prospectively categorized as low risk by Unstructured clinical impression, and by HEART, TIMI, GRACE, and EDACS scores, in combination with two negative contemporary cardiac troponins (cTn) available in the U.S. Methods: We enrolled 434 patients with chest pain presenting to one of seven emergency departments (ED). Risk scores were prospectively calculated and included the first two cTn. Low risk was defined for each score as HEART <= 3, TIMI <= 0, GRACE <= 50, and EDACS <= 15. AMI incidence was calculated for low risk patients and com- pared across scores using ?2 tests and C statistics.

Results: The patients’ median age was 57, 58% were male, 60% white, and 80 (18%) had AMI. The missed AMI rate in low risk patients for each of the scores when combined with 2 cTn were HEART 3.6%, TIMI 0%, GRACE 6.3%, EDACS 0.9%, and unstructured clinical impression 0%. The C-statistic was greatest for the EDACS score, 0.94 (95% CI, 0.92-0.97).

Conclusions: Using their recommended cutpoints and non high sensitivity cTn, TIMI and unstructured clinical im- pression were the only scores with no missed cases of AMI. Using lower cutpoints (GRACE <= 48, TIMI = 0, EDACS <= 11, HEART <= 2) missed no case of AMI, but classified less patients as low-risk.

(C) 2017

  1. Introduction

Of roughly 136 million emergency department (ED) visits to the U.S. in 2011, over 7 million were due to chest pain [1]. Despite the high inci- dence of chest pain, most patients presenting to the ED with chest pain do not have acute myocardial infarction (AMI) or unstable angina, oth- erwise known as acute coronary syndromes [2]. While the overall incidence of ACS is low, most patients undergo extensive evaluation that may include serial biomarker testing, Stress testing, coronary com- puted tomography (CT) angiography, prolonged ED observation, or

* Corresponding author.

E-mail address: [email protected] (A.J. Singer).

hospital admission, all of which results in significant resource utiliza- tion, patient inconvenience, and the potential for iatrogenic misadven- tures. In order to limit advanced testing and admission to those patients most likely to have ACS, a number of risk stratification and pre- dictive scores, which include historical, electrocardiographic and cardi- ac troponin (cTn) parameters, have been used to predict the likelihood of ACS [3].

The Thrombolysis in Myocardial Infarction risk score was de- veloped over 15 years ago to predict the risk of adverse outcomes in pa- tients with ACS within the next 14 days [4]. Similarly, the Global Registry of Acute coronary events risk score was developed based on a large observational registry of ACS patients to predict 6- month outcomes [5]. Several other predictive scores were specifically

0735-6757/(C) 2017

developed to predict poor outcomes in ED patients with chest pain in- cluding the History, ECG, Age, Risk Factors, Troponin (HEART) and the Emergency Department Assessment of Chest Pain (EDACS) risk scores [6,7]. Despite the large number of studies evaluating these and other Predictive tools, few have prospectively compared them all head-to- head in ED patients with acute chest pain. Furthermore, most recent studies have attempted to enhance the accuracy of diagnosis and risk prediction scores by including a second troponin test collected 1-3 h after the initial test [8-11]. Thus the prediction scores have evolved into decision pathways incorporating multiple components. For exam- ple, Mahler et al. have used the HEART pathway, which combines the original HEART score with troponin measures at 0 and 3 h to identify pa- tients for early discharge [12]. A potential impediment to this strategy for U.S. physicians is that troponin assays available for clinical use out- side the U.S. have markedly higher sensitivity than the FDA-approved assays available for routine American clinical practice.

As part of a multi-center study evaluating novel cardiac biomarkers, the goal of the current study was to compare the predictive value of de- cision pathways incorporating four commonly used clinical prediction scores at identifying patients with AMI when combined with serial con- temporary U.S. cTn.

  1. Methods
    1. Study design

This study was part of a prospective, multicenter, observational study aimed at validating a novel cTn for the diagnosis of AMI in ED pa- tients with symptoms suggestive of ACS. Only results of conventional clinically deployed cTn results are used in this analysis.


Conducted at seven geographically diverse academic emergency de- partments with annual censuses of 50,000-120,000, this study com- plied with the good clinical practice guidelines as per the declaration of Helsinki, and was approved by the institutional review board of each participating center.


Adult patients ages 21 years and older presenting to the ED with symptoms suggestive of ACS including sharp or dull chest pain, tight- ness, pressure, squeezing, aching, burning or stabbing, or sensations of a heavy or crushing weight on the chest, pain in the jaw or neck, pain radiating down the arms, shortness of breath or nausea were eligible for enrollment. Patients with an implantable defibrillator discharge within 24 h, a cardiac procedure (e.g., percutaneous coronary interven- tion [PCI] or Coronary artery bypass grafting [CABG]) within 30 days, or recent chest trauma were excluded.


An electronic data capture instrument was used to collect informa- tion including baseline patient demographics, physician assessments, medical history, medications, and Laboratory test results including local hospital cTn levels. Also recorded were the clinical signs and symp- toms, time of symptom onset, ECG results, and all additional tests per- formed to determine the diagnosis (such as cardiac CT angiography). Structured data collection was performed by trained Research staff using a dictionary of defined terms, measures and outcomes [13]. ECGs were classified as normal, non-specific ST-T wave changes, abnor- mal but not diagnostic of ischemia, ischemia or previous infarction known to be old, ischemia or previous infarction not known to be old, or consistent with ST-elevation AMI [13].

At each site, the attending physician was asked to estimate the like- lihood of AMI as low, moderate or high based on clinical gestalt. Addi- tionally, the total scores for each of the following prediction scores were calculated: TIMI, GRACE, HEART, and EDACS [9,14,15]. Since most recent studies and clinical pathways include measurement of seri- al cTn levels, with at least 1-3h between the first and second measure- ments, prediction scores that included cTn were calculated based on the highest of the first two cTn measurements. The HEART score was also calculated using a single cTn measurement since it was originally and continues to be described this way, which we termed HEART-1. When two cTn measurements were included in calculating the HEART score we termed this HEART-2. All scores were stratified into low, intermedi- ate, or high-risk categories (Table 1).

The EDACS score results in a score which categorizes patients into low risk and not low risk groups; it does not give points for elevated car- diac markers, but simply states that if they are elevated the patient is au- tomatically assigned to the not low risk group, regardless of score value. To generate an EDACS score which incorporated cardiac markers, 24 points were added to the score if the cTn was elevated; this would auto- matically put any case into the High risk group since the lowest possible EDACS score is -8.

Troponin assays

A variety of cTn assays were used at the various study institutions and included two point of care (POC) assays (Table 2). The local upper reference limit (99th centile) of the normal reference population was used to determine the threshold for an elevated cTn. Serial cTn assays were measured on arrival, at 3 +- 1 h and 6 +- 1 h after the initial CTn.


The main outcome was the presence or absence of an AMI. The final diagnosis was adjudicated by a central committee consisting of three members, two of which were board certified cardiologists, and the other an emergency physician. All adjudicators were experienced in clinical trial adjudication and blinded to each other’s determinations and the clinical site diagnoses. In cases where there was disagreement among the adjudicators, the final diagnosis was determined by consen- sus. The criteria used for diagnosing AMI were based on the ESC/ACCF/ AHA/WHF Third Universal Definition of Myocardial Infarction guide- lines [16].

Data analysis

Continuous data were reported as means and standard deviations or medians and inter-quartile ranges for parametric and non-parametric data respectively. Binary data were summarized as counts with fre- quencies as appropriate. Continuous data were compared with Student’ T and Mann Whitney U tests as appropriate. Binary data were compared with Chi-square or Fisher’s exact tests. The missed AMI rates in low risk patients were given with 95% confidence intervals around the point es- timate. C-statistics for each cardiac risk score were also calculated from receiver operating characteristic curves. SPSS version 23.0 (IBM Inc., Armonk, NY) was used for all statistical analyses. Because of missing data, some of the cardiac scores could not be calculated for all patients; results for each score are based on all of the cases which had sufficient data to calculate that score.

Sample size calculation

The original study was designed to enroll enough subjects to capture at least 45 patients with a confirmed diagnosis of AMI. Assuming a prev- alence of 10% for AMI in the study population, the original enrollment target was 450 patients.

Table 1

Cut-off values for grouping of prediction scores.

HEART-1 (single cTn)

HEART-2 (serial cTn)




Low risk






No new ischemia on ECG

1st and 2nd troponin both negative

intermediate risk






High risk




>= 100


  1. Results
    1. General characteristics

During the study we enrolled 459 patients across the seven Clinical sites. The median (IQR) age of study patients was 57 (49-64), 58% were male, 60% white, 29% black, and 9% Hispanic. Twenty-five patients (5%) withdrew prior to the end of the study and were excluded from analysis (Fig. 1).

Of the 434 final included patients, 80 (18.4%) were diagnosed with AMI. A breakdown of the number of patients included for calculation of the different scores is summarized in Fig. 1. A summary of baseline demographic characteristics and past medical history among patients with and without AMI are presented in Table 3. Compared to patients without AMI, those with AMI were more likely male, and with a previ- ous history of CAD, AMI, PCI, or cardiac surgery. They were also less like- ly to have a normal 12-lead ECG on presentation (Table 3). For those with two serial cTn, the median (IQR) time interval between measure- ments was 4.1 (2.7-7.8) h.

Cardiac risk scores

Unstructured emergency physician suspicion of AMI (general ge- stalt) was high in 16%, moderate in 38% and low in 33% of all study pa- tients, with 60 (14%) undocumented. The numbers (%) of patients classified as low risk based on the various prediction scores in combina- tion with cTn (Table 4) were HEART-2 138 (32%), HEART-1 140 (32%) TIMI 30 (7%), GRACE 16 (4%), EDACS 224 (52%), and unstructured clin- ical impression 113 (26%).

AMI rates for patients classified as low risk are presented in Table 4. The “missed” rates of AMI in low risk patients based on the various clin- ical prediction scores were: HEART-2, 3.6% (95% CI 1.3-8.7); HEART-1,

4.3 (95% CI 1.8-9.5); TIMI 0% (95% CI 0-14.1); GRACE 6.3% (95% CI

0.3-32.3); EDACS 0.9% (95% CI 0.2-3.5); unstructured impression with- out cTn 5.7% (95% CI 2.7-11.2); and unstructured clinical impression with serial cTn 0% (95% CI 0-3.9), none of which were significantly dif- ferent. Receiver operating characteristic curves demonstrated that EDACS performed best, with a C-statistic of 0.94 [95% CI, 0.92-0.97], followed by the HEART-2 score with a C-statistic of 0.87 [95% CI, 0.82- 0.92] (Table 5, Fig. 2). They both were significantly higher than those of TIMI, and clinical impression (GESTALT) (p b 0.05), which performed similarly to each other, and GRACE, which had the lowest C-statistic.

Table 2

Institutional cTn assays and cut-offs.

Standard-of-care cTn tests used at

99th percentile URL

10% CV

study sites



Siemens Dimension Vista TnI



Abbott i-STAT POC cTnI



Abbott Architect TnI



Roche Cobas TnT



Beckman Access AccuTnI+3






Siemens ADVIA Centaur TnI-Ultra



Alere Triage POC cTnI



By lowering the suggested risk score cutoff points we were able to reduce the risk of missed AMI to zero but with varying upper bounds of the 95% confidence limit. The revised low risk cutoff for HEART-2 was 0-2, resulting in 0% missed AMI (95% CI 0-6.6%), for HEART-1 was 0 with 0% missed (95% CI 0-6.6%), GRACE was b 49 with 0% missed (95% CI 0-5.7), and for EDACS was b 12 with 0% missed (95% CI 0-5.8). Unstructured clinical assessment, in combination with troponin, already had a zero miss rate. The percentage of patients that would have been classified as low risk using these modified cutpoints, and therefore po- tentially eligible for ED discharge were; HEART-1, 1%; GRACE 3.3%; TIMI 7%; HEART-2, 19.5%; and EDACS 33.6%.

It is also worth noting that half or more of the false negative cases for both the EDACS (1 case) and HEART-2 score with 2 cTn measurements (3 cases) involved the use of point of care assays.

  1. Discussion

Our results demonstrate that with all of the Predictive tools, using the recommended low risk cutpoints, a TIMI score of zero or a low un- structured clinical suspicion combined with two negative cTn measure- ments approximately 3-4 h later had the lowest False negative rates and did not misclassify any AMI patients. This resulted in 30% of patients classified as low risk by unstructured clinical suspicion and being eligi- ble for discharge. Surprisingly, using the recommend low risk cutpoints previously suggested by the various tools, none of the other prediction scores were sensitive enough to reduce the risk of AMI to an acceptable missed rate, generally considered to be b 1% [17].

Because the relationship between the rate of missed AMI and being categorized as low risk is relatively linear, we sought to determine

Fig. 1. Patient flow.

Table 3

Baseline characteristics of AMI vs. non-AMI.

Table 5

C-statistics (area under the curve) for the cardiac risk scores.






95% Confidence interval

N = 80

N = 354





Mean (SD) age, years

58 (11)

56 (13)



































Current history

Chest pain 98 94 0.19

Chest pain characteristics

Constant 49 49 0.92

Intermittent 51 51

Left sided 41 38 0.57

Right sided 4 6 0.49

Sub-sternal 61 47 0.02

Dyspnea 59 51 0.19

Diaphoresis 35 19 0.002

Nausea 31 27 0.43

Vomiting 9 5 0.20

Weakness 19 13 0.16

Dizziness 21 19 0.68

Prior history

CAD 58 29 b0.001

AMI 35 18 0.001

PCI 28 16 0.01

Cardiac surgery 18 9 0.02

Atrial fibrillation 13 12 0.84

HF 19 17 0.62

CP 45 35 0.10

COPD 13 11 0.67

cardiac risk factors













Hypercholesterolemia 55 47 0.19

Family history CAD 32 32 0.97

ECG 0.03

No ECG performed 0 2

Paced 0 2

Normal 31 50

Nonspecific ST or T wave changes 28 18

Abnormal but non diagnostic of ischemia 23 17

Ischemia or previous infarction known to be old 8 5

Ischemia or previous infarction NOT known to be 9 5


Consistent with AMI 3 1

a Numbers represent percentages.

where the equipoise of safety (no missed AMI) and the maximal low risk definition intersect. As applying lower risk score cutpoints to the various Prediction tools would result in smaller numbers of patients

being potentially categorized as low risk, while at the same time resulting in no missed AMIs, we determined that using the cutpoints for HEART-1 = 1, GRACE <= 48, TIMI = 0, HEART-2 <= 2, EDACS <= 11,

would result in zero missed AMI for each score. Furthermore, these lower cutpoints classified 1%, 3.2%, 7%, 18.9%, and 34.3% of patients as having “zero missed MIs,” respectively. We suggest that physicians con- sider these results when evaluating suspected ACS patients.

Both the original HEART (termed HEART-1 in our study based on a single cTn) and EDACS scores have been evaluated prospectively in dif- ferent environments and performed much better than in this investiga- tion. Originally described in a European population, the HEART score has been prospectively evaluated in only a single U.S. study [12]. EDACS was derived and validated in New Zealand [14] and showed 100% sensitivity and high classification rates, but cannot be compared with our study because they used hs cTnI [18,19]. The differences be- tween these results and our findings may be multifactorial, due to the various types of CTn assays used and also possibly to statistical variation, as our confidence intervals were wide in this smaller study. In our study, no high sensitivity assays were used, as none are approved for use in the

U.S. Thus, assay specific interpretation of various cTn assays used at any institution is critical when interpreting the various prediction scores.

Numerous ED based studies of cardiac risk scores have been pub- lished over the last decade. A 2006 study of 1481 ED patients with suspected ACS found that the incidence of 30-day death, AMI, and revas- cularization among patients with a TIMI score of 0 was 1.7% (95% CI, 0.42-2.95) [20]. A similar study of 3929 patients found that the inci- dence of adverse events at 30 days was 2.1% [21]. A more recent meta-analysis including 10 studies and 17,265 patients found an ad- verse cardiac event rate within 30 days of 1.8% among those with a TIMI score of 0 [22].

A recent review comparing cardiac risk scores found that the diag- nostic performance of the original HEART score was similar to the TIMI risk score with negative likelihood ratios in low risk patients of

0.20 (95% CI, 0.13-0.30) and 0.31 (95% CI, 0.23-0.43) respectively [3].

In this, and many other studies, a HEART score (based on a single cTn) of b 3 was considered low risk. In our analysis, a HEART score of 3 was associated with a missed AMI rate of 4.8% with a single cTn and 4.1%

Table 4

Prevalence of AMI in patients with low risk scores.a

Low risk score

No. patients

% with AMIa [1-NPV]




Clinical suspicion







(n = 374)





Clinical suspicion plus troponins







(n = 394)












(n = 374)












(n = 374)












(n = 434)












(n = 430)












(n = 411)





a 95% Confidence intervals in parentheses.

b For EDACS 24 points were added if one or both cTn was positive.

Fig. 2. receiver operating characteristics curve analyses.

with serial cTn. One of several major drawbacks of both the TIMI and the HEART scores which in our view affect their face validity is age classifi- cation: for HEART, a patient of age 46 gets the same age score as one of 64, and for TIMI, a patient of age 40 gets the same score as a patient of age 64.

A secondary analysis that included 4743 ED patients with potential ACS, found that, among the least discriminatory risk scores (TIMI, GRACE, and PURSUIT), the TIMI risk score had the best ability to predict 30-day cardiovascular events, with respective areas under the curve of 0.757 (95% CI, 0.728-0.78), 0.728 (95% CI, 0.701-0.755) and 0.691

(95% CI, 0.662-0.720) [23]. The higher C-statistics in our study are likely due to the inclusion of a second measurement of cTn in calculation of the scores, emphasizing the advantage of serial measurements.

We recognize that the thrust for the development of risk scores was historically centered on the positive predictive value for AMI, and with some scores, recurrent AMI and Cardiovascular death. While none of the cardiac risk scores completely exclude cardiovascular events in low-risk patients, combining several risk prediction scores may enhance their negative predictive value. For example, a secondary analysis of a prospective cohort of 8815 ED patients with possible ACS found no major Cardiovascular adverse events in the 485 patients with both a TIMI score of 0 and a HEART score of 0 [24]. However, by combining scores, the number classified as low risk diminishes substantially. In our study we did not attempt to combine scores due to the relatively small number of patients with confirmed AMI.

The relatively high rate of missed AMI in patients classified as low- risk in our study may be the result of incorporating older, less sensitive cTn assays at some of the study sites, especially point-of-care (POC) as- says. Among contemporary assays, there is great variation, and some of the available assays have excellent analytical characteristics, with 10% CV beneath the 99% Upper limit of normal. Others have unacceptable characteristics but are nevertheless in use [25]. Finally, some sites used both POC cTnI and central laboratory cTnI, which makes it chal- lenging to evaluate rise and fall between two different assays, as they are not standardized [26].

Studies assessing various predictive scores that included hs-cTn have demonstrated better performance of cardiac risk scores at exclud- ing AMI. A recent accelerated Diagnostic protocol defined patients with normal levels of hs-cTnT at presentation and 2 h, a TIMI score <= 1, and normal ECG findings as candidates for rapid rule-out of AMI and rapid

discharge. The protocol classified 641 patients (40.3%) as low-risk. Six of these patients had a major adverse cardiac event at 30 days, resulting in a negative predictive value of 99.1% (95% CI, 98.0-99.6%) and a sensi- tivity of 97.4% (95% CI, 94.5-98.8%) [27]. In a study of 2544 patients combining data from the ADAPT and APACE studies, and using 2 serial Abbott Architect hs TnI and a non-ischemic ECG, a TIMI score of zero classified 23% as low risk and identified all 403 patients with 30-day MACE. A TIMI score of 0/1 classified 41.2% as low risk, missing only 3 of 403 (sensitivity 99.2%) [28]. Another small study of ED patients found that, at presentation, 68 patients (76%) had a hs-cTn below cut- off value of 14 ng/L. Of these, 31 had a HEART score between 1 and 3, and no adverse cardiac events occurred in this group [29]. A prior study of 1005 ED patients found that the addition of a serial 3 h cTn to the HEART score (which we would have termed HEART-2) resulted in the identification of 1 patient with ACS that would have been missed using the HEART-1 approach (a score with a single cTn measurement) [30]. Thus reliance on cardiac risk scores and a single cTn may not be ad- equate to exclude AMI or ACS and repeat testing (as early as 1-3 h after ED arrival) should be strongly considered at this time in order to have a miss rate of 1% or less, especially when using non high sensitivity assays. In our study clinical gestalt combined with two cTn performed as well as most prediction scores in identifying patients at low risk of AMI. However, inter-rater agreement for clinical gestalt may be less than for the cardiac risk scores. A prior study of 255 patients, 75 of whom had ACS, found that the area under the curve of the HEART score was not significantly different from clinical gestalt, (0.81 (95% CI, 0.76-0.86) vs. 0.79 [95% CI, 0.73-0.84], p = 0.13) [31]. In contrast,

Body et al. have also demonstrated that by combining clinical gestalt with the admission ECG and hs-cTnT level below the 99% cutoff (14 ng/mL), 100% sensitivity (95% CI 95.6% to 100%) for the diagnosis of AMI was achieved [32]. Although accurate in the current study as well as the ones referenced, there may be greater variability with the clinical gestalt, which may be decreased by using formalized criteria in addition to clinical gestalt.


Our study has several notable limitations. The analysis is based on a relatively small sample size with only 82 confirmed cases of AMI. Thus the precision around the point estimates is wide. Secondly, we excluded many patients due to incomplete data, thus introducing potential selec- tion bias that may have led to an underestimation or overestimation of the accuracies of the scoring systems. Consent and participation in the study also introduced selection bias as our population may not be repre- sentative of the general population of patients being evaluated for chest pain at community centers. Furthermore, our only outcome was AMI di- agnosed on the index visit. Thus we probably missed other cases of ACS and longer-term major cardiac events. However, the inability of a pre- diction score to identify AMI during the index visit would severely limit or preclude its use. Additionally, because we used each hospital’s troponin assay as the biomarker definition, we introduced a variability in ascertainment of the outcome due to the troponin itself, which may account for the decrement in risk score performance, as the local tropo- nins used in the United States have various “contemporary” sensitivities in terms of the detection of troponin at low concentrations. It is also worth noting that half the false negative cases for both the EDACS (1 case) and HEART-2 score with 2 cTn measurements (3 cases) involved the use of point of care assays, which are additionally limited due to rel- atively high coefficients of variation. As a result, such assays have inferi- or analytical reliability compared with central laboratory based assays. This may also have adversely affected the performance of the risk tools. Nevertheless, our results confirm that ED patients classified as low-risk of adverse cardiac events based on any of the currently avail- able predictive scores still have an unacceptable high rate of AMI pre- cluding early discharge. Thus additional testing, including more prolonged serial cTn (especially when using non high sensitivity

assays), stress testing and advanced imaging (cardiac CT angiography, MRI, echocardiography) may need to be considered prior to safe ED discharge.

  1. Conclusions

Our prospective, multi-center observational study, using non-high- sensitivity cTn assays currently available in the U.S., compared the accu- racy and predictive values of general clinical gestalt and validated cardi- ac risk scores in a population of ED patients with suspected AMI. Using their recommended cutpoints and non high sensitivity cTn, TIMI and unstructured clinical impression were the only scores with no missed cases of AMI. Using lower cutpoints (GRACE <= 48, TIMI = 0, EDACS <= 11, HEART <= 2) missed no case of AMI, but classified less pa- tients as low-risk. Our data suggest that predictive scores and pathways should be used cautiously in terms of negative predictive value with clinical judgment being essential to determine the safety of discharging ED patients with suspected AMI and ACS.


Presented at the Society for Academic Emergency Medicine, May 2016, New Orlean, LA.


This study was funded by Alere, San Diego, CA.

Conflicts of interest

AJS has received research funding from Alere and APOC and is a con- sultant for Alere and APOC.

Author contributions

AJS and WFP conceived the study and designed the study. AJS and WFP supervised study conduct and data collection. AWD undertook site recruitment and supervision. HCT provided statistical advises and analyzed the data. AJS drafted the manuscript and all authors contribut- ed substantially to its revision. AJS takes responsibility for the paper as a whole.


  1. National Hospital Ambulatory Medical Care Survey: 2011 Emergency Department Summary Tables.
  2. Penumesta SC, Mallidi J, Friderici JL, Hiser W, Rothberg MB. Outcomes of patients ad- mitted for observation of chest pain. Arch Intern Med 2012;172:873-7.
  3. Fanaroff AC, Rymer JA, Goldstein SA, Simel DL, Newby LK. Does this patient with chest pain have acute coronary syndrome?: the rational clinical examination sys- tematic review. J Am Med Assoc 2015;314:1955-65.
  4. Antman EM, Cohen M, Bernink PJ, McCabe CH, Horacek T, Papuchis G, et al. The TIMI risk score for unstable angina/non-ST elevation MI: a method for prognostication and therapeutic decision making. J Am Med Assoc 2000;284:835-42.
  5. Elbarouni B, Goodman SG, Yan RT, Welsh RC, Kornder JM, Deyoung JP, et al, Canadian Global Registry of acute coronary events (GRACE/GRACE(2)) Investigators. Validation of the Global Registry of Acute Coronary Event (GRACE) risk score for in-hospital mortality in patients with acute coronary syndrome in Can- ada. Am Heart J 2009;158:392-9.
  6. Six AJ, Be B, Kelder JC. Chest pain in the emergency room: value of the HEART. Neth Heart J 2008;16:191-6.
  7. Sanders S, Flaws D, Than M, Pickering JW, Doust J, Glasziou P. Simplification of a scoring system maintained overall accuracy but decreased the proportion classified as low risk. J Clin Epidemiol 2015 May 14.

05.006 [pii: S0895-4356(15)00227-9, Epub ahead of print].

  1. Six AJ, Cullen L, Backus BE, Greenslade J, Parsonage W, et al. The HEART score for the assessment of patients with chest pain in the emergency department: a multina- tional validation study. Crit Pathw Cardiol 2013;12:121-6.
  2. Cullen L, Greenslade J, Hammett CJ, Brown AF, Chew DP, Bilesky J, et al. Comparison of three risk stratification rules for predicting patients with acute coronary syn- drome presenting to an Australian emergency department. Heart Lung Circ 2013; 22:844-51.
  3. Aldous SJ, Richards M, Cullen L, Troughton R, Than M. A 2-hour thrombolysis in myocardial infarction score outperforms other risk stratification tools in patients presenting with possible acute coronary syndromes: comparison of chest pain risk stratification tools. Am Heart J 2012;164:516-23.
  4. Meller B, Cullen L, Parsonage WA, Greenslade JH, Aldous S, et al. Accelerated diag- nostic protocol using high-sensitivity cardiac troponin T in acute chest pain patients. Int J Cardiol 2015;184:208-15.
  5. Mahler SA, Riley RF, Hiestand BC, Russell GB, Hoekstra JW, et al. The HEART pathway randomized trial. Identifying emergency department patients with acute chest pain for early discharge. Circ Cardiovasc Qual Outcomes 2015;8:195-203.
  6. Cullen L, Than M, Brown AFT, et al. Comprehensive standardized data definitions for acute coronary syndrome research in emergency departments in Australasia. Emerg Med Australas 2010;22:35-55.
  7. Than M, Flaws D, Sanders S, Doust J, Glasziou P, et al. Development and validation of the emergency department assessment of chest pain score and 2 h accelerated diag- nostic protocol. Emerg Med Australas 2014;26:34-44.
  8. Six AJ, Cullen L, Backus BE, Greenslade J, Parsonage W, et al. The HEART score for the assessment of patients with chest pain in the emergency department. Crit Pathw Cardiol 2013;12:121-6.
  9. Thygesen K, Alpert JS, Jaffe AS, Simoons ML, Chaitman BR, White HD, et al. Third uni- versal definition of myocardial infarction. Glob Health 2012;7:275-95.
  10. Than M, Herbert M, Flaws D, Cullen L, Hess E, et al. What is an acceptable risk of

major adverse cardiac event in chest pain patients soon after discharge from the emergency department?: a clinical survey. Int J Cardiol 2013;166:752-4.

  1. Pickering JW, Greenslade JH, Cullen L, Flaws D, Parsonage W, et al. Validation of pre- sentation and 3 h High-sensitivity troponin To rule-in and rule-out acute myocardial infarction. Heart 2016;0:1-9.
  2. Than MP, Pickering JW, Aldous SJ, Cullen L, Frampton CMA, et al. Effectiveness of EDACS versus ADAPT accelerated Diagnostic pathways for chest pain: a pragmatic randomized controlled trial embedded within practice. Ann Emerg Med 2016 [in press].
  3. Chase M, Robey JL, Zogby KE, Sease KL, Shofer FS, Hollander JE. Prospective valida- tion of the Thrombolysis in Myocardial Infarction risk score in the emergency depart- ment chest pain population. Ann Emerg Med 2006;48:252-9.
  4. Pollack Jr CV, Sites FD, Shofer FS, Sease KL, Hollander JE. Application of the TIMI risk score for unstable angina and non-ST elevation acute coronary syndrome to an un- selected emergency department chest pain population. Acad Emerg Med 2006;13: 13-8.
  5. Hess EP, Agarwal D, Chandra S, Murad MH, Erwin PJ, Hollander JE, et al. Diagnostic accuracy of the TIMI risk score in patients with chest pain in the emergency depart- ment: a meta-analysis. Can Med Assoc J 2010;182:1039-44.
  6. Lee B, Chang AM, Matsuura AC, Marcoon S, Hollander JE. Comparison of cardiac risk scores in ED patients with potential acute coronary syndrome. Crit Pathw Cardiol 2011;10:64-8.
  7. Marcoon S, Chang AM, Lee B, Salhi R, Hollander JE. HEART score to further risk strat- ify patients with low TIMI scores. Crit Pathw Cardiol 2013;12:1-5.
  8. Apple FS. A new season for cardiac troponin assays: it’s time to keep a scorecard. Clin

Chem 2009;55:1303-6.

  1. Singh J, Akbar MS, Adabag S. Discordance of cardiac troponin I assays on the point- of-care i-STAT and architect assays from Abbott diagnostics. Clin Chim Acta 2009; 403:259-60.
  2. Cullen L, Mueller C, Parsonage WA, Wildi K, Greenslade JH, et al. Validation of high- sensitivity troponin I in a 2-hour diagnostic strategy to assess 30-day outcomes in emergency department patients with possible acute coronary syndrome. J Am Coll Cardiol 2013;62:1242-9.
  3. Meller B, Cullen L, Parsonage WA, Greenslade JH, Aldous S, Reichlin T, et al. Acceler- ated diagnostic protocol using high-sensitivity cardiac troponin T in acute chest pain patients. Int J Cardiol 2015 Apr 1;184:208-15.
  4. Willems MN, van de Wijngaart DJ, Bergman H, Adiyaman A, Telting D, Willems FF. Addition of heart score to high-sensitivity troponin T versus conventional troponin T in risk stratification of patients with chest pain at the coronary emergency rooms. Neth Heart J 2014;22:552-6.
  5. Mahler SA, Miller CD, Hollander JE, Nagurney JT, Birkhahn R, Singer AJ, et al. Identi- fying patients for early discharge: performance of decision rules among patients with acute chest pain. Int J Cardiol 2013;168(2):795-802.
  6. Visser A, Wolthuis A, Breedveld R, ter Avest E. HEART score and clinical gestalt have similar diagnostic accuracy for diagnosing ACS in an Unselected population of pa- tients with chest pain presenting in the ED. Emerg Med J 2015;32:595-600.
  7. Body R, Cook G, Burrows G, Carley S, Lewis PS. Can emergency physicians ‘rule in’ and ‘rule out’ acute myocardial infarction with clinical judgement? Emerg Med J 2014;31:872-6.

Leave a Reply

Your email address will not be published. Required fields are marked *