Article Text

PDF

What is the incidence of major adverse cardiac events in emergency department chest pain patients with a normal ECG, Thrombolysis in Myocardial Infarction score of zero and initial troponin ≤99th centile: an observational study?
  1. Anne-Maree Kelly
  1. Correspondence to Professor A-M Kelly, Joseph Epstein Centre for Emergency Medicine Research at Western Health, Sunshine Hospital, Furlong Road, St Albans, VIC 3021, Australia; anne-maree.kelly{at}wh.org.au

Abstract

Objective To determine the rate of major adverse cardiac events (MACE) in patients assessed in an emergency department (ED) for chest pain with a non-ischaemic ECG, Thrombolysis in Myocardial Infarction (TIMI) score of 0 and initial troponin I (TnI) ≤99th centile.

Methods This was a sub-study of a prospective observational study of adult patients with potentially cardiac chest pain who underwent evaluation for acute coronary syndrome in an urban teaching hospital. Adult patients with non-traumatic chest pain were eligible for inclusion. Those with ECG evidence of acute ischaemia or an alternative diagnosis were excluded. Data collected included demographic, clinical, ECG, biomarker and outcome data. Low risk was defined as a TIMI risk score of 0 and initial TnI ≤99th centile. Primary outcome of interest was defined as MACE within 7 days. MACE included death, cardiac arrest, revascularisation, cardiogenic shock, arrhythmia, and prevalent (cause of presentation) and incident (occurring within the follow-up period) myocardial infarction (MI). Analysis was by descriptive and clinical performance analyses.

Results 651 patients were studied of whom 215 met the low risk criteria. There was one MACE in this group (0.47%, 95% CI 0.08% to 2.6%)—a revascularisation within 7 days without prevalent MI. Negative predictive value of low risk classification was 99.5% (95% CI 97% to 100%) at both 7 and 30 days. Negative likelihood ratio, weighted by prevalence, was 0.005 at both intervals.

Conclusion Risk stratification for early discharge based on ECG, TIMI score of 0 and presentation TnI ≤99th centile appears to identify a group at very low risk of MACE. Research to prospectively validate this is warranted.

Statistics from Altmetric.com

Introduction

Evaluation of chest pain is a growing challenge for emergency departments (ED), in particular ruling out acute coronary syndrome (ACS) and determining which patients can safely be discharged for further management in the community with a low risk of adverse cardiac events. Recently, Than and colleagues1 reported the use of an accelerated diagnostic protocol which included use of a structured pretest probability (using the Thrombolysis in Myocardial Infarction (TIMI) score), ECG and a point of care biomarker panel. Patients with a TIMI score of 0, no new ischaemic changes on ECG and negative biomarker assays at presentation and at 2 h where defined as low risk; 9.8% of patients met the low risk criteria, of whom three had a major adverse cardiac event (MACE) within 30 days (0.08%). They reported a sensitivity of the protocol of 99.3% (95% CI 97.9% to 99.8%) with a negative predictive value (NPV) of 99.1% (95% CI 97.3% to 99.8%). They concluded that this protocol identified a group at very low risk who might be safe for early discharge and that this approach has the potential to reduce observation periods for chest pain assessment, enhancing ED flow.

The Antithrombotics in the Secondary Prevention of Events in Coronary Thrombosis (ASPECT) study1 used a point of care multimarker panel including creatine kinase, troponin and myoglobin. Troponin is the biomarker considered to be most sensitive and specific for myocardial injury. Newer so-called highly sensitive troponin assays have much increased sensitivity compared with their predecessors. Some reports suggest that they are highly sensitive for the diagnosis of myocardial infarction at ED presentation (∼90%)2 ,3 while others report lower sensitivities.4 Promisingly, Body et al 5 reported that using a single high sensitivity troponin T (TnT) assay, an initial TnT value below the limit of detection had a NPV for myocardial infarction (MI) of 100% (95% CI 98.1% to 100%) in a research cohort and 99.4% (95% CI 96.6% to 100%) in a clinical audit cohort. Unfortunately, our previously published data found troponin I (TnI) ≤99th centile had unacceptable sensitivity of 76.7% (95% CI 68.5% to 83.7%) with NPV 96.3% (95% CI 94.6% to 97.4%).4

The TIMI score has also been shown to correlate with outcome in ED patients.6 ,7 A TIMI score of 0 has been found in meta-analysis to have a 1.8% rate of adverse cardiac events at 30 days.8 In a recent study, Goodacre et al 9 reported that patients with a TIMI score of 0 had a rate of MACE of 0.6% at 30 days and 0.9% at 90 days. The high sensitivity of new troponin assays and the high negative predictive value of a TIMI score of 0 raises the question of whether their combination might prove effective in identifying a group of ED chest pain patients at very low risk of adverse events using an accelerated assessment strategy.

The aim of the study was to determine the rate of MACE at 7 and 30 days in patients assessed in an ED for chest pain with a non-ischaemic ECG, a TIMI score of 0 and an initial TnI ≤99th centile.

Methods

Design and setting

This was an unplanned post hoc sub-study of a prospective observational study of consecutive adult patients with potentially cardiac chest pain who underwent evaluation for ACS. That study focused on clinical features, risk scores, performance of biomarker assays and outcomes. The study was conducted in the ED of a community teaching hospital with an annual ED census of 32 000 patients. The study was approved according to institutional ethics approval requirements. Consent was not required for collection of data from routine clinical care.

Population and data

Adult patients (aged >18 years) presenting with non-traumatic chest pain where ACS was considered by the clinician as a possible diagnosis were eligible for inclusion. Patients who did not undergo ECG or cardiac biomarker testing, those with clear ECG evidence of acute ischaemia (as adjudicated by the treating clinician), those who declined or were unavailable for follow-up and those with a clear alternative diagnosis were excluded. Data collected included demographic, clinical, ECG, biomarker and outcome data. While patients were identified prospectively and most data (ECG, biomarker and outcome data) were collected prospectively, some demographic and clinical data were collected by explicit medical record review using a piloted proforma. Risk scores were calculated retrospectively by a research officer blinded to outcome. Only data available in ED were used to calculate scores. Follow-up was by structured records review and telephone interview by a trained research nurse after 7 and 30 days. The low risk group was defined as non-ischaemic ECG, TIMI 0 and presentation TnI ≤99th centile. Clinical care was according to established institutional protocols with clinical assessment, serial ECG and serial biomarker assays at presentation and at least 4 h thereafter (with at least one test 6 h from symptom onset). Provocative testing or CT coronary angiography as outpatients was arranged for patients who were ruled out for ACS, as appropriate.

The troponin assay used by the laboratory was TnI-Ultra (Siemens Diagnostics Erlanger, Germany) performed on an Advia Centaur analyser in Erlanger, Germany. The test has a reported range of 0.006–50 ng/ml. Coefficient of variation is 10% at TnI 0.03 ng/ml, 5.3% at 0.08 ng/ml and 4.1% at 0.18 ng.ml. The 99th percentile is 0.04 ng/ml (95% CI 0.03 ng/ml to 0.05 ng/ml).10 The assay manufacturer was not involved in any aspect of this study.

Outcomes and analysis

The primary outcome of interest was defined as MACE within 7 days. MACE included death (unless clearly non-cardiac), cardiac arrest, revascularisation, cardiogenic shock, arrhythmia requiring intervention, and prevalent (being the cause for the patient's presentation) and incident (occurring within the follow-up period) MI. Secondary outcomes were MACE at 30 days. MI diagnosis was as adjudicated for admitted patients by the treating cardiologist with access to all clinical and investigational data. Treating cardiologists were unaware of the study.

Data were analysed by descriptive statistics, χ2 and Fisher's tests for comparisons of proportions, Mann–Whitney U test for non-parametric continuous comparisons and clinical performance analysis; sensitivity, specificity, NPV and negative likelihood ratio (LR).11 No a priori sample size calculation was performed.

Results

Sample derivation is shown in figure 1 and characteristics of the cohort in table 1. A total of 550 patients had initial TnI ≤99th centile and 215 patients had a TIMI risk score of 0. All TIMI 0 patients had initial TnI ≤99th centile.

Figure 1

Sample derivation. MI, myocardial infarction; TIMI, Thrombolysis in Myocardial Infarction; TnI, troponin I.

Table 1

Characteristics of the sample

Eighty-seven patients suffered a MI (13.6%, 95% CI 10.9% to 16.3%). There were 90 MACE at 7 days (13.8%, 95% CI 11.4% to 16.7%) and 95 MACE at 30 days (14.6%, 95% CI 12.1% to 17.6%) (figure 1). Two hundred and fifteen patients were classified as low risk according to our criteria (33.0%, 95% CI 29.5% to 36.7%). There was one MACE (a revascularisation within 7 days without prevalent MI) in the defined low risk group (0.47%, 95% CI 0.08% to 2.6%).

At 7 days, the risk classification system used had a sensitivity 98.9% (95% CI 93.1% to 99.9%), specificity 38.1% (95% CI 34.1% to 42.3%), NPV 99.5% (95% CI 97% to 100%) and negative LR, weighted by prevalence, of 0.005 (95% CI 0.0006 to 0.033).

At 30 days, the risk classification system used had a sensitivity of 98.9% (95% CI 93.4% to 99.9%), specificity 38.5% (95% CI 34.4% to 42.7%), NPV 99.5% (95% CI 97% to 100%) and negative LR, weighted by prevalence, of 0.005 (95% CI 0.0007 to 0.033).

Discussion

These results suggest that in patients without ECG evidence of ischaemia, the combination of TIMI score 0 and initial TnI ≤99th centile identifies a group at very low risk of MACE who may be suitable for an accelerated ED assessment process. As current Australasian diagnostic protocols using high sensitivity assays recommend serial TnI assays at least 3 h apart (with one at least 6 h from symptom onset),12 such an approach might allow discharge several hours earlier than is now possible with reduced inconvenience to patients and improved ED flow. It should be noted however that as this was a post hoc analysis, it is hypothesis generating rather than hypothesis testing. Prospective validation would be required before this approach to risk stratification and clinical care could be recommended.

Comparing the ASPECT accelerated diagnostic protocol low risk group and the low risk group defined in this study, the difference in MACE at 30 days was 0.38%, favouring the TIMI 0–initial TnI stratification approach (95% CI −1.8% to 2.1%). Similarly, the difference in NPV was 0.38% favouring the TIMI 0–initial TnI stratification approach (95% CI −1.8% to 2.1%). It should be noted however that this is not a scientifically robust comparison as the cohorts studied were different and unidentified heterogeneities cannot be excluded. Considered together, our data and those from the ASPECT study1 strongly suggest that the combination of ECG, TIMI score and biomarker assays over a short time period can identify a group that are safe for early discharge. Body et al 5 took a different approach and found that a single ‘undetectable’ level of TnT using a high sensitivity test at ED presentation had a very high NPV for detection of MI (>99%). Prospective validation of these approaches is required and should ideally be combined so that they can be validly compared.

In this sample, 33% of patients met low risk criteria. This is believed to be an overestimate due to a relatively high number of exclusions for language and telephone contact issues and decline of consent to follow-up. The true number is likely to be closer to the 17.5% reported in the Manchester trial5 which had higher follow-up performance. A shorter diagnostic process for this proportion of patients of the order of 4 h would be of significant clinical and operational value given the high prevalence of this condition. This needs to be weighed against the cost of potentially missed MACE.

This study had a similar proportion of MACE at 30 days to the ASPECT study (14.6% vs 11.8%), suggesting that the cohorts are reasonably similar despite ASPECT being conducted at 14 sites in nine countries, principally from Asia, versus this single site study. There was a minor difference in the definition of MACE between the studies with ASPECT including emergent revascularisation and this study including any revascularisation which may account for the small difference. The threshold for investigation for ACS might be different in different countries or different hospitals, fuelled in part by differing medicolegal environments. The rate of MI is lower than reported by Body5 (18.5%) but that was a research population rather than a clinical cohort. All three studies might be challenged (for different reasons) on their generalisability to broader international rule out ACS practice, particularly in the North American and European context. Notwithstanding, they report identification of a very low risk group worthy of further validation.

The risk stratification approach used had a high NPV (99.5%) that was similar to that reported by the ASPECT study (99.1%)1 and the Manchester clinical cohort (99.4%).5 As safety for early discharge is the aim, these data suggest the approaches have similar high safety. This study also had a similar sensitivity to that reported by the other studies (98.9% vs 99.3% and 99.8%) and a relatively low specificity (of the order of 34%). This is not unexpected as the aim of the approach is safe rule out rather than rule in of ACS and risk of MACE. This finding is similar to other instruments to rule out ACS.1 ,13 ,14 Unlike the ASPECT trial which had a median initial hospital length of stay of 26 h (IQR 9.9–37.0 h), 90% of the low risk cohort in this study were discharged within 24 h. This likely reflects differences in diagnostic processes between institutions, but may also reflect some patient inclusion differences between the studies.

In this study, serial biomarker testing intervals according to Australasian guidelines for high sensitivity troponin assays were used.12 This differs from recommended biomarker intervals in other countries or guidelines. The National Institute for Clinical Evidence guideline recommends serial biomarker assays10–12 h apart.15 The European Cardiac Society recommends an interval of 6–12 h16 and American guidelines recommend assays at presentation and 8–12 h after symptom onset.17 None of these approaches has been shown to be superior to others. The differences are probably explained by rapidly evolving and sometimes conflicting evidence, the range of biomarker assays available which may vary by jurisdiction and the development of new and more sensitive assays.

The argument could be made that the elements examined here are interconnected rather than discrete variables, in that the TIMI risk score includes specified ECG abnormalities and ‘abnormal’ biomarkers as criteria in the score. The approach taken in this analysis was deliberately designed to align with clinical process in ED. A diagnostic ECG at presentation rules in ACS, making the remainder of the risk score irrelevant and the troponin level of less diagnostic import. If the ECG is not diagnostic, a process of biomarker assays is usual practice to identify MI. Depending on the setting, processes for observation and testing may be guided by a risk classification (TIMI or other). This was the case at the study institution although the risk classification system used was that described in national guidelines.11

The TIMI score has also been shown to correlate with outcome in ED patients.6 ,7 A TIMI score of 0 has been found in meta-analysis to have a 1.8% rate of adverse cardiac events at 30 days.8 The studies involved in this meta-analysis used a variety of troponin assays, some with less sensitivity than current tests. A recent study9 reported MACE of 0.6% at 30 days in a UK cohort. As tests have become more sensitive, fewer patients are TIMI 0 so the meta-analysis might be overestimating adverse events in TIMI 0 patients who are assessed using newer assays. Alternatively, the approach described in this study may not perform as well as reported here in a more heterogeneous population. Further research will be required to clarify this.

There are some limitations to this study which should be considered when interpreting its results. While patients were identified prospectively, some data were collected from the medical record with the inherent weaknesses associated. It was conducted at a single site so may not be generalisable to other settings. Determination of risk factors and past history was by patient self-report. No attempt was made to confirm the information provided, reflecting the ‘real world’ ED setting. It is also possible that there were some cases of missed diagnosis of MI, principally because of atypical presentations, in particular lack of chest pain. The shorter interval of biomarker testing required by the Australian guidelines might also have contributed to missed MI. The TIMI score was not a routine part of clinical practice at the study institution but individual cardiologists may have calculated it in particular cases and this may have influenced their management decision making.

Conclusion

Risk stratification for early discharge based on ECG, TIMI score of 0 and presentation TnI ≤99th centile appears to identify a group at very low risk of MACE. Research to prospectively validate is warranted.

Acknowledgments

The author notes with thanks the efforts in data collection of Sharon Klim and the data management services of People Strategy Innovation Pty Ltd.

View Abstract

Footnotes

  • Funding This project was supported by the Morson Taylor Award of the Emergency Medicine Research Foundation, supplemented by departmental funds.

  • Competing interests A-MK is the coauthor of the guidelines referred to in the discussion.

  • Ethics approval Ethics approval was provided by the Western Health HREP.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The author is open to requests to share data.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

References