Single high-sensitivity troponin-I for ruling out acute coronary syndrome: a detection limit approach

Abstract

Aims

The aim of this study was determine the incidence of major adverse cardiac events within 30 and 365-days among patients discharged from emergency departments (EDs), following a single high-sensitivity cardiac troponin I test result below or close to the limits of detection (LoD).

Methods and results

Patients ≥20 years old who presented to four EDs from mid-2014 to end-2015, underwent a single high-sensitivity troponin test and were discharged were included. Data from ED presentations, hospital admissions, mortality records, and pathology laboratories were linked and harmonized. High-sensitivity troponin levels were categorized as below (<2 ng/L) or close to (<5 ng/L) the LoD. The primary outcome was cardiovascular death and myocardial infarction (MI), identified using ICD-10-AM codes. In a cohort of 6633 patients, 49% had high-sensitivity troponin levels below the LoD (<2 ng/L), and 79% had levels <5 ng/L. There were no primary outcome events at 30-day follow-up among patients with high-sensitivity troponin results below 2 or 5 ng/L. At 365-days, there were 5 (0.15%) and 11 (0.21%) primary outcome events for patients with high-sensitivity troponin results below 2 and 5 ng/L, indicating negative predictive values of 99.85% and 99.79%.

Conclusion

These findings confirm that patients with a single very low level of high-sensitivity troponin on presentation to EDs are at low risk of MI and cardiovascular death at 30 and 365 days, supporting the safety of a triage strategy incorporating a single high-sensitivity troponin result below the LoD to identify patients at low-risk, who may be suitable for expedited discharge.

Graphical Abstract

Open in new tabDownload slide

Introduction

Symptoms that incorporate a differential diagnosis of acute coronary syndrome (ACS) are a leading cause of presentation to Emergency Departments (ED). However, the incidence of acute myocardial infarction (MI) in this population is low.¹ Current European guidelines for management of ACS recommend using serial (at presentation and 1 or 2 h later) high-sensitivity cardiac troponin (hs-cTn) assays to rule-in or rule-out MI, with individuals thus identified as low risk and considered suitable for expedited discharge.² They do, however, acknowledge that a single very low levels of hs-cTn at presentation may be adequate to rule out MI, particularly if measured >3 h after symptom onset.

Several studies have concluded that patients with exceptionally low levels of hs-cTn assay at presentation are at very low risk of MI or death at 30 days.^3,4 However, whilst these studies suggest excellent outcomes for patients with very low hs-cTn levels at presentation, many were not testing the real-world application of discharging patients based only on a single reassuring hs-cTn result. Instead, many of the patients with low hsTn at presentation in these initial studies underwent serial testing with longer observation, possible hospital admission, and further investigations to inform decisions. Other more recent studies have addressed this, and further support safe discharge of low-risk patients with a single hs-cTn at the limits of detection (LoD).^5–7 There are, however, limited real-world data on the outcomes of patients who have been directly discharged home after a single very low level hs-cTn measurement, with little information on events extending beyond 30-day follow up.

The aim of this study was to assess 30-day and 365-day outcomes among patients presenting to EDs, investigated with only a single hs-cTnI test result below or close to the LoD, and who were directly discharged.

Methods

Study design and setting

The study cohort was derived from patients who visited one of four participating EDs across Australia between mid-2014 and end-2015. All participating hospitals used the Abbott ARCHITECT hs-cTnI assay, and the study included consecutive patients (≥20 years old), who underwent a single hs-cTnI test at the discretion of the attending medical team and were discharged directly from the ED. The APACE and ADAPT cohorts informed localized ED protocols for high-sensitivity troponin testing in those with any symptoms, which suggest possible ACS.⁸ We obtained linked administrative data from each jurisdiction and troponin assay details from the respective accredited pathology laboratories. The cross-jurisdictional datasets were harmonized and combined into a single file. The data were coded using the International Classification of Diseases Tenth Revision Australian Modification (ICD-10-AM). Initial hs-cTnI concentrations below 2 ng/L were considered undetectable, following the manufacturers guidelines for the Abbott ARCHITECT assay. The assay has an LoD range of 1.1–1.9 ng/L, a 99th percentile upper reference limit of 26.2 ng/L, and a coefficient of variation of <10% at 4.7 ng/L.⁹

As previous studies and recent rule-out pathways have used a higher threshold of hs-TnI <5 ng/L,^5,6 participants were also categorized on this threshold. This approach results in overlapping groups, with individuals with hs-TnI levels <2 ng/L also included in the cohort with levels below 5 ng/L.

Outcomes

Outcomes were identified using ICD-10-AM codes in the principal discharge diagnosis field or mortality field where death occurred. The primary outcome was the composite of cardiovascular death (ICD10-AM: I00-99) and MI (ICD-10-AM: I21). Secondary outcomes included all-cause death and stroke (ICD-10-AM: I60-61, I63-64). Coding of administrative data in Australia has been previously validated, demonstrating a positive predictive value of over 90% for cardiovascular death, 83% for MI, and 85% for stroke.¹⁰

Data analyses

All analyses were completed in SAS (version 9.4 M7). Categorical variables were presented as counts and proportions. Due to the sample size and the low incidence of outcomes, statistical comparisons, and meaningful modelling or hypothesis testing were limited. The study was approved by the South Metropolitan Health Service Human Research Ethics Committee (RGS: 000000490) and conducted in accordance with the Declaration of Helsinki.

Results

There were 37 501 presentations to the four participating EDs during the study period, with 3.9% diagnosed with MI on ED departure. The study cohort, a subset of these individuals, comprises 6633 patients (47% female, mean age of 51.7 years, SD 18.2: Table 1) who underwent a single troponin test only and were discharged directly from the ED.

Approximately, half of the study cohort (n = 3283) had hs-cTnI test results below the LoD (<2 ng/L), whilst 79% (n = 5263) of the cohort had levels <5 ng/L (Table 2). At 30-day follow-up, no cardiovascular deaths or MIs were observed. At 365 days after discharge from ED there were no cardiovascular deaths and five cases (0.15%) of MI for patients with a single hs-cTnI level <2 ng/L, yielding a negative predictive value (NPV) of 99.85% (95% CI: 99.72–99.98). These patients ranged in age from 47–65 years (mean age 54 years) and three were male. At the higher threshold of hs-cTnI <5 ng/L, an additional five patients with MIs were identified (10 in total; 0.19%) and there was one cardiovascular death. The NPV for the primary outcome was 99.79% (95% CI: 99.69–99.89). These patients ranged in age from 37 to 83 years (mean age 58 years) and included six males.

At 1-year, there were 9 cases of stroke (0.17%) and 25 deaths (0.48%) among the 5263 individuals with hs-cTnI <5 ng/L (Table 2). Of those that died, cancer was the predominant cause.

To capture all presentations evaluated for possible ACS, our sample included consecutive unselected patients presenting to the ED where this was a potential differential diagnosis, rather than only those with suggestive symptoms, such as chest pain. To address this, we examined a subset of patients from a single hospital who had a presenting complaint incorporating symptoms suggestive of a differential diagnosis of ACS (n = 1004). The results in this subgroup were consistent with the primary analysis (Table 2).

A further sensitivity analysis was conducted including all patients presenting to the four EDs during the study period who had presenting hs-cTnI <2 ng/L, independent of hospitalisation status (n = 7193). The incidence of combined cardiovascular death or MI at 30-days and 365-day were 4 (0.06%) and 16 (0.22%), respectively.

Discussion

Scarce real-world data are available addressing the outcomes of patients discharged home from ED after only a single hs-cTn measurement. This multi-centre study addresses this and confirms very low rates of cardiovascular death and MI even after 1 year.

Imprecision in hs-cTnI assays at the low thresholds used for single-sample rule-out pathways could influence their performance, with risk of misdiagnosis, yet there is little guidance on acceptable imprecision levels. Li et al. recently addressed this issue using the HighSTEACS cohort and demonstrated that whilst assay imprecision could lead to some patient reclassification, the likelihood of missing an MI diagnosis for this reason was very low (0.01%) at a rule-out threshold of <5 ng/L, with an assay coefficient of variation of 10% at this level.¹¹ Our results extend and support the findings of recent studies that have prospectively tested a single-sample rule-out strategy.^5–7 They do not, however, imply that expedited discharge should be based on biomarker results alone. Any such decisions need to be made in conjunction with careful clinical risk assessment, based on history, examination and other investigations.

Limitations

Whilst independent adjudication of outcome events was not conducted in this study coding of administrative data in Australia has been previously validated.¹⁰ Additionally, the time from onset of symptoms to troponin testing, nature and duration of symptoms were not collected or included in the analysis.

Whilst this study validated the practice of discharging some patients from ED, the study population is likely biased towards those at lowest risk and rigorous clinical risk stratification should continue. Likewise, the low risk of the study cohort means that, despite the large sample size, the limited events may reduce precision in estimating missed cases of MI or cardiovascular death. Likewise, due to the small number of adverse outcomes, a separate analysis by sex was not performed.

Although the data were collected in 2014 and 2015, the hs-cTnI assay, central to our analysis, remains among the commonest used to diagnose MI in contemporary clinical practice. Additionally, there have been no substantial changes to guidelines for assessment of patients with possible ACS. Thus, these findings hold relevance for clinicians in their current decision-making, but may not be applicable to other troponin assays, such as hs-cTnT.

Conclusions

An undetectable or extremely low circulating level of hs-cTnI at presentation may provide an opportunity for treating clinicians to safely discharge patients directly from the ED after a single measurement. This can considerably reduce the time spent in hospital and eliminate the need for many patients to wait for serial troponin results to rule out MI.

Lead author biography

Siobhan Hickling is a research dietitian and epidemiologist in public health teaching, research and practice. Her research spans large national and multi-national studies, such as TropED and VITATOPS, as well as smaller qualitative and quantitative research in cardiovascular and nutritional epidemiology. Siobhan serves on the Stroke Foundation Australia's Stroke Prevention Advisory Committee, the Scientific Advisory Board of Western Australia's Environmental Protection Authority, as an associate editor for the European Journal of Cardiovascular Prevention and is a Senior Fellow of the Higher Education Academy.

Chelsea Francis completed her BPhil Hons at The University of Western Australia in 2022 and is currently completing her MD. Her academic and research interests focus on cardiovascular disease and public health. Chelsea has engaged in various research projects and initiatives aimed at understanding the broader determinants of health, and she is keen to continue contributing to this field throughout her medical career.

Data availability

The datasets may be available from the corresponding author upon request and subject to approval from the WADOH HREC and relevant custodians.

Author contributions

Siobhan Hickling: methodology, analysis, supervision, writing—original draft preparation and review and editing. Chelsea J. Francis: methodology, analysis, writing—original draft preparation. Derek P. Chew: writing—reviewing and editing Biswadev Mitra: writing—reviewing and editing. Graham S. Hillis: conceptualisation, writing—reviewing and editing.

Acknowledgements

The authors wish to thank the Linkage, Data Outputs and Research Data Services Teams at the Data Linkage Services Western Australia, the Centre for Victorian Data Linkage and the Centre for Health Record Linkage for the provision of data. The datasets generated and/or analysed during the current study are not publicly available due to the terms of the ethics approval granted by the Western Australian Department of Health Human Research Ethics Committee (WADOH HREC), Victorian Data Linkage and the Centre for Health Record Linkage and data disclosure policies of the Data Providers.

Funding

This study was funded by the Australian National Health and Medical Research Council (Project Grant 1122792).

References

Author notes