https://orcid.org/0000-0002-6640-6260
Abstract
Quality-of-care and safety of patients with suspected acute coronary syndrome (ACS) would benefit if management was independent of which high-sensitivity cardiac troponin (hs-cTn) assay was used for risk stratification. We aimed to determine the concordance of hs-cTn assays to risk-stratify patients with suspected ACS according to the European Society of Cardiology (ESC) 2020 Guidelines.
Blood samples were obtained at arrival and at 2 h from patients with suspected ACS using four hs-cTn assays. The patients were classified into rule-out/observe/rule-in strata based on the ESC 2020 Guidelines. Concordance was determined among the assays for rule-out/observe/rule-in strata. The prevalences of significant underlying disease (≥50% stenosis on coronary computed tomography or inducible myocardial ischaemia on stress testing) and adjudicated ACS, plus quality-of-care outcomes, were compared. Among 238 patients (52.7 ± 8.0 years; 40.3% female), the overall concordance across assays to classify patients into rule-out/observe/rule-in strata was 74.0% (176/238). Platforms significantly differed for rule-out (89.9 vs. 76.5 vs. 78.6 vs. 86.6%, P < 0.001) and observe strata (6.7 vs. 20.6 vs. 17.7 vs. 9.2%, P < 0.001), but not for rule-in strata (3.4 vs. 2.9 vs. 3.8 vs. 4.2%, P = 0.62). Among patients in ruled-out strata, 19.1–21.6% had significant underlying disease and 3.3–4.2% had ACS. The predicted disposition of patients and cost-of-care differed across the assays (all P < 0.001). When compared with observed strata, conventional troponin-based management and predicted quality-of-care outcomes significantly improved with hs-cTn-based strategies (direct discharge: 21.0 vs. 80.3–90.8%; cost-of-care: $3889 ± 4833 vs. $2578 ± 2896–2894 ± 4371, all P < 0.001).
Among individuals with suspected ACS, patient management may differ depending on which hs-cTn assay is utilized. More data are needed regarding the implications of inter-assay differences.
NCT01084239.
Introduction
Chest discomfort is among the most common complaints in the emergency department (ED). The key elements of the diagnostic algorithm for the evaluation of acute coronary syndrome (ACS) are clinical assessment, including electrocardiogram (ECG), and measurement of cardiac troponin with high-sensitivity troponin (hs-cTn) assays.1,2 The European guidelines for the management of ACS in patients presenting without persistent ST-segment elevation1 published in 2020 (ESC 2020 Guidelines) recommend single-sample rule-in or rule-out and serial blood testing for troponin measurement at either 0/1 or at 0/2 h after presentation. The ESC 2020 Guidelines endorse assay-specific thresholds for several commercially available hs-cTn platforms to achieve standardized care across different assays used for triaging patients with suspected ACS. The purpose of assay-specific thresholds is to eliminate differences among assays emerging from their distinct analytic attributes and thus to establish common cut-points that allow for rendering similar management recommendations across platforms.
In a recent report, using standard analytical benchmarks, e.g. the limit of detection (LOD) and 99th percentile reference values, three hs-cTn assays were significantly discordant when stratifying intermediate-risk patients into the same analytical categories.3 The implications of this finding include the possibility that this heterogeneity in hs-cTn classification might result in differences in patient management and downstream costs of care.
The aim of this analysis was to assess the agreement among four hs-cTn assays when applying the ESC 2020 Guidelines–recommended assay-specific thresholds and to assess whether non-invasive diagnostic testing results, clinical outcomes, and predicted quality-of-care outcomes of patients are consistent across assays and whether they are improved when compared with a conventional troponin-based strategy.
Methods
Patient population and study design
The Rule Out Myocardial Infarction/Ischaemia Using Computer Assisted Tomography (ROMICAT) II trial4 (NCT01084239) was a multi-centre, randomized controlled trial, which included patients with suspected ACS who were referred for further non-invasive diagnostic testing after inconclusive initial ED triage. The subjects were randomized to undergo standard of care vs. coronary computed tomography angiography (CTA). The included patients provided written informed consent, and the study was approved by the local institutional review board.
In this secondary analysis, we included patients who consented to blood draws for biomarker assessment and whose blood samples were analysed with four hs-cTn assays (Figure 1).
Consort flow diagram of the studied patient population. hs-cTn, high-sensitivity cardiac troponin.
Cardiac troponin measurements
Blood samples collected at the time of ED presentation and at 2 h were analysed.5 Blood was collected into EDTA tubes, immediately centrifuged, and stored in microcentrifuge tubes at −80°C until sample assessment. The samples were analysed in a blinded fashion without clinical information.
High-sensitivity cardiac troponin
All blood samples were tested with four hs-cTn assays: Roche Elecsys Cobas Gen 5 assays (e411; Roche Diagnostics, Penzberg, Germany), Abbott ARCHITECT (Abbott Laboratories, Irving, TX, USA), a pre-commercial version of the Siemens Vista (Siemens Diagnostics, Newark, DE, USA), and the Beckman ACCESS (Beckman Coulter, Brea, CA, USA; Table 1).
Analytic characteristics of high-sensitivity troponin assays6
| LOD | LOQ | Overall 99th %tile | Sex-specific 99th %tile, female/male | |
|---|---|---|---|---|
| Roche Elecsys, TnT, ng/L | 5 | 6 | 19 | 14/22 |
| Abbott Architect, TnI, ng/L | 1.7 | 2.3 | 28 | 17/35 |
| Siemens Vista, TnI, ng/L | 2 | 3 | 58.9 | 53.7/78.5 |
| Beckmann Access, TnI, ng/L | 1–2 | 0.9–2.3 | 18.2 | 11.8/19.7 |
| LOD | LOQ | Overall 99th %tile | Sex-specific 99th %tile, female/male | |
|---|---|---|---|---|
| Roche Elecsys, TnT, ng/L | 5 | 6 | 19 | 14/22 |
| Abbott Architect, TnI, ng/L | 1.7 | 2.3 | 28 | 17/35 |
| Siemens Vista, TnI, ng/L | 2 | 3 | 58.9 | 53.7/78.5 |
| Beckmann Access, TnI, ng/L | 1–2 | 0.9–2.3 | 18.2 | 11.8/19.7 |
AMI, acute myocardial infarction; TnI, troponin I; TnT, troponin T; LOD, limit of detection; LOQ, limit of quantification; CV, coefficient of variance.
Analytic characteristics of high-sensitivity troponin assays6
| LOD | LOQ | Overall 99th %tile | Sex-specific 99th %tile, female/male | |
|---|---|---|---|---|
| Roche Elecsys, TnT, ng/L | 5 | 6 | 19 | 14/22 |
| Abbott Architect, TnI, ng/L | 1.7 | 2.3 | 28 | 17/35 |
| Siemens Vista, TnI, ng/L | 2 | 3 | 58.9 | 53.7/78.5 |
| Beckmann Access, TnI, ng/L | 1–2 | 0.9–2.3 | 18.2 | 11.8/19.7 |
| LOD | LOQ | Overall 99th %tile | Sex-specific 99th %tile, female/male | |
|---|---|---|---|---|
| Roche Elecsys, TnT, ng/L | 5 | 6 | 19 | 14/22 |
| Abbott Architect, TnI, ng/L | 1.7 | 2.3 | 28 | 17/35 |
| Siemens Vista, TnI, ng/L | 2 | 3 | 58.9 | 53.7/78.5 |
| Beckmann Access, TnI, ng/L | 1–2 | 0.9–2.3 | 18.2 | 11.8/19.7 |
AMI, acute myocardial infarction; TnI, troponin I; TnT, troponin T; LOD, limit of detection; LOQ, limit of quantification; CV, coefficient of variance.
Conventional cardiac troponin
Conventional troponin T (Stat T, Roche Diagnostics) was measured with a fourth-generation immunoassay on an Elecsys 2010 platform. This assay has a 99th percentile of 0.01 ng/mL, a recommended diagnostic threshold for acute myocardial infarction (MI) of 0.03 ng/mL, and coefficients of variation of 6.6 and 3.8% at concentrations of 0.07 and 2.2 ng/mL, respectively.
Patient management recommendations
The 0/2 h algorithm as defined by the ESC 2020 Guidelines was used to determine the management recommendations (thus rule-out/observe/rule-in classifications) of ACS in haemodynamically stable patients presenting with acute chest complaints (Figure 2). ‘Rule-out’ was defined as low likelihood for non-ST segment MI (NSTEMI) with very low hs-cTn measurement at presentation or with low baseline levels and a lack of a relevant increase after serial hs-cTn testing. ‘Rule-in’ was defined as high likelihood for NSTEMI with at least moderately elevated hs-cTn concentration at presentation or with a clear rise in hs-cTn concentrations. Any patient who did not meet the criteria for rule-out or rule-in was stratified into ‘observe’. Agreement across assays to stratify patients according to ESC 2020 Guidelines–defined assay-specific thresholds at 0 and at 2 h was determined.
The 0/2 h rule-out and rule-in algorithm recommended by the ESC 2020 Guidelines to define the management recommendations for patients with suspected acute coronary syndrome based on fixed thresholds for the assessed four high-sensitivity cardiac troponin assays. ACS, acute coronary syndrome; CCTA, coronary computed tomography angiography; CCU, coronary care unit; ICA, invasive coronary angiography; hs-cTn, high-sensitivity cardiac troponin; NSTEMI, non-ST-segment elevation myocardial infarction.
Non-invasive diagnostic testing
The association of management recommendations with non-invasive diagnostic test findings and the agreement across the studied assays among patients who underwent either coronary CTA or nuclear myocardial stress perfusion imaging [single-photon emission computed tomography (SPECT)] was assessed. A positive result of coronary CTA was defined as the presence of obstructive coronary artery disease (CAD, defined as a luminal narrowing ≥50%). A positive result of SPECT was defined as the presence of any stress-induced ischaemia (reversible myocardial perfusion defect).
Clinical outcomes
The clinical outcomes of patients and their hs-cTn–based management recommendations as per the ESC 2020 Guidelines were assessed. The main outcome of the ROMICAT II trial was clinically adjudicated ACS, which was defined as either MI or unstable angina pectoris (UAP) as adjudicated by an independent events committee (detailed in Supplementary material online).
Observed and predicted quality-of-care outcomes
Quality-of-care outcomes were assessed across the four assays to evaluate the agreement between the platforms. Further, quality-of-care outcomes were compared between hs-cTn–based vs conventional troponin-based strategies.
Quality-of-care outcomes included utilization of any advanced cardiac testing including coronary CTA, exercise treadmill testing, stress echocardiography, SPECT, or invasive coronary angiography (ICA). Consistent with the ESC 2020 Guidelines, it was assumed that patients stratified as ruled-out based on the first hs-cTn measurement (i.e. very low risk) would be diagnosed after 1 h of waiting time and could be discharged from the ED without further testing after 2 h. Patients stratified as ruled-out after the second hs-cTn measurement, and thus were diagnosed after 3 h (blood drawn at 2 h plus 1 h waiting time), could be discharged without further testing 4 h after presentation. All remaining patients were assumed to be managed as in the ROMICAT II trial.
Diagnostic yield was calculated as the number of patients in whom diagnostic testing was abnormal divided by the number of patients who underwent (i.e. observed) or would undergo (i.e. predicted) cardiac testing. Cumulative radiation exposure from coronary CTA, SPECT, and ICA was calculated. Disposition of patients, length of stay, time-to-diagnosis, and need for non-invasive testing were evaluated.
Healthcare costs during the index care episode were assessed using reports from hospital cost-accounting systems and physician billing records. Cost data were available in 649 of 1000 ROMICAT II patients using 2012 US dollars.4,7 To predict the costs when applying the ESC 2020 Guidelines–derived management recommendations, regression coefficients were generated based on a multivariable linear regression model (using information from 649 patients, the regression model explained 86% of the variability in costs with R2 = 0.856) that included total cost as dependent variable and length of stay of >8 h, number of non-invasive cardiac tests, ICA, in-patient hospitalization, percutaneous coronary intervention, and coronary artery bypass graft surgery as independent variables. The total healthcare costs were then calculated by multiplying the predicted cost coefficients by their corresponding parameters for each patient included in our study.
Statistical analysis
Continuous data are presented as mean and standard deviation (SD) or median and inter-quartile range (IQR). Categorical and ordinal variables are presented as numbers and percentages. Comparisons for unpaired data were performed with the use of an independent sample t-test or Wilcoxon rank-sum test for continuous variables, Fisher’s exact test for categorical variables, and Wilcoxon rank-sum test for ordinal variables. For paired data, a paired t-test, Wilcoxon matched-pairs signed-ranks test, the Friedman test, the exact McNemar’s test, or Cochran’s Q test for the overall comparison of management recommendations among the assays was used as appropriate. Kappa values were provided to indicate the statistical strength of concordance or discordance and were interpreted as per Landis and Koch.8 Simulated length of stay, time to diagnosis, healthcare costs, and radiation exposure for all assays were compared using the non-parametric Friedman test, which is an extension of the sign test for more than two paired groups. Because of the exploratory character of this study, the inference was guided with a two-sided 5% false-positive error rate without adjustment for multiple comparisons and clinically meaningful effect sizes. Statistical analyses were performed using Stata 14.2 (StataCorp LP). For all analyses, a two-tailed P value <0.05 was required to reject the null hypothesis.
Results
Study population
Of 1000 randomized subjects of the ROMICAT II trial, 238 (23.8%) had their blood samples analysed with all 4 assays (Figure 1). These patients were not significantly different from those not included in the analysis (see Supplementary material online, Table S1).
The baseline characteristics of the patients are summarized in Table 2. The patients were on average 52.7 ± 8.0 years old, 40.3% (96/238) were female, and most had 0–3 cardiovascular risk factors (90.7%, 216/238). Of those who underwent anatomical testing with coronary CTA, 25.6% (30/117) had obstructive CAD, and of those who were tested with SPECT, 16.3% (n = 7/43) had inducible myocardial ischaemia. In a majority, 91.2% (217/238), of the patients, low (<2) thrombolysis was detected in the MI risk score, and adjudicated ACS was diagnosed in 7.6% (18/238).
Demographic data
| ROMICAT-II (N = 238) | |
|---|---|
| Age, years | 52.7 ± 8.0 |
| Female sex, n (%) | 96 (40.3) |
| BMI, kg/m2 | 29.1 ± 4.7 |
| Cardiovascular risk factors | |
| Hypertension, n (%) | 124 (52.1) |
| Diabetes mellitus, n (%) | 40 (16.8) |
| Dyslipidaemia, n (%) | 105 (44.1) |
| Former/current smoker, n (%) | 118 (49.6) |
| Family history of premature CAD, n (%) | 87 (36.6) |
| Number of cardiovascular risk factors, n (%) | |
| 0–1 | 81 (34.0) |
| 2–3 | 135 (56.7) |
| ≥4 | 22 (9.2) |
| TIMI score, n (%) | |
| 0 | 145 (60.9) |
| 1 | 72 (30.3) |
| 2 | 19 (8.0) |
| ≥3 | 2 (0.8) |
| Prior medication | |
| Aspirin, n (%) | 55 (23.1) |
| Beta-blocker, n (%) | 42 (17.7) |
| Statin, n (%) | 65 (27.3) |
| Non-invasive diagnostic testing | |
| Positive test, n (%) | 34/145 (23.5) |
| Positive CCTAa, n (%) | 30/117 (25.6) |
| Positive SPECTb, n (%) | 7/43 (16.3) |
| Clinical events | |
| ACS, n (%) | 18 (7.6) |
| AMI, n (%) | 5 (2.1) |
| UAP, n (%) | 13 (5.5) |
| ROMICAT-II (N = 238) | |
|---|---|
| Age, years | 52.7 ± 8.0 |
| Female sex, n (%) | 96 (40.3) |
| BMI, kg/m2 | 29.1 ± 4.7 |
| Cardiovascular risk factors | |
| Hypertension, n (%) | 124 (52.1) |
| Diabetes mellitus, n (%) | 40 (16.8) |
| Dyslipidaemia, n (%) | 105 (44.1) |
| Former/current smoker, n (%) | 118 (49.6) |
| Family history of premature CAD, n (%) | 87 (36.6) |
| Number of cardiovascular risk factors, n (%) | |
| 0–1 | 81 (34.0) |
| 2–3 | 135 (56.7) |
| ≥4 | 22 (9.2) |
| TIMI score, n (%) | |
| 0 | 145 (60.9) |
| 1 | 72 (30.3) |
| 2 | 19 (8.0) |
| ≥3 | 2 (0.8) |
| Prior medication | |
| Aspirin, n (%) | 55 (23.1) |
| Beta-blocker, n (%) | 42 (17.7) |
| Statin, n (%) | 65 (27.3) |
| Non-invasive diagnostic testing | |
| Positive test, n (%) | 34/145 (23.5) |
| Positive CCTAa, n (%) | 30/117 (25.6) |
| Positive SPECTb, n (%) | 7/43 (16.3) |
| Clinical events | |
| ACS, n (%) | 18 (7.6) |
| AMI, n (%) | 5 (2.1) |
| UAP, n (%) | 13 (5.5) |
ACS, acute coronary syndrome; AMI, acute myocardial infarction; BMI, body mass index; CAD, coronary artery disease; CCTA, coronary computed tomography angiography; SPECT, single-photon emission computed tomography; TIMI, thrombolysis in myocardial infarction; UAP, unstable angina pectoris.
aPositive coronary CTA: >50% luminal narrowing.
bPositive SPECT: evidence of stress-induced ischaemia defined as reversible myocardial perfusion defect.
Demographic data
| ROMICAT-II (N = 238) | |
|---|---|
| Age, years | 52.7 ± 8.0 |
| Female sex, n (%) | 96 (40.3) |
| BMI, kg/m2 | 29.1 ± 4.7 |
| Cardiovascular risk factors | |
| Hypertension, n (%) | 124 (52.1) |
| Diabetes mellitus, n (%) | 40 (16.8) |
| Dyslipidaemia, n (%) | 105 (44.1) |
| Former/current smoker, n (%) | 118 (49.6) |
| Family history of premature CAD, n (%) | 87 (36.6) |
| Number of cardiovascular risk factors, n (%) | |
| 0–1 | 81 (34.0) |
| 2–3 | 135 (56.7) |
| ≥4 | 22 (9.2) |
| TIMI score, n (%) | |
| 0 | 145 (60.9) |
| 1 | 72 (30.3) |
| 2 | 19 (8.0) |
| ≥3 | 2 (0.8) |
| Prior medication | |
| Aspirin, n (%) | 55 (23.1) |
| Beta-blocker, n (%) | 42 (17.7) |
| Statin, n (%) | 65 (27.3) |
| Non-invasive diagnostic testing | |
| Positive test, n (%) | 34/145 (23.5) |
| Positive CCTAa, n (%) | 30/117 (25.6) |
| Positive SPECTb, n (%) | 7/43 (16.3) |
| Clinical events | |
| ACS, n (%) | 18 (7.6) |
| AMI, n (%) | 5 (2.1) |
| UAP, n (%) | 13 (5.5) |
| ROMICAT-II (N = 238) | |
|---|---|
| Age, years | 52.7 ± 8.0 |
| Female sex, n (%) | 96 (40.3) |
| BMI, kg/m2 | 29.1 ± 4.7 |
| Cardiovascular risk factors | |
| Hypertension, n (%) | 124 (52.1) |
| Diabetes mellitus, n (%) | 40 (16.8) |
| Dyslipidaemia, n (%) | 105 (44.1) |
| Former/current smoker, n (%) | 118 (49.6) |
| Family history of premature CAD, n (%) | 87 (36.6) |
| Number of cardiovascular risk factors, n (%) | |
| 0–1 | 81 (34.0) |
| 2–3 | 135 (56.7) |
| ≥4 | 22 (9.2) |
| TIMI score, n (%) | |
| 0 | 145 (60.9) |
| 1 | 72 (30.3) |
| 2 | 19 (8.0) |
| ≥3 | 2 (0.8) |
| Prior medication | |
| Aspirin, n (%) | 55 (23.1) |
| Beta-blocker, n (%) | 42 (17.7) |
| Statin, n (%) | 65 (27.3) |
| Non-invasive diagnostic testing | |
| Positive test, n (%) | 34/145 (23.5) |
| Positive CCTAa, n (%) | 30/117 (25.6) |
| Positive SPECTb, n (%) | 7/43 (16.3) |
| Clinical events | |
| ACS, n (%) | 18 (7.6) |
| AMI, n (%) | 5 (2.1) |
| UAP, n (%) | 13 (5.5) |
ACS, acute coronary syndrome; AMI, acute myocardial infarction; BMI, body mass index; CAD, coronary artery disease; CCTA, coronary computed tomography angiography; SPECT, single-photon emission computed tomography; TIMI, thrombolysis in myocardial infarction; UAP, unstable angina pectoris.
aPositive coronary CTA: >50% luminal narrowing.
bPositive SPECT: evidence of stress-induced ischaemia defined as reversible myocardial perfusion defect.
Agreement among high-sensitivity cardiac troponin assays to stratify patients into rule-out/observe/rule-in strata per the ESC 2020 Guidelines
After the arrival of troponin measurement, the assays differed in stratifying patients into rule-out (Roche 60.9% vs. Abbott 63.5% vs. Siemens 55.0% vs. Beckman 86.6%, P < 0.001) and observe (Roche 37.0% vs. Abbott 33.6% vs. Siemens 42.0% vs. Beckman 10.5%, P < 0.001) but did not differ for rule-in (Roche 2.1% vs. Abbott 2.9% vs. Siemens 2.9% vs. Beckman 2.9%, P = 0.57), (Table 3 and Supplementary material online, Table S3). Kappa values for rule-out, observe, and rule-in strata were 0.40, 0.35, and 0.76, respectively (Supplementary material online, Table S2). The overall concordance across the assays was 49.6% (118/238; kappa: 0.40), and all four assays agreed to stratify patients to rule-out, observe, and rule-in in 46.3, 11.1, and 40.0%, respectively (Figure 3A).
Agreement among assays in patient management recommendations based on the ESC 2020 Guidelines provided by the 0/2 h algorithm. (A) First hs-cTn measurement. (B) First and the second high-sensitivity cardiac troponin measurements. Concordant: defined as agreement among all four assays, and anything else is considered as discordant. Discordant cases are stratified to more than one management recommendation; therefore, the overall sum of the columns is not equal to the overall sum of the studied patients (n = 238), but it is higher because of the redundancy. hs-cTn, high-sensitivity cardiac troponin.
Agreement between assays in stratifying patients based on baseline (0 h) and serial (0/2 h) high-sensitivity cardiac troponin measurements
| Management recommendation | Roche Elecsys, n (%) | Abbott Architect, n (%) | Siemens Vista, n (%) | Beckman Coulter, n (%) | P-valuesa | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Roche vs. Abbott | Roche vs. Siemens | Abbott vs. Siemens | Roche vs. Beckman | Siemens vs. Beckman | Abbott vs. Beckman | Overall comparison | |||||
| 0 h | |||||||||||
| Rule-out | 145 (60.9) | 151 (63.5) | 131 (55.0) | 206 (86.6) | 0.431 | 0.080 | 0.002 | <0.001 | <0.001 | <0.001 | <0.001 |
| Observe | 88 (37.0) | 80 (33.6) | 100 (42.0) | 25 (10.5) | 0.310 | 0.146 | 0.003 | <0.001 | <0.001 | <0.001 | <0.001 |
| Rule-in | 5 (2.1) | 7 (2.9) | 7 (2.9) | 7 (2.9) | 0.317 | 0.317 | 1.000 | 0.317 | 1.000 | 1.000 | 0.572 |
| Total | 238 (100.0) | 238 (100.0) | 238 (100.0) | 238 (100.0) | — | — | — | — | — | — | — |
| 0/2 h | |||||||||||
| Rule-out | 214 (89.9) | 182 (76.5) | 187 (78.6) | 206 (86.6) | <0.001 | <0.001 | 0.251 | 0.103 | <0.001 | <0.001 | <0.001 |
| Observe | 16 (6.7) | 49 (20.6) | 42 (17.7) | 22 (9.2) | <0.001 | <0.001 | 0.144 | 0.221 | <0.001 | <0.001 | <0.001 |
| Rule-in | 8 (3.4) | 7 (2.9) | 9 (3.8) | 10 (4.2) | 0.655 | 0.706 | 0.317 | 0.480 | 0.655 | 0.180 | 0.623 |
| Total | 238 (100.0) | 238 (100.0) | 238 (100.0) | 238 (100.0) | — | — | — | — | — | — | — |
| Management recommendation | Roche Elecsys, n (%) | Abbott Architect, n (%) | Siemens Vista, n (%) | Beckman Coulter, n (%) | P-valuesa | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Roche vs. Abbott | Roche vs. Siemens | Abbott vs. Siemens | Roche vs. Beckman | Siemens vs. Beckman | Abbott vs. Beckman | Overall comparison | |||||
| 0 h | |||||||||||
| Rule-out | 145 (60.9) | 151 (63.5) | 131 (55.0) | 206 (86.6) | 0.431 | 0.080 | 0.002 | <0.001 | <0.001 | <0.001 | <0.001 |
| Observe | 88 (37.0) | 80 (33.6) | 100 (42.0) | 25 (10.5) | 0.310 | 0.146 | 0.003 | <0.001 | <0.001 | <0.001 | <0.001 |
| Rule-in | 5 (2.1) | 7 (2.9) | 7 (2.9) | 7 (2.9) | 0.317 | 0.317 | 1.000 | 0.317 | 1.000 | 1.000 | 0.572 |
| Total | 238 (100.0) | 238 (100.0) | 238 (100.0) | 238 (100.0) | — | — | — | — | — | — | — |
| 0/2 h | |||||||||||
| Rule-out | 214 (89.9) | 182 (76.5) | 187 (78.6) | 206 (86.6) | <0.001 | <0.001 | 0.251 | 0.103 | <0.001 | <0.001 | <0.001 |
| Observe | 16 (6.7) | 49 (20.6) | 42 (17.7) | 22 (9.2) | <0.001 | <0.001 | 0.144 | 0.221 | <0.001 | <0.001 | <0.001 |
| Rule-in | 8 (3.4) | 7 (2.9) | 9 (3.8) | 10 (4.2) | 0.655 | 0.706 | 0.317 | 0.480 | 0.655 | 0.180 | 0.623 |
| Total | 238 (100.0) | 238 (100.0) | 238 (100.0) | 238 (100.0) | — | — | — | — | — | — | — |
Bold values indicate statistical significance.
aIndicating the differences among the assays.
Agreement between assays in stratifying patients based on baseline (0 h) and serial (0/2 h) high-sensitivity cardiac troponin measurements
| Management recommendation | Roche Elecsys, n (%) | Abbott Architect, n (%) | Siemens Vista, n (%) | Beckman Coulter, n (%) | P-valuesa | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Roche vs. Abbott | Roche vs. Siemens | Abbott vs. Siemens | Roche vs. Beckman | Siemens vs. Beckman | Abbott vs. Beckman | Overall comparison | |||||
| 0 h | |||||||||||
| Rule-out | 145 (60.9) | 151 (63.5) | 131 (55.0) | 206 (86.6) | 0.431 | 0.080 | 0.002 | <0.001 | <0.001 | <0.001 | <0.001 |
| Observe | 88 (37.0) | 80 (33.6) | 100 (42.0) | 25 (10.5) | 0.310 | 0.146 | 0.003 | <0.001 | <0.001 | <0.001 | <0.001 |
| Rule-in | 5 (2.1) | 7 (2.9) | 7 (2.9) | 7 (2.9) | 0.317 | 0.317 | 1.000 | 0.317 | 1.000 | 1.000 | 0.572 |
| Total | 238 (100.0) | 238 (100.0) | 238 (100.0) | 238 (100.0) | — | — | — | — | — | — | — |
| 0/2 h | |||||||||||
| Rule-out | 214 (89.9) | 182 (76.5) | 187 (78.6) | 206 (86.6) | <0.001 | <0.001 | 0.251 | 0.103 | <0.001 | <0.001 | <0.001 |
| Observe | 16 (6.7) | 49 (20.6) | 42 (17.7) | 22 (9.2) | <0.001 | <0.001 | 0.144 | 0.221 | <0.001 | <0.001 | <0.001 |
| Rule-in | 8 (3.4) | 7 (2.9) | 9 (3.8) | 10 (4.2) | 0.655 | 0.706 | 0.317 | 0.480 | 0.655 | 0.180 | 0.623 |
| Total | 238 (100.0) | 238 (100.0) | 238 (100.0) | 238 (100.0) | — | — | — | — | — | — | — |
| Management recommendation | Roche Elecsys, n (%) | Abbott Architect, n (%) | Siemens Vista, n (%) | Beckman Coulter, n (%) | P-valuesa | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Roche vs. Abbott | Roche vs. Siemens | Abbott vs. Siemens | Roche vs. Beckman | Siemens vs. Beckman | Abbott vs. Beckman | Overall comparison | |||||
| 0 h | |||||||||||
| Rule-out | 145 (60.9) | 151 (63.5) | 131 (55.0) | 206 (86.6) | 0.431 | 0.080 | 0.002 | <0.001 | <0.001 | <0.001 | <0.001 |
| Observe | 88 (37.0) | 80 (33.6) | 100 (42.0) | 25 (10.5) | 0.310 | 0.146 | 0.003 | <0.001 | <0.001 | <0.001 | <0.001 |
| Rule-in | 5 (2.1) | 7 (2.9) | 7 (2.9) | 7 (2.9) | 0.317 | 0.317 | 1.000 | 0.317 | 1.000 | 1.000 | 0.572 |
| Total | 238 (100.0) | 238 (100.0) | 238 (100.0) | 238 (100.0) | — | — | — | — | — | — | — |
| 0/2 h | |||||||||||
| Rule-out | 214 (89.9) | 182 (76.5) | 187 (78.6) | 206 (86.6) | <0.001 | <0.001 | 0.251 | 0.103 | <0.001 | <0.001 | <0.001 |
| Observe | 16 (6.7) | 49 (20.6) | 42 (17.7) | 22 (9.2) | <0.001 | <0.001 | 0.144 | 0.221 | <0.001 | <0.001 | <0.001 |
| Rule-in | 8 (3.4) | 7 (2.9) | 9 (3.8) | 10 (4.2) | 0.655 | 0.706 | 0.317 | 0.480 | 0.655 | 0.180 | 0.623 |
| Total | 238 (100.0) | 238 (100.0) | 238 (100.0) | 238 (100.0) | — | — | — | — | — | — | — |
Bold values indicate statistical significance.
aIndicating the differences among the assays.
After the second hs-cTn measurement, disagreement among the assays remained for rule-out (Roche 89.9% vs. Abbott 76.5% vs. Siemens 78.6% vs. Beckman 86.6%, P < 0.001) and observe strata (Roche 6.7% vs. Abbott 20.6% vs. Siemens 17.7% vs. Beckman 9.2%, P < 0.001), while for rule-in strata, the assays remained similar (Roche 3.4% vs. Abbott 2.9% vs. Siemens 3.8% vs. Beckman 4.2%, P = 0.62; Table 3). Kappa values for rule-out, observe, and rule-in strata were 0.53, 0.42, and 0.65, respectively. The overall concordance across the assays was 74.0% (176/238; kappa: 0.50), and assays agreed in 74.3 9.4, and 31.3% for rule-out, observe, and rule-in strata, respectively (Figure 3B).
Non-invasive diagnostic testing results and clinical outcomes
A total of 60.9% (145/238) of the studied patients received non-invasive testing (102/145 CTA, 28/145 SPECT, and 15/145 CTA and SPECT; see Supplementary material online, Table S4). Among the patients stratified to rule-out, 21.6% (29/134) vs. 20.0% (23/115) vs. 21.4% (25/117) vs. 19.1% (24/126) for Roche, Abbott, Siemens, and Beckman, respectively, had obstructive CAD or inducible myocardial ischaemia. Among these patients, 21/29 (72.4%) patients would have been risk-stratified as rule-out by all assays.
Among patients who were stratified as rule-out by at least one of the four assays, 4.2 (9/214) vs. 3.3% (6/182) vs. 3.7% (7/187) vs. 3.4% (7/206) for Roche, Abbott, Siemens, and Beckman, respectively, were diagnosed with ACS (2.8–3.7% UAP and 0.0–0.6% MI).
Quality-of-care outcomes
Agreement among the four assays
The predicted rate of non-invasive testing differed modestly among the four assays, i.e. the rate of no testing was 92.0% (219/238) vs. 80.7% (238/192) vs. 82.8% (197/238) vs. 88.7% (211/238) for Roche, Abbott, Siemens, and Beckman, respectively (Table 4). However, the rates of invasive testing and interventions and radiation exposure were similar across the hs-cTn assays (all P > 0.05). Disposition rates and treatment times differed moderately across the assays, with ED discharge rates of 80.3–90.8% (P < 0.001), the mean length-of-stay of 6.7 ± 15.8–10.1 ± 26.1 h (P < 0.001), and the mean time-to-diagnosis of 2.6 ± 6.2–4.3 ± 7.5 h (P < 0.001). Mean healthcare costs differed slightly across the assays with $2571 ± 2897 vs. $2784 ± 3234 vs. $2894 ± 4371 vs. $2651 ± 3138 per patient (P < 0.001) and $1988 ± 702 vs. $2070 ± 729 vs. $2051 ± 704 vs. $1973 ± 681 when excluding ACS patients (P < 0.001) for Roche, Abbott, Siemens, and Beckman, respectively.
Advanced cardiac testing, radiation exposure, disposition, length of hospital stay, time-to-diagnosis, and healthcare costs as observed in the Rule Out Myocardial Infarction/Ischaemia Using Computer Assisted Tomography II trial vs. as predicted using decision rules based on highly sensitive troponin and cardiovascular risk factors
| As observed (N = 238) | As predicted | ||||
|---|---|---|---|---|---|
| Roche Elecsys (N = 238) | Abbott Architect (N = 238) | Siemens Vista (N = 238) | Beckman Coulter (N = 238) | ||
| Non-invasive diagnostic testing, n (%)a | |||||
| No testing | 21 (8.8) | 219 (92.0) | 192 (80.7) | 197 (82.8) | 211 (88.7) |
| 1 test | 171 (71.9) | 9 (3.8) | 29 (12.2) | 26 (10.9) | 12 (5.0) |
| ≥ 2 tests | 46 (19.3) | 10 (4.2) | 17 (7.1) | 15 (6.3) | 15 (6.3) |
| Diagnostic yield, n (%)b | 14.8 | 31.6 | 28.3 | 26.8 | 37.0 |
| Invasive coronary angiography, n (%) | 19 (8.0) | 8 (3.4) | 10 (4.2) | 10 (4.2) | 10 (4.2) |
| Intervention, n (%) | |||||
| PCI | 12 (5.0) | 7 (2.9) | 9 (3.8) | 8 (3.4) | 8 (3.4) |
| CABG | 1 (0.4) | 0 (0.0) | 0 (0.0) | 1 (0.4) | 0 (0.0) |
| Cumulative radiation exposure (mSv/patient)c | 9.1 ± 11.8 | 1.7 ± 7.9 | 2.7 ± 8.9 | 2.7 ± 8.7 | 2.4 ± 9.0 |
| Disposition | |||||
| ED discharge | 50 (21.0) | 216 (90.8) | 191 (80.3) | 197 (82.8) | 212 (89.1) |
| Observational unit admission | 147 (61.8) | 8 (3.4) | 31 (13.0) | 26 (10.9) | 13 (5.5) |
| Hospital admission | 38 (16.0) | 13 (5.5) | 15 (6.3) | 14 (5.9) | 12 (5.0) |
| Left against medical advice | 3 (1.3) | 1 (0.4) | 1 (0.4) | 1 (0.4) | 1 (0.4) |
| Length of hospital stay, h | |||||
| Median (IQR) | 23.3 (8.4–28.7) | 2.0 (2.0–4.0) | 2.0 (2.0–4.0) | 2.0 (2.0–4.0) | 2.0 (2.0–2.0) |
| Mean ± SD | 25.1 ± 28.5 | 6.7 ± 15.8 | 9.2 ± 17.6 | 10.1 ± 26.1 | 6.9 ± 17.2 |
| Time-to-diagnosis, h | |||||
| Median (IQR) | 8.2 (6.0–22.3) | 1.0 (1.0–3.0) | 1.0 (1.0–3.0) | 1.0 (1.0–3.0) | 1.0 (1.0–1.0) |
| Mean ± SD | 14.5 ± 12.9 | 2.8 ± 5.4 | 4.3 ± 7.5 | 4.2 ± 7.2 | 2.6 ± 6.2 |
| Healthcare cost per patient in US $ | |||||
| Median (IQR) | 2698 (1837–2698) | 1837 (1837–1837) | 1837 (1837–2698) | 1837 (1837–2698) | 1837 (1837–1837) |
| Mean ± SD | 3889 ± 4833 | 2571 ± 2896 | 2784 ± 3234 | 2894 ± 4371 | 2651 ± 3138 |
| Non-ACS patients only | |||||
| Median (IQR) | 2698 (1837–2698) | 1837 (1837–1837) | 1837 (1837–1837) | 1837 (1837–1837) | 1837 (1837–1837) |
| Mean ± SD | 2784 ± 1341 | 1980 ± 696 | 2070 ± 729 | 2051 ± 704 | 1965 ± 670 |
| As observed (N = 238) | As predicted | ||||
|---|---|---|---|---|---|
| Roche Elecsys (N = 238) | Abbott Architect (N = 238) | Siemens Vista (N = 238) | Beckman Coulter (N = 238) | ||
| Non-invasive diagnostic testing, n (%)a | |||||
| No testing | 21 (8.8) | 219 (92.0) | 192 (80.7) | 197 (82.8) | 211 (88.7) |
| 1 test | 171 (71.9) | 9 (3.8) | 29 (12.2) | 26 (10.9) | 12 (5.0) |
| ≥ 2 tests | 46 (19.3) | 10 (4.2) | 17 (7.1) | 15 (6.3) | 15 (6.3) |
| Diagnostic yield, n (%)b | 14.8 | 31.6 | 28.3 | 26.8 | 37.0 |
| Invasive coronary angiography, n (%) | 19 (8.0) | 8 (3.4) | 10 (4.2) | 10 (4.2) | 10 (4.2) |
| Intervention, n (%) | |||||
| PCI | 12 (5.0) | 7 (2.9) | 9 (3.8) | 8 (3.4) | 8 (3.4) |
| CABG | 1 (0.4) | 0 (0.0) | 0 (0.0) | 1 (0.4) | 0 (0.0) |
| Cumulative radiation exposure (mSv/patient)c | 9.1 ± 11.8 | 1.7 ± 7.9 | 2.7 ± 8.9 | 2.7 ± 8.7 | 2.4 ± 9.0 |
| Disposition | |||||
| ED discharge | 50 (21.0) | 216 (90.8) | 191 (80.3) | 197 (82.8) | 212 (89.1) |
| Observational unit admission | 147 (61.8) | 8 (3.4) | 31 (13.0) | 26 (10.9) | 13 (5.5) |
| Hospital admission | 38 (16.0) | 13 (5.5) | 15 (6.3) | 14 (5.9) | 12 (5.0) |
| Left against medical advice | 3 (1.3) | 1 (0.4) | 1 (0.4) | 1 (0.4) | 1 (0.4) |
| Length of hospital stay, h | |||||
| Median (IQR) | 23.3 (8.4–28.7) | 2.0 (2.0–4.0) | 2.0 (2.0–4.0) | 2.0 (2.0–4.0) | 2.0 (2.0–2.0) |
| Mean ± SD | 25.1 ± 28.5 | 6.7 ± 15.8 | 9.2 ± 17.6 | 10.1 ± 26.1 | 6.9 ± 17.2 |
| Time-to-diagnosis, h | |||||
| Median (IQR) | 8.2 (6.0–22.3) | 1.0 (1.0–3.0) | 1.0 (1.0–3.0) | 1.0 (1.0–3.0) | 1.0 (1.0–1.0) |
| Mean ± SD | 14.5 ± 12.9 | 2.8 ± 5.4 | 4.3 ± 7.5 | 4.2 ± 7.2 | 2.6 ± 6.2 |
| Healthcare cost per patient in US $ | |||||
| Median (IQR) | 2698 (1837–2698) | 1837 (1837–1837) | 1837 (1837–2698) | 1837 (1837–2698) | 1837 (1837–1837) |
| Mean ± SD | 3889 ± 4833 | 2571 ± 2896 | 2784 ± 3234 | 2894 ± 4371 | 2651 ± 3138 |
| Non-ACS patients only | |||||
| Median (IQR) | 2698 (1837–2698) | 1837 (1837–1837) | 1837 (1837–1837) | 1837 (1837–1837) | 1837 (1837–1837) |
| Mean ± SD | 2784 ± 1341 | 1980 ± 696 | 2070 ± 729 | 2051 ± 704 | 1965 ± 670 |
CABG, coronary artery bypass graft; ED, emergency department; PCI, percutaneous coronary intervention.
aCardiac testing included coronary CTA, exercise treadmill test, stress echocardiography, nuclear myocardial perfusion imaging, and ICA.
bDiagnostic yield was calculated as number of patients in whom diagnostic test was abnormal (stress test positive for ischaemia, coronary CTA, or ICA with >50% stenosis) divided by the number of patients who underwent (as observed) or could undergo (as predicted) cardiac testing.
cRadiation exposure included exposure from coronary computed tomography, nuclear myocardial perfusion imaging, and ICA.
Advanced cardiac testing, radiation exposure, disposition, length of hospital stay, time-to-diagnosis, and healthcare costs as observed in the Rule Out Myocardial Infarction/Ischaemia Using Computer Assisted Tomography II trial vs. as predicted using decision rules based on highly sensitive troponin and cardiovascular risk factors
| As observed (N = 238) | As predicted | ||||
|---|---|---|---|---|---|
| Roche Elecsys (N = 238) | Abbott Architect (N = 238) | Siemens Vista (N = 238) | Beckman Coulter (N = 238) | ||
| Non-invasive diagnostic testing, n (%)a | |||||
| No testing | 21 (8.8) | 219 (92.0) | 192 (80.7) | 197 (82.8) | 211 (88.7) |
| 1 test | 171 (71.9) | 9 (3.8) | 29 (12.2) | 26 (10.9) | 12 (5.0) |
| ≥ 2 tests | 46 (19.3) | 10 (4.2) | 17 (7.1) | 15 (6.3) | 15 (6.3) |
| Diagnostic yield, n (%)b | 14.8 | 31.6 | 28.3 | 26.8 | 37.0 |
| Invasive coronary angiography, n (%) | 19 (8.0) | 8 (3.4) | 10 (4.2) | 10 (4.2) | 10 (4.2) |
| Intervention, n (%) | |||||
| PCI | 12 (5.0) | 7 (2.9) | 9 (3.8) | 8 (3.4) | 8 (3.4) |
| CABG | 1 (0.4) | 0 (0.0) | 0 (0.0) | 1 (0.4) | 0 (0.0) |
| Cumulative radiation exposure (mSv/patient)c | 9.1 ± 11.8 | 1.7 ± 7.9 | 2.7 ± 8.9 | 2.7 ± 8.7 | 2.4 ± 9.0 |
| Disposition | |||||
| ED discharge | 50 (21.0) | 216 (90.8) | 191 (80.3) | 197 (82.8) | 212 (89.1) |
| Observational unit admission | 147 (61.8) | 8 (3.4) | 31 (13.0) | 26 (10.9) | 13 (5.5) |
| Hospital admission | 38 (16.0) | 13 (5.5) | 15 (6.3) | 14 (5.9) | 12 (5.0) |
| Left against medical advice | 3 (1.3) | 1 (0.4) | 1 (0.4) | 1 (0.4) | 1 (0.4) |
| Length of hospital stay, h | |||||
| Median (IQR) | 23.3 (8.4–28.7) | 2.0 (2.0–4.0) | 2.0 (2.0–4.0) | 2.0 (2.0–4.0) | 2.0 (2.0–2.0) |
| Mean ± SD | 25.1 ± 28.5 | 6.7 ± 15.8 | 9.2 ± 17.6 | 10.1 ± 26.1 | 6.9 ± 17.2 |
| Time-to-diagnosis, h | |||||
| Median (IQR) | 8.2 (6.0–22.3) | 1.0 (1.0–3.0) | 1.0 (1.0–3.0) | 1.0 (1.0–3.0) | 1.0 (1.0–1.0) |
| Mean ± SD | 14.5 ± 12.9 | 2.8 ± 5.4 | 4.3 ± 7.5 | 4.2 ± 7.2 | 2.6 ± 6.2 |
| Healthcare cost per patient in US $ | |||||
| Median (IQR) | 2698 (1837–2698) | 1837 (1837–1837) | 1837 (1837–2698) | 1837 (1837–2698) | 1837 (1837–1837) |
| Mean ± SD | 3889 ± 4833 | 2571 ± 2896 | 2784 ± 3234 | 2894 ± 4371 | 2651 ± 3138 |
| Non-ACS patients only | |||||
| Median (IQR) | 2698 (1837–2698) | 1837 (1837–1837) | 1837 (1837–1837) | 1837 (1837–1837) | 1837 (1837–1837) |
| Mean ± SD | 2784 ± 1341 | 1980 ± 696 | 2070 ± 729 | 2051 ± 704 | 1965 ± 670 |
| As observed (N = 238) | As predicted | ||||
|---|---|---|---|---|---|
| Roche Elecsys (N = 238) | Abbott Architect (N = 238) | Siemens Vista (N = 238) | Beckman Coulter (N = 238) | ||
| Non-invasive diagnostic testing, n (%)a | |||||
| No testing | 21 (8.8) | 219 (92.0) | 192 (80.7) | 197 (82.8) | 211 (88.7) |
| 1 test | 171 (71.9) | 9 (3.8) | 29 (12.2) | 26 (10.9) | 12 (5.0) |
| ≥ 2 tests | 46 (19.3) | 10 (4.2) | 17 (7.1) | 15 (6.3) | 15 (6.3) |
| Diagnostic yield, n (%)b | 14.8 | 31.6 | 28.3 | 26.8 | 37.0 |
| Invasive coronary angiography, n (%) | 19 (8.0) | 8 (3.4) | 10 (4.2) | 10 (4.2) | 10 (4.2) |
| Intervention, n (%) | |||||
| PCI | 12 (5.0) | 7 (2.9) | 9 (3.8) | 8 (3.4) | 8 (3.4) |
| CABG | 1 (0.4) | 0 (0.0) | 0 (0.0) | 1 (0.4) | 0 (0.0) |
| Cumulative radiation exposure (mSv/patient)c | 9.1 ± 11.8 | 1.7 ± 7.9 | 2.7 ± 8.9 | 2.7 ± 8.7 | 2.4 ± 9.0 |
| Disposition | |||||
| ED discharge | 50 (21.0) | 216 (90.8) | 191 (80.3) | 197 (82.8) | 212 (89.1) |
| Observational unit admission | 147 (61.8) | 8 (3.4) | 31 (13.0) | 26 (10.9) | 13 (5.5) |
| Hospital admission | 38 (16.0) | 13 (5.5) | 15 (6.3) | 14 (5.9) | 12 (5.0) |
| Left against medical advice | 3 (1.3) | 1 (0.4) | 1 (0.4) | 1 (0.4) | 1 (0.4) |
| Length of hospital stay, h | |||||
| Median (IQR) | 23.3 (8.4–28.7) | 2.0 (2.0–4.0) | 2.0 (2.0–4.0) | 2.0 (2.0–4.0) | 2.0 (2.0–2.0) |
| Mean ± SD | 25.1 ± 28.5 | 6.7 ± 15.8 | 9.2 ± 17.6 | 10.1 ± 26.1 | 6.9 ± 17.2 |
| Time-to-diagnosis, h | |||||
| Median (IQR) | 8.2 (6.0–22.3) | 1.0 (1.0–3.0) | 1.0 (1.0–3.0) | 1.0 (1.0–3.0) | 1.0 (1.0–1.0) |
| Mean ± SD | 14.5 ± 12.9 | 2.8 ± 5.4 | 4.3 ± 7.5 | 4.2 ± 7.2 | 2.6 ± 6.2 |
| Healthcare cost per patient in US $ | |||||
| Median (IQR) | 2698 (1837–2698) | 1837 (1837–1837) | 1837 (1837–2698) | 1837 (1837–2698) | 1837 (1837–1837) |
| Mean ± SD | 3889 ± 4833 | 2571 ± 2896 | 2784 ± 3234 | 2894 ± 4371 | 2651 ± 3138 |
| Non-ACS patients only | |||||
| Median (IQR) | 2698 (1837–2698) | 1837 (1837–1837) | 1837 (1837–1837) | 1837 (1837–1837) | 1837 (1837–1837) |
| Mean ± SD | 2784 ± 1341 | 1980 ± 696 | 2070 ± 729 | 2051 ± 704 | 1965 ± 670 |
CABG, coronary artery bypass graft; ED, emergency department; PCI, percutaneous coronary intervention.
aCardiac testing included coronary CTA, exercise treadmill test, stress echocardiography, nuclear myocardial perfusion imaging, and ICA.
bDiagnostic yield was calculated as number of patients in whom diagnostic test was abnormal (stress test positive for ischaemia, coronary CTA, or ICA with >50% stenosis) divided by the number of patients who underwent (as observed) or could undergo (as predicted) cardiac testing.
cRadiation exposure included exposure from coronary computed tomography, nuclear myocardial perfusion imaging, and ICA.
Predicted, high-sensitivity cardiac troponin–based strategies vs. observed, conventional troponin-based strategy
The predicted rates of any testing (8.0–19.3 vs. 91.2%, P < 0.001) and radiation exposure (1.7 ± 7.9–2.7 ± 8.9 vs. 9.1 ± 11.8 mSv/patient, P < 0.001) per each hs-cTn assay were markedly lower compared with the observed, conventional troponin-based measurements (Table 4). However, the diagnostic yield of testing (26.8–37.0 vs. 14.8%, P > 0.05), the invasive testing rates (3.4–4.2 vs. 8.0%, P > 0.05), and intervention rates (2.9–3.8 vs. 5.4%, P > 0.05) were similar compared with the observed data. Predicted patient pathways based on the hs-cTn–based strategies resulted in a higher discharge rate compared with the observed conventional troponin-based strategy (80.3–90.8 vs. 21.0%). Consequently, length of stay (6.7 ± 15.8–10.1 ± 26.1 vs. 25.1 ± 28.5 h, P < 0.001) was substantially lower with the hs-cTn–based strategies, with a faster time-to-diagnosis compared with the observed data (2.6 ± 6.2–4.3 ± 7.5 vs. 14.5 ± 12.9 h, P < 0.001). Ultimately, the predicted hs-cTn–based strategies improved utility-of-care with lower costs compared with the observed data ($2571 ± 2896–2894 ± 4371 vs. $3889 ± 4,833, P < 0.001).
Discussion
In this secondary analysis of the ROMICAT II trial, we assessed the agreement across four hs-cTn assays to stratify patients with suspected ACS to rule-out/observe/rule-in strata based on the ESC 2020 Guidelines. We report the following key findings: (i) significant differences are present among four hs-cTn assays to stratify patients to rule-out and observe clinical management recommendations but not for rule-in; (ii) by using the ESC 2020 Guidelines-recommended fixed assay-specific thresholds, ∼20% of patients with positive stress tests or obstructive CAD on CTA were stratified to ruled-out and ∼4% had clinically adjudicated ACS (2.8–3.7% UAP and 0.0–0.6% MI); (iii) a significant disagreement was present among the four assays across the predicted quality-of-care outcomes, but utilizing the hs-cTn–based management strategies improved all quality-of-care outcomes compared with conventional troponin-based management.
ESC 2020 Guidelines–recommended assay-specific thresholds
Prior studies have demonstrated discrepancies among assays when using the 99th percentile as a threshold for rule-in ACS and described misclassification rates of 3–17% and greater agreement after re-deriving the 99th percentile.9,10 Additionally, hs-cTn assays showed significant disagreement for rule-out management recommendation when using thresholds suggested by studies establishing the 0/2 h algorithms.3 Our current analysis extends our understanding of the observed discordances among assays, because, to the best of our knowledge, this is the first assessment of the concordance among hs-cTn assays by using the ESC-recommended assay-specific thresholds to triage patients with suspected ACS. Our findings corroborate prior observations in that we identified significant differences among the assays. When assessing the agreement across four hs-cTn–based assay-specific thresholds derived from large population-based studies as included in the ESC 2020 Guidelines, we found significant disagreement in stratifying patients to rule-out and observe, with kappa values suggesting only a fair to moderate agreement at both 0 and 2 h timepoints (kappa for observe and rule-in 0.40, 0.35, and 0.53, 0.42, respectively). This finding can be explained in part by the fact that the suggested cut-points were derived by using different reference populations.11,12 Another potential reason for the observed discordances is that different assays measure different troponin isotypes (i.e. troponin I: Abbott, Siemens, and Beckmann and troponin T: Roche), which behave differently to some extent.13,14 However, given that the thresholds in the ESC 2020 Guidelines recommended 0/2 h are assay specific, the differences between the isotypes are anticipated to have no impact on the assay performances as all were calibrated to detect troponin changes/dynamics along the same clinical outcomes. Further, the agreement between troponin-I assays was not higher compared with troponin-I vs. troponin-T assays (Abbott vs. Siemens: 90.3%; Abbott vs. Beckmann: 82.8%; Abbott vs. Roche: 82.4%; Siemens vs. Roche: 82.8%, Siemens vs. Beckmann: 85.3%; Beckmann vs. Roche: 88.2%); thus, we speculate that the observed differences are mainly occurring in the rule-out strata, more likely to be associated with threshold-related discordance vs. being the result of differences in troponin release dynamics between isotypes. Thresholds for rule-in strata rendered similar results across the assays with substantial agreement at 0 and 2 h (kappa 0.76 and 0.65, respectively); thus, thresholds for the identification of patients with myocardial ischaemia seem to be universally more fine-tuned—with the only caveat of having a few observations for the rule-in strata and thus the lack of discordance could be the result of Type 2 statistical error. It is important to emphasize though that the ESC algorithms are triage algorithms and not discharge/admit strategies. Further adjudication in the rule-out or observe zone and individualized risk assessment is mandatory.
Non-invasive diagnostic testing results
Despite disagreement across the assays in the number of patients who were stratified to rule-out clinical management recommendation (76.5–89.9%, P < 0.001), ∼20% of patients had obstructive CAD or inducible myocardial ischaemia. When assessing the overlap of these patients among assays, we found that 72% (n = 21/29) were stratified as rule-out by all four assays. Two considerations are important to note: first, in a prior analysis using analytical benchmarks and LOD as a threshold to identify those for whom MI can be ruled-out, we found a similar extent of underlying CAD among patients stratified to rule-out.3 This implies that assay-specific thresholds endorsed by the ESC 2020 Guidelines are similar in their performance for rule-out patients as each assay’s LOD. Second consideration is that the ESC 2020 Guideline management recommendations are not constrained to triaging of patients based on troponin but suggest further non-invasive or invasive testing options for each risk stratum (rule-out/observe/rule-in). Among patients who were stratified to rule-out, further diagnostic testing may be triggered and lead to the recognition of undetected disease. While the lack of detectable troponin is a good predictor of major adverese cardiovascular events-free survival in the short term and not the lack of significant ischaemia or CAD, this is especially important because of the prognostic value of inducible myocardial ischaemia/obstructive CAD.15 However, the ESC hs-cTn algorithms are designed to identify ACS and not to determine underlying CAD.
Quality-of-care outcomes
Predicted quality-of-care outcomes were affected substantially by the differences across the assays in the proportion of patients stratified as rule-out and observe and consequently resulted in similar discrepancies in quality-of-care outcomes. For example, diagnostic yield, disposition of patients, length of stay, and time-to-diagnosis endpoints would be significantly different based on the hs-cTn assay used to place individuals in the rule-out category. On the other hand, endpoints mainly determined by the rule-in management strata were similar across hs-cTn methods; thus, the rate of invasive testing, rate of interventions, and radiation exposure were also similar across groups. However, even if the care of those who were recommended to be in ruled-in strata entails greater healthcare costs, the differences among the assays in the proportion of patients who were stratified to rule-out still affected the overall healthcare costs and thus cost-of-care was modestly but significantly different across the assays.
When projected hs-cTn assay-based strategies were compared with conventional troponin-based management, we observed an improvement in all quality-of-care outcomes consistent with most16–18 but not with all established data.19 This inconsistency may be related to the pre-test probability of ACS in various study cohorts. In this study, which assessed a population with low-to-intermediate risk for ACS, the use of hs-cTn assay-based alternatives markedly decreased the admission rate, suggesting the ESC 2020 Guidelines perform well compared with a conventional troponin-based alternative. The nearly four-fold difference in discharge rate between conventional troponin-based vs. hs-cTn–based strategies potentially further explains the substantial improvement observed for other subsequent predicted quality-of-care outcomes, such as length of hospital stay, time-to-diagnosis, and cost-of-care.
Strengths and limitations
This study is the first to quantify the degree of agreement across four hs-cTn assays when using the ESC 2020 0/2 h diagnostic algorithm with assay-specific fixed thresholds. These key findings are based on the assessment of a cohort with low to intermediate likelihood for ACS. As the low to intermediate ACS risk cohort represents a large population for whom objective measures may be of greatest utility, our findings of improved results using hs-cTn further support the adoption of hs-cTn as a standard of care and are consistent with the most recent guidelines from other societies.20
There are limitations to this study. First, ROMICAT II was a multi-centre study based in the USA. This is a limitation, given that the ESC 2020 Guidelines comprised studies and results derived based on European cohorts. However, given that the main aim of this analysis is to assess agreement among hs-cTn assays to stratify patients and not to predict patient outcomes, the cohort that is used is less relevant. Second, our patient cohort represents a selected patient population presenting to EDs with suspected ACS. However, patients with an intermediate likelihood of ACS referred to further non-invasive testing after inconclusive initial triage (normal conventional troponin-T and non-ischaemic ECG) represent ∼20% of all-comer chest pain patients and are the highest diagnostic challenge for safe and efficient triage. Thus, differences/similarities among the hs-cTn assays are the most critical to assess in this population. Third, chest pain onset in the ROMICAT II trial was not recorded, and the ESC 2020 0/2 h algorithm was designed for patients with chest pain onset >3 h. We anticipate that the majority of our patients was fulfilling this criterion by the design of the trial and potential delays in patient evaluation. Fourth, the Siemens Dimension Vista platform used for hs-cTn testing in the ROMICAT II trial was a pre-commercial model, and the 0/2 h algorithm was developed for the Siemens Centaur platform. However, the differences between the two assays are relatively small with the Centaur being slightly more sensitive.21 This may have resulted in a minor difference in the stratification of patients, although it is not likely that it significantly influenced the agreement of the assays. However, this result warrants further investigation. Fourth, the costs of care were based on the US healthcare system. Because the demonstrated differences in quality-of-care outcomes reflect patient management and are independent of financing, they are not affected by differences in financing between countries; therefore, the conclusion that hs-cTn–based approaches are cost saving compared with conventional troponin-based strategies can be universally drawn.
Conclusions
In this secondary analysis of the ROMICAT II trial, four hs-cTn assays significantly disagreed in stratifying patients with suspected ACS to rule-out and observe management strategies but not for rule-in. A substantial proportion of patients stratified to rule-out had positive findings in non-invasive imaging and had clinical events. Quality-of-care outcomes differed across the four hs-cTn assays but markedly improved when compared with conventional troponin-based data.
Supplementary material
Supplementary material is available at European Heart Journal: Acute Cardiovascular Care.
Funding
The original ROMICAT II study was funded by grants from the National Heart, Lung, and Blood Institute (NHLBI; U01HL092040 and U01HL092022). The content of this manuscript is solely the responsibility of the authors and does not necessarily reflect the views of the National Heart, Lung, and Blood Institute, the National Institutes of Health, or the US Department of Health and Human Services.
Data availability
The data underlying this article were provided by a third party under license/by permission. Data will be shared on request to the corresponding author with permission from the third party.
References
Author notes
Conflict of interest: J.T.N. has received research funds from Roche Diagnostics, Ortho Diagnostics, and Alere/Biosite to the Massachusetts General Hospital unrelated to this research. W.F.P. has received grant support from Abbott, Boehringer Ingelheim, Braincheck, CSL Behring, Daiichi-Sankyo, ImmunArray, Janssen, Ortho Clinical Diagnostics, Portola, Relypsa, Roche, Salix, and Siemens; consulting income from Abbott, Astra-Zeneca, Bayer, Beckman, Boehringer Ingelheim, Ischemia Care, Dx, ImmunArray, Instrument Labs, Janssen, Nabriva, Ortho Clinical Diagnostics, Relypsa, Roche, Quidel, Salix, and Siemens; expert testimony from Johnson and Johnson; and reports stock/ownership interest for AseptiScope Inc., Brainbox Inc., Comprehensive Research Associates LLC, Emergencies in Medicine LLC, and Ischemia DX LLC unrelated to this research. M.T.L. reports funding to his institution from Astra-Zeneca/MedImmune and Kowa and consulting fees from PQBypass unrelated to this research. B.M. reports personal fees from Biotronik, Abbott, Astra-Zeneca, Boehringer Ingelheim, and Novartis and institutional grants from Medtronic and Boston Scientific unrelated to this research. J.L.J. is a Trustee of the American College of Cardiology; is a Board member of Imbria Pharmaceuticals; has received grant support from Abbott, Applied Therapeutics, Innolife, HeartFlow, Janssen, Novartis Pharmaceuticals, and Roche Diagnostics; has received consulting income from Abbott, Janssen, Novartis, Pfizer, Merck, and Roche Diagnostics; and participates in clinical endpoint committees/data safety monitoring boards for Abbott, AbbVie, Boehringer Ingelheim, CVRx, Intercept, Eidos, Janssen, and Takeda unrelated to this research. W.K. reports personal fees from AstraZeneca, Novartis, Pfizer, The Medicines Company, DalCor, Kowa, Amgen, Corvidia, Genentech, Esperion, Amarin, Daiichi-Sanky, OMEICOS, Novo Nordisk, and Sanofi and grants and non-financial support from Beckmann, Singulex, Abbott, and Roche Diagnostics unrelated to this research. M.F. was supported by the grant from the American Heart Association (Fellow to Faculty Award #13FTF16450001) and has received consulting income from Biograph, Inc., Siemens Healthineers, and Elucid unrelated to this research. U.H. reported receiving research support from Duke University (Abbott), HeartFlow, Kowa Company Limited, and MedImmune/Astrazeneca and receiving consulting fees from Duke University (NIH), Recor Medical, Clinical Cardiovascular Sciences, and MedTrace unrelated to this research. The remaining authors have reported nothing to disclose.
Comments