Association of High-Sensitivity Cardiac Troponin I Concentration With Cardiac Outcomes in Patients With Suspected Acute Coronary Syndrome

Question  What is the optimal high-sensitivity cardiac troponin I concentration at presentation to risk-stratify patients with suspected acute coronary syndrome?

Findings  In an individual patient-level meta-analysis of 22 457 patients from 9 countries, troponin I concentrations were less than 5 ng/L in 49%, among whom 5 per 1000 patients had a myocardial infarction or cardiac death at 30 days.

Meaning  Among patients with suspected acute coronary syndrome, a high-sensitivity cardiac troponin I threshold of less than 5 ng/L identified patients at low risk of cardiac events; further research is needed to assess the clinical utility of this test.

Importance  High-sensitivity cardiac troponin I testing is widely used to evaluate patients with suspected acute coronary syndrome. A cardiac troponin concentration of less than 5 ng/L identifies patients at presentation as low risk, but the optimal threshold is uncertain.

Objective  To evaluate the performance of a cardiac troponin I threshold of 5 ng/L at presentation as a risk stratification tool in patients with suspected acute coronary syndrome.

Data Sources  Systematic search of MEDLINE, EMBASE, Cochrane, and Web of Science databases from January 1, 2006, to March 18, 2017.

Study Selection  Prospective studies measuring high-sensitivity cardiac troponin I concentrations in patients with suspected acute coronary syndrome in which the diagnosis was adjudicated according to the universal definition of myocardial infarction.

Data Extraction and Synthesis  The systematic review identified 19 cohorts. Individual patient-level data were obtained from the corresponding authors of 17 cohorts, with aggregate data from 2 cohorts. Meta-estimates for primary and secondary outcomes were derived using a binomial-normal random-effects model.

Main Outcomes and Measures  The primary outcome was myocardial infarction or cardiac death at 30 days. Performance was evaluated in subgroups and across a range of troponin concentrations (2-16 ng/L) using individual patient data.

Results  Of 11 845 articles identified, 104 underwent full-text review, and 19 cohorts from 9 countries were included. Among 22 457 patients included in the meta-analysis (mean age, 62 [SD, 15.5] years; n = 9329 women [41.5%]), the primary outcome occurred in 2786 (12.4%). Cardiac troponin I concentrations were less than 5 ng/L at presentation in 11 012 patients (49%), in whom there were 60 missed index or 30-day events (59 index myocardial infarctions, 1 myocardial infarction at 30 days, and no cardiac deaths at 30 days). This resulted in a negative predictive value of 99.5% (95% CI, 99.3%-99.6%) for the primary outcome. There were no cardiac deaths at 30 days and 7 (0.1%) at 1 year, with a negative predictive value of 99.9% (95% CI, 99.7%-99.9%) for cardiac death.

Conclusions and Relevance  Among patients with suspected acute coronary syndrome, a high-sensitivity cardiac troponin I concentration of less than 5 ng/L identified those at low risk of myocardial infarction or cardiac death within 30 days. Further research is needed to understand the clinical utility and cost-effectiveness of this approach to risk stratification.

Chest pain is one of the most common reasons for presentation to hospitals worldwide.¹ Despite the majority of patients not having myocardial infarction,² hospital admission for observation and serial cardiac troponin testing is required in many patients to identify those with and without myocardial infarction.³ Novel strategies to identify low-risk patients at presentation have been proposed to reduce hospital admissions, serial testing, and resource utilization as well as to improve care for patients.^4,5

High-sensitivity assays are able to quantify cardiac troponin at low concentrations and provide an opportunity to rule out myocardial infarction at an earlier stage. In a prospective study of consecutive patients with suspected acute coronary syndrome, a risk stratification threshold was defined using a high-sensitivity cardiac troponin I assay. In 4870 patients, a threshold of less than 5 ng/L had a negative predictive value (NPV) of 99.6%, misclassifying less than 1 myocardial infarction for every 200 patients tested.⁶ This threshold identified more than half of all patients with suspected acute coronary syndrome as low risk, reducing the proportion of patients who require admission for serial testing.⁷

Recent studies have questioned whether 5 ng/L is the optimal threshold to risk-stratify patients and have proposed alternative thresholds that may miss fewer patients with myocardial infarction.^8-11 To investigate these concerns, a systematic review of all studies of high-sensitivity cardiac troponin I testing in patients with suspected acute coronary syndrome was undertaken and individual patient-level data were obtained. Across multiple cohorts with varying prevalence of myocardial infarction, the aim was to evaluate the performance of this threshold, to evaluate other risk stratification thresholds, and to determine the association with other clinical risk characteristics.

A systematic search of the MEDLINE, EMBASE, Cochrane, and Web of Science databases was performed without language restriction from January 1, 2006, to March 18, 2017, using detailed search terms for chest pain, acute coronary syndrome, acute myocardial infarction, troponin, high sensitive/sensitivity, and emergency department (Figure 1; eAppendix 1 in the Supplement contains the full search strategy). Studies were included if they met the following prespecified eligibility criteria: (1) were prospective studies of patients investigated in the emergency department for suspected acute coronary syndrome; (2) measured cardiac troponin using the Abbott ARCHITECT_STAT high-sensitive cardiac troponin I assay (Abbott Laboratories) at presentation; and (3) had an adjudicated end point of myocardial infarction on index hospitalization (eAppendixes 2 and 3 in the Supplement). All findings are reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-analysis of Individual Participant Data (PRISMA-IPD).¹²

Two investigators (A.R.C. and K.K.L.) performed the initial screening of titles and abstracts. Full-text reports of potentially relevant articles were obtained and assessed by both investigators using a prespecified protocol (PROSPERO register CRD42017059128). A third investigator (A.S.V.S.) adjudicated all disagreements. When there were multiple articles describing the same cohort, the article that included the largest number of participants was included. The corresponding authors of each eligible cohort were contacted with a request for anonymized data including cardiac troponin concentrations, adjudicated diagnosis, outcomes, and prespecified covariates (age, sex, chest pain, time from symptom onset to presentation sample, myocardial ischemia on electrocardiogram, cigarette smoking, diabetes mellitus, hypertension, hyperlipidemia, known angina, previous myocardial infarction, percutaneous coronary intervention, coronary artery bypass graft surgery, and stroke). All studies were prospective and conducted in accordance with the Declaration of Helsinki with approval from the regional ethics committee or institutional review board, and written consent was obtained where required. This approval permitted each contributor to share individual-level data or aggregate data for inclusion in this meta-analysis. Bias was assessed by 2 investigators independently, with consensus from a third, using the Quality Assessment of Diagnostic Accuracy Studies version 2 (QUADAS-2) framework (eAppendix 4 in the Supplement).

The analysis population comprised patients with cardiac troponin concentrations at or below the 99th percentile at presentation (those above the 99th percentile have evidence of myocardial injury and are not eligible for risk stratification at presentation). Patients with ST-segment elevation myocardial infarction and those who presented in cardiac arrest were excluded from this analysis. The prespecified primary outcome was a composite of type 1 myocardial infarction or cardiac death at 30 days. The prespecified secondary outcomes were recurrent myocardial infarction and cardiac death at 1 year. In addition, we evaluated the performance of cardiac troponin thresholds for the diagnosis of type 1 or type 2 myocardial infarction on index presentation. The number of patients available for each analysis is shown in Figure 1.

Baseline characteristics are summarized as mean (standard deviation) or median (interquartile range) as appropriate. The primary outcome measure was the NPV of a high-sensitivity cardiac troponin I concentration of less than 5 ng/L at presentation. All cardiac troponin concentrations were rounded to integer values in line with clinical standards for reporting. When individual patient-level data were available, this was checked for consistency and completeness, and cohort-level summary counts of patients with and without the primary outcome were derived for a high-sensitivity cardiac troponin I concentration of less than 5 ng/L at presentation. In cohorts in which raw data were not available, the corresponding authors were asked to provide these summaries. The NPV was calculated at a cohort level using a Bayesian approach, with a binomial likelihood and beta prior (a noninformative Jeffreys prior with both shape parameters equal to 0.5), as this produces confidence intervals with better coverage when proportions are close to 0 or 1.¹³ Heterogeneity is reported using the I² statistic.¹⁴ Survival free from cardiac death at 30 days and at 1 year is reported for patients with cardiac troponin I concentrations of less than 5 ng/L, 5 ng/L to the 99th percentile, and greater than the 99th percentile at presentation.

For the primary outcome, the NPV was evaluated in prespecified subgroups stratified by age (≤65 or >65 years), sex, history of ischemic heart disease, time since symptom onset (≤2 or >2 hours), and presence of myocardial ischemia on electrocardiogram. Most cohorts defined myocardial ischemia as at least 2-mm ST-segment depression in 2 consecutive leads or new T-wave inversion. To explore the clinical implications of differences in performance between subgroups, we undertook these subgroup analyses in patients without myocardial ischemia on electrocardiogram. Studies have demonstrated imperfect calibration between high-sensitivity cardiac troponin I and T assays, with up to 17.5% of patients greater than the 99th percentile on the T assay shown to be less than the 99th percentile on the I assay.¹⁵ Therefore, a further analysis evaluated whether the assay used to adjudicate the index diagnosis affected the performance of the risk stratification threshold. In addition, we determined whether the assessed risk of bias and site of patient recruitment affected the NPV.

Meta-estimates of the NPV were derived in the analysis population for all primary and secondary outcomes by modeling cohort-level proportions (true negative/[true negative + false negative]) in a binomial-normal random-effects model, with an additional term when cohort-level characteristics (adjudication assay, assessment of bias or location of recruitment) were compared. We estimated odds ratios for the difference in NPV between prespecified subgroups, meta-analyzing this across cohorts to obtain the mean odds ratio and a P value for the null hypothesis of no association. For cohorts in which individual patient-level data were available, the cardiac troponin threshold that would identify the highest proportion of patients as at low risk for an NPV at or above 99.5% was determined. For this analysis, we prespecified an NPV of 99.5% as being clinically acceptable and equivalent to a miss rate of 5 per 1000 low-risk patients.¹⁶ To evaluate how the inclusion of a risk stratification threshold would affect the overall diagnosis in all patients with suspected acute coronary syndrome, meta-estimates of NPV, positive predictive value (PPV), and sensitivity were derived for risk stratification thresholds alone (2-16 ng/L) and in conjunction with a nonischemic electrocardiogram result at presentation. At each threshold, the proportion of the total population classified as low risk and the miss rate per 1000 patients was reported. All analyses were performed in R version 3.2.2, with the meta-analyses performed using the metafor package.¹⁷ The analysis code is available online (eAppendix 5 in the Supplement).

The initial search identified 11 845 articles, of which 104 articles underwent full-text review. A total of 36 articles met inclusion criteria, reporting observations from 19 individual cohorts across 9 different countries (Figure 1). Five articles reported outcomes for a high-sensitivity cardiac troponin I concentration of less than 5 ng/L.^6,8-10,18

All corresponding authors from the 19 individual cohorts identified in the systematic review agreed to provide data for the meta-analysis. Individual patient-level data were obtained from 17 cohorts^6-9,11,18-29 and aggregate data from 2 cohorts,^10,30 for a total study population of 22 457 patients with suspected acute coronary syndrome (mean age, 62 [SD, 16] years; 41.5% women) (Table 1, Table 2, and Table 3). In 11 cohorts, data were available for the prespecified primary outcome of type 1 myocardial infarction or cardiac death at 30 days (Table 4). In the remainder, the outcome was index type 1 myocardial infarction (n = 1) or non–ST-segment elevation myocardial infarction on index presentation (n = 5) or at 30 days (n = 2). The assessed risk of bias was high in 11 cohorts because of patient selection or use of a contemporary reference standard (eAppendixes 3 and 4 in the Supplement). Across all cohorts, the proportion with the primary outcome was 12.4% (range, 2.4%-24.0%). The analysis population comprised 18 248 of 22 457 patients in which high-sensitivity cardiac troponin I concentrations were below the 99th percentile at presentation, and the prevalence of the primary outcome was 3.5% (range, 0.6%-6.1%).

High-sensitivity cardiac troponin I concentrations were less than 5 ng/L at presentation in 11 012 patients (49%), with an NPV of 99.5% (95% CI, 99.3%-99.6%) (Figure 2 and Table 2) for the primary outcome and a total of 60 missed index or 30-day events (59 index myocardial infarctions, 1 myocardial infarction at 30 days, and no cardiac deaths at 30 days) (eTable 1 in the Supplement). The NPV was similar across cohorts with varying prevalence of myocardial infarction. The estimate of heterogeneity (I²) was 31.9%. Cohort-level 2×2 summary tables are provided for the analysis population in eTable 2 in the Supplement. When data were available in the analysis population (n = 16 537 [90.6%]), we estimated the NPV for the secondary outcome of index non–ST-segment elevation myocardial infarction (type 1 or type 2 myocardial infarction). Cardiac troponin I concentrations were less than 5 ng/L at presentation in 9574 patients (48%), with an NPV of 99.4% (95% CI, 99.2%-99.6%) and a total of 58 missed events.

Meta-estimates of NPV were obtained in a number of prespecified subgroups (Figure 3). The NPV was lower in those with (98.2%; 95% CI, 96.4%-99.1% [n = 2178]) compared with those without (99.7%; 95% CI, 99.4%-99.8% [n = 13 709]) myocardial ischemia on electrocardiogram (P < .001) and in those who presented within 2 hours of symptom onset (99.0% [95% CI, 97.7%-99.5%] [n = 2303] vs 99.6% [95% CI, 99.4%-99.8%] [n = 11 101]; P = .003). Differences in the NPV were also observed between patients older than 65 years (NPV, 99.1%; 95% CI, 98.5%-99.5% [n = 6818]) compared with those aged 65 years or younger (99.6%; 95% CI, 99.4%-99.8% [n = 11 430]; P = .02) and in those with (NPV, 98.8%; 95% CI, 98.1%-99.3% [n = 3990]) compared with those without (NPV, 99.6%; 95% CI, 99.4%-99.7% [n = 10 170]) a history of ischemic heart disease (P = .03). When this analysis was restricted to patients without myocardial ischemia on electrocardiogram, estimates of NPV were higher than 99% for all subgroups (eFigure 1 in the Supplement). Performance of the risk stratification threshold was similar regardless of the assay used for adjudication (high-sensitivity cardiac troponin I: NPV, 99.6% [95% CI, 99.3%-99.7%] [n = 7046]; contemporary cardiac troponin I or T: NPV, 99.6% [95% CI, 99.3%-99.7%] [n = 5907]; high-sensitivity cardiac troponin T: NPV, 99.2% [95% CI, 98.6%-99.6%] [n = 5295]; P = .27), the assessed risk of bias (high risk of bias: NPV, 99.5% [95% CI, 99.1%-99.7%] [n = 7043]; low risk of bias: NPV, 99.3% [95% CI, 99.3%-99.6%] [n = 11 205]; P = .37), and the site of patient recruitment (Europe: NPV, 99.5% [95% CI, 99.1%-99.7%] [n = 11 714]; North America: NPV, 99.5% [95% CI, 99.0%-99.8%] [n = 2999]; Asia-Pacific: NPV, 99.5% [95% CI, 99.1%-99.7%] [n = 3535]; P = .30) (eFigure 2 in the Supplement).

Follow-up data for cardiac death at 30 days and at 1 year were available in 12 953 patients (57.7%) and 9271 patients (41.3%), respectively (eTables 3 and 4 in the Supplement). In patients with cardiac troponin concentrations of less than 5 ng/L at presentation (n = 6956), there were no cardiac deaths at 30 days (NPV, 100% [95% CI, 99.9%-100%]; sensitivity, 99.4% [95% CI, 97.7%-100%]) and 7 cardiac deaths (0.1%) at 1 year (NPV, 99.9% [95% CI, 99.7%-99.9%]; sensitivity, 96.1% [95% CI, 92.9%-98.3%]). In patients with cardiac troponin concentrations between 5 ng/L and the 99th percentile at presentation (n = 3817), there were 19 cardiac deaths at 30 days (0.5%) and 58 (2.1%) at 1 year. In comparison, in those with troponin concentrations above the 99th percentile (n = 2180), there were 62 cardiac deaths at 30 days (2.8%) and 125 (8.2%) at 1 year. In patients with troponin concentrations of less than 5 ng/L at presentation and an index or 30-day myocardial infarction, there were no cardiac deaths at 30 days or at 1 year. Because the majority of studies did not adjudicate recurrent myocardial infarction events at 1 year, we were not able to conduct this prespecified analysis.

In all patients with suspected acute coronary syndrome for whom individual patient-level data were available (n = 18 601 [82.8%]), we evaluated how different risk stratification thresholds would affect the NPV and sensitivity for the primary outcome. When used in isolation, a troponin I concentration of less than 5 ng/L gave an NPV of 99.5% (95% CI, 99.3%-99.7%) and a sensitivity of 98.0% (95% CI, 96.4%-98.9%), identifying 49.1% of patients as low risk with a miss rate of 5.4 (95% CI, 4.0-7.0) per 1000 patients. At a threshold of less than 2 ng/L, the NPV was 99.8% (95% CI, 99.0%-100%) and the sensitivity was 100% (95% CI, 98.9%-100%), but the proportion of patients identified as low risk was lower at 13.7%. Although the absolute number of missed cases was lower, the miss rate was similar at 4.1 (95% CI, 2.0-6.9) per 1000 patients (eTable 5 in the Supplement).

In a subgroup analysis combining risk stratification thresholds and a nonischemic electrocardiogram result (Figure 4), a cardiac troponin I concentration of less than 5 ng/L gave an NPV of 99.7% (95% CI, 99.4%-99.8%) and a sensitivity of 99.0% (95% CI, 97.3%-99.6%), identifying 45.9% of patients as low risk, with 4.4 (95% CI, 3.0-6.0) false negatives per 1000 patients and a PPV of 24.5% (95% CI, 20.3%-29.2%). The combination of a cardiac troponin I concentration of less than 2 ng/L and a nonischemic electrocardiogram result gave a similar NPV of 99.9% (95% CI, 98.5%-100%) and a sensitivity of 100% (95% CI, 96.6%-100%) but identified just 13.1% of patients as low risk, with 4.1 (95% CI, 1.8-7.3) false negatives per 1000 patients and a lower PPV of 14.2% (95% CI, 11.3%-17.6%).

In 19 cohorts across 9 countries and encompassing more than 22 000 patients, a cardiac troponin I concentration of less than 5 ng/L at presentation identified half of all patients with suspected acute coronary syndrome as at low risk of myocardial infarction or cardiac death at 30 days, with 5 false negatives per 1000 patients tested.

There are a number of strengths to the analysis. This was a prespecified systematic review and meta-analysis that included individual patient-level data from all cohorts identified. The findings were consistent across a range of health care settings and geographic regions with considerable differences in the prevalence of myocardial infarction. Individual patient-level data were included from more than 22 000 patients, allowing a meaningful analysis of important subgroups. All studies were prospective, and in all studies the final diagnosis was adjudicated according to the universal definition of myocardial infarction.

Two recent meta-analyses have suggested an approach to risk stratification using the limit of detection of the high-sensitivity cardiac troponin T assay, which identifies up to 31% of patients with an NPV of 99.3%.^31,32 The limit of detection of the high-sensitivity cardiac troponin I assay identifies 19% to 27% of patients as low risk with an NPV of 99.5% or greater.^8-10 The major limitation of this approach for both assays is analytical, with biases and analytical variation at the limit of detection associated with rates of misclassification that are twice that observed at 5 ng/L.^33,34

In clinical practice, cardiac troponin concentrations are interpreted in conjunction with the electrocardiogram and clinical assessment. When a risk stratification threshold of less than 5 ng/L was evaluated in the subgroup of patients without myocardial ischemia on electrocardiogram, the NPV and sensitivity were excellent. To ensure that safety estimates were conservative, performance was evaluated not just for an index diagnosis but for a composite end point that included events up to 30 days. Although there were 81 cardiac deaths at 30 days, none occurred in the 6956 patients with cardiac troponin I concentrations less than 5 ng/L. Furthermore, performance was similar for both spontaneous type 1 and secondary type 2 myocardial infarction. This is relevant because the diagnosis of type 2 myocardial infarction is more challenging and is associated with a worse prognosis.^35-37

At a threshold of 5 ng/L, the analytical performance of the high-sensitivity cardiac troponin I assay is excellent.^6,38 The use of lower thresholds did not improve diagnostic accuracy. A miss rate of 5 per 1000 patients was observed when applying less than 5 ng/L as the risk stratification threshold, with a miss rate of 4 per 1000 patients observed at a threshold of less than 2 ng/L. Although the true risk of missing an individual patient with myocardial infarction is the same at both thresholds, lower thresholds reduce the proportion of patients classified as at low risk; only 1 in 10 patients had a troponin I concentration of less than 2 ng/L compared with 5 in 10 patients a with concentration of less than 5 ng/L. Use of lower thresholds would result in more patients without myocardial infarction being admitted for serial testing and further investigation, with an increase in health care expenditures.

Despite recent changes to guidelines,⁵ the majority of clinicians continue to rely on the 99th percentile to rule in and rule out myocardial infarction.³⁹ A pathway incorporating a risk stratification threshold of less than 5 ng/L alongside nonischemic electrocardiogram findings misses 5-fold fewer index myocardial infarctions or 30-day events than guideline-approved pathways based exclusively on the 99th percentile.⁷ This limitation of the 99th percentile has now been demonstrated in multiple studies.^25,40 This approach to risk stratification using low high-sensitivity cardiac troponin concentrations has major potential to improve both the efficiency of health care delivery and patient safety and is being formally evaluated in a prospective multicenter clinical trial (NCT03005158).

This study has several limitations. First, not all cohorts used identical protocols, with differences both in the inclusion criteria and the diagnostic criteria used for adjudication (eAppendix 3 in the Supplement). However, no significant differences in NPV were observed when stratified by adjudicating assay, and the NPV was high across individual cohorts, suggesting that these findings are generalizable. Second, the percentage of patients who presented early after onset of symptoms was low at just 10% of the study population. Despite observing an NPV of 99% in this subgroup, inconsistencies in the documentation of symptom onset across cohorts may affect the analysis, and until further research is available, serial testing is recommended in patients presenting within 2 hours of symptom onset.⁵ The greatest number of false negatives was observed in the cohort with the shortest median symptom onset to sample time (179 [interquartile range, 119-349] minutes), which may explain the lower NPV and sensitivity reported at this threshold in a previous study.⁸ Third, while it is reassuring that patients with troponin I concentrations of less than 5 ng/L had a much lower rate of cardiac death at 1 year than did patients with concentrations between 5 ng/L and the 99th percentile, this observation needs to be verified in prospective studies in which patient care is guided by this approach.

Among patients with suspected acute coronary syndrome, a high-sensitivity cardiac troponin I concentration of less than 5 ng/L at presentation identified those at low risk of myocardial infarction or cardiac death within 30 days. Further research is needed to understand the clinical utility and cost-effectiveness of this approach to risk stratification.

Corresponding Author: Nicholas L. Mills, MD, BHF/University Centre for Cardiovascular Science, University of Edinburgh, Edinburgh EH16 4SA, Scotland (nick.mills@ed.ac.uk).

Correction: This article was corrected on March 20, 2018, for an author name misspelling.

Accepted for Publication: October 22, 2017.

Published Online: November 11, 2017. doi:10.1001/jama.2017.17488

Author Contributions: Drs Chapman and Mills had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Chapman, Lee, Shah, and Mills contributed equally.

Concept and design: Chapman, Lee, McAllister, Shah, Mills.

Acquisition, analysis, or interpretation of data: Chapman, Lee, McAllister, Cullen, Greenslade, Parsonage, Worster, Kavsak, Blankenberg, Neumann, Sörensen, Westermann, Buijs, Verdel, Pickering, Than, Twerenbold, Badertscher, Sabti, Mueller, Anand, Adamson, Ferry, Sandeman, Gray, Body, Keevil, Carlton, Greaves, Korley, Metkus, Sandoval, Apple, Newby, Shah, Mills.

Drafting of the manuscript: Chapman, Lee, Shah, Mills.

Critical revision of the manuscript for important intellectual content: Chapman, Lee, McAllister, Cullen, Greenslade, Parsonage, Worster, Kavsak, Blankenberg, Neumann, Sörensen, Westermann, Buijs, Verdel, Pickering, Than, Twerenbold, Badertscher, Sabti, Mueller, Anand, Adamson, Strachan, Sandeman, Body, Keevil, Carlton, Greaves, Korley, Metkus, Sandoval, Apple, Newby, Shah, Mills.

Statistical analysis: Chapman, Lee, McAllister, Shah.

Obtained funding: Cullen, Worster, Twerenbold, Mueller, Body, Apple, Mills.

Administrative, technical, or material support: Chapman, Lee, Greenslade, Parsonage, Worster, Westermann, Verdel, Pickering, Sabti, Mueller, Anand, Strachan, Sandeman, Gray, Keevil, Carlton, Korley, Apple, Newby.

Supervision: McAllister, Cullen, Worster, Kavsak, Blankenberg, Westermann, Than, Badertscher, Mueller, Greaves, Sandoval, Apple, Newby, Shah, Mills.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Chapman has received honoraria from Abbott Diagnostics and AstraZeneca. Dr Cullen has received funding from Abbott Diagnostics, Roche, Alere, Siemens, and Radiometer Pacific for research on diagnostic protocols and from Alere, Boehringer-Ingelheim, Pfizer, AstraZeneca, Abbott Diagnostics, and Radiometer Pacific for speaking and education. Dr Parsonage has received funding from Abbott Diagnostics, Roche, Alere, and Beckmann Coulter for research on diagnostic protocols, honoraria, travel expenses, and consultancy fees from Abbott, AstraZeneca, Hospira, and Sanofi-Aventis; and travel, accommodation, consulting fees, or honoraria from Abbott Laboratories. Dr Metkus performs consulting unrelated to this subject matter for BestDoctors Inc and Oakstone/EBIX and receives royalties for a textbook publication for McGraw-Hill publishing unrelated to this subject matter. Dr Kavsak has received grants, reagents, consultancy, advisory fees, and honoraria from Abbott Laboratories, Abbott Point of Care, Abbott Diagnostics Division Canada, Beckman Coulter, Ortho Clinical Diagnostics, Randox Laboratories, Roche Diagnostics, and Siemens Healthcare Diagnostics. In addition, McMaster University has filed patents with Dr Kavsak listed as an inventor in the acute cardiovascular biomarker field. Dr Anand has received honoraria from Abbott Diagnostics. Dr Carlton has undertaken research under collaborative agreements with Abbott Diagnostics and Randox. Dr Apple has acted as a consultant for Phillips Healthcare Incubator and Metanomics Healthcare, is on the board of directors at HyTest Ltd, has received honoraria from Instrumentation Laboratory and Abbott POC, has been a research principal investigator through the Minneapolis Medical Research Foundation, and has had nonsalaried relationships with Abbott Diagnostics, Roche Diagnostics, Siemens Healthcare, Alere, Ortho-Clinical Diagnostics, Nanomix, Becton Dickinson, and Singulex. Dr Than has accepted travel, accommodation, consulting fees, or honoraria from Abbott Laboratories. Dr Shah has received honoraria from Abbott Diagnostics. Dr Mills has acted as a consultant for Abbott Diagnostics, Beckman-Coulter, Roche, and Singulex. No other disclosures were reported.

Funding/Support: This research was funded by the British Heart Foundation (SP/12/10/29922 and PG/15/51/31596). Drs Chapman and Mills are supported by a Clinical Research Training Fellowship (FS/16/75/32533) and the Butler Senior Clinical Research Fellowship (FS/16/14/32023), respectively, from the British Heart Foundation. Dr McAllister is supported by an Intermediate Clinical Fellowship (and Beit Award; 201492/Z/16/Z) from the Wellcome Trust.

Role of the Funder/Sponsor: The study funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

Meeting Presentation: Presented in part at the American Heart Association Scientific Sessions; November 11, 2017; Anaheim, California.

Additional Contributions: We thank A. Mark Richards, MD, PhD, University of Otago, and Richard W. Troughton, MD, PhD, and Sally Aldous, MD, PhD, Christchurch Hospital, Christchurch, New Zealand, who contributed to the study design, analysis, and interpretation of the ADAPT trial, and all the patients and investigators who contributed to the original cohort studies. No financial compensation was received for named contributors’ roles in the study.