High-Sensitivity Cardiac Troponin I as a Gatekeeper for Coronary Computed Tomography Angiography and Stress Testing in Patients with Acute Chest Pain

Abstract

BACKGROUND:

Most patients presenting to the emergency department (ED) with suspected acute coronary syndrome (ACS) undergo noninvasive cardiac testing with a low diagnostic yield. We determined whether a combination of high-sensitivity cardiac troponin I (hs-cTnI) and cardiovascular risk factors might improve selection of patients for cardiac testing.

METHODS:

We included patients from the Rule Out Myocardial Infarction/Ischemia Using Computer Assisted Tomography (ROMICAT) I and II trials who presented to the ED with acute chest pain and were referred for cardiac testing. Based on serial hs-cTnI measurements and cardiovascular risk factors, we derived and validated the criterion for no need of cardiac testing. We predicted the effect of this criterion on the effectiveness of patient management.

RESULTS:

A combination of baseline hs-cTnI (<4 ng/L) and cardiovascular risk factors (<2) ruled out ACS with a negative predictive value of 100% in ROMICAT I. We validated this criterion in ROMICAT II, identifying 29% patients as not needing cardiac testing. An additional 5% of patients were identified by adding no change or a decrease between baseline and 2 h hs-cTnI as a criterion. Assuming those patients would be discharged from the ED without cardiac testing, implementation of hs-cTnI would increase ED discharge rate (24.3% to 50.2%, P < 0.001) and decrease the length of hospital stay (21.4 to 8.2 h, P < 0.001), radiation dose (10.2 to 7.7 mSv, P < 0.001), and costs of care (4066 to 3342 US$, P < 0.001).

CONCLUSIONS:

We derived and validated a criterion for combined hs-cTnI and cardiovascular risk factors that identified acute chest pain patients with no need for cardiac testing and could improve effectiveness of patient management.

ClinicalTrials.gov Identifiers:

and

The evaluation of acute chest pain patients in the emergency department (ED)¹⁷ remains inefficient, as baseline information (e.g., chest pain characteristics, cardiovascular risk factors, electrocardiogram, conventional cardiac troponin) is often insufficient to safely rule out acute coronary syndrome (ACS), including unstable angina pectoris (1–3). Because in the US follow-up and outpatient testing of ED patients occurs in fewer than half of patients (4), ED providers need information on whether obstructive coronary artery disease (CAD) or myocardial ischemia is present to safely discharge patients. Stress testing or coronary computed tomography angiography (CTA) permits exclusion of obstructive CAD or myocardial ischemia, and their use is supported by guidelines (2, 5). Although this approach has proven to be safe (missed ACS rate <1%), it results in overcrowding of EDs, a low rate of direct ED discharges, prolonged hospital stays, and significant healthcare costs, all at a low diagnostic yield (5, 6). In multicenter studies in the US, noninvasive cardiovascular testing was positive in only 8% to 18% of low-to-intermediate risk patients with acute chest pain (7–10). Given that approximately 6 million patients undergo noninvasive cardiac testing in this setting every year (11, 12), there is an unmet need for a better selection of patients who need testing.

Recent research has established high-sensitivity cardiac troponin (hs-cTn) assays, which are characterized by an increased analytic sensitivity and precision at low concentrations (13–19). Several studies showed that serial measurements of hs-cTn (at the time of presentation and after 1–3 h) using both the absolute values and dynamic changes had excellent diagnostic accuracy with sensitivity of 97% to 100% for myocardial infarction during index hospitalization and major adverse cardiovascular events at 30 days (14, 17, 18, 20–23). hs-cTnI measurement also permitted reclassification of up to one-third of patients previously diagnosed with unstable angina pectoris to a new diagnosis of myocardial infarction (24, 25).

However, another potential use of hs-cTnI, that of improving the selection of patients who need stress testing or coronary CTA to rule out myocardial ischemia or obstructive CAD, has been less researched. Several studies demonstrated that higher concentrations of hs-cTnI were associated with both the presence and extent of coronary atherosclerosis, obstructive CAD, and high-risk plaque morphology, as well as with the presence and extent of reversible myocardial ischemia, suggesting that hs-cTnI could improve selection of patients for testing (24–27).

The aim of our study was to derive and validate a selection algorithm for stress testing and coronary CTA in patients with acute chest pain based on hs-cTnI and cardiovascular risk factors, and to predict downstream effects of such practice on safety and efficiency of treatment for patients with intermediate risk of ACS.

Methods

We studied patients from the Rule Out Myocardial Infarction/Ischemia Using Computer Assisted Tomography (ROMICAT) I trial () to derive a criterion based on the lowest upper threshold for hs-cTnI values at the time of presentation and the number of cardiovascular risk factors that permitted exclusion of ACS, including unstable angina, with a sensitivity and negative predictive value (NPV) of 100%. ROMICAT I enrolled patients with suspected ACS in the ED but who did not have ischemic electrocardiographic changes and whose conventional troponin concentration at the time of presentation was below the 99th percentile value for the apparently healthy population (28). The rationale and inclusion criteria for the ROMICAT I trial were described previously (28). We then validated this threshold in patients from the ROMICAT II trial (Fig. 1) () (10).

Fig. 1.

Study population enrollment, as well as exclusion and inclusion in the ROMICAT II patients.

All study participants provided written informed consent for participation in the ROMICAT I or II trial and for additional blood sampling. The local institutional review boards approved both studies. Patient demographics and cardiovascular risk factors (arterial hypertension, dyslipidemia, diabetes mellitus, smoking, family history of premature CAD), results of diagnostic testing, timeline of hospitalization, other process measures, and outcomes were prospectively collected for both trials (10, 28). An independent clinical events committee using the local conventional troponin assay results adjudicated the presence of ACS during the index hospitalization. ACS was defined as acute myocardial infarction or unstable angina pectoris according to the American College of Cardiology/American Heart Association Guidelines (see Methods in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol63/issue11). Unstable angina was identified based on the presence of obstructive CAD or reversible myocardial ischemia in combination with chest pain characteristics (10, 28).

NONINVASIVE CARDIAC TESTING

Noninvasive cardiac testing included coronary CTA in the ROMICAT I trial and either coronary CTA or functional stress testing (exercise treadmill stress test, stress echocardiography, or nuclear myocardial perfusion imaging) in the ROMICAT II trial. All patients included in our study underwent cardiac testing. Results of testing are based on the reports by the study sites (10, 28). A functional test was defined as positive if there was evidence of stress-induced myocardial ischemia (stress electrocardiogram with ST segment changes diagnostic for ischemia, new stress-induced wall motion abnormalities on stress echocardiography, or reversible defect on myocardial perfusion imaging). A coronary CTA was defined as positive if there was evidence of >50% stenosis in any major epicardial coronary artery.

hs-cTnI MEASUREMENTS

Blood was collected into tubes containing EDTA and immediately processed and frozen at −80 °C. The samples were obtained at the time of ED presentation (in the ROMICAT I and II trials) and at approximately 2 h (range 90–155 min; in the ROMICAT II trial). The blood was analyzed in a blinded fashion using a preclinical highly sensitive method for detecting troponin I (STAT hs-cTnI Abbott ARCHITECT i2000SR, Abbott Laboratories) (19, 29). The limit of detection of the assay ranged from 1.1 to 1.9 ng/L. The 99th percentile value for the apparently healthy population was 26 ng/L (30). The CV was 5.86% in the range of 12 to 28 ng/L. A 10% CV was achieved at 4.7 ng/L.

DERIVATION OF THE CRITERION FOR NO NEED OF NONINVASIVE CARDIAC TESTING IN THE ROMICAT I TRIAL

To determine the optimal criterion for no need for stress testing or coronary CTA, we generated receiver-operating characteristic curves for the diagnosis of ACS during the index hospitalization based on the combination of baseline hs-cTnI value and number of cardiovascular risk factors. The goal was to identify the combination of the lowest upper threshold values for hs-cTnI and the number of cardiovascular risk factors that would permit exclusion of ACS with a sensitivity and NPV of 100% while maintaining the best specificity.

VALIDATION OF THE CRITERION FOR NO NEED OF NONINVASIVE CARDIAC TESTING IN THE ROMICAT II TRIAL

We applied the criterion for no need of noninvasive cardiac testing in the ROMICAT II trial in an advanced diagnostic pathway (Fig. 2). Patients meeting the criterion for no need of noninvasive cardiac testing based on the baseline hs-cTnI could be discharged from the ED without further testing 2 h after the baseline hs-cTnI. Additional patients with hs-cTnI below the 99th percentile for the normal healthy population at baseline and no increase of hs-cTnI at 2 h and <2 cardiovascular risk factors could be discharged from the ED without further testing 4 h after the baseline hs-cTnI (31). The remaining patients could be treated as observed in the ROMICAT II trial. These rules were applied to calculate the predicted discharges from the ED, length of stay, time to diagnosis, and need for cardiac testing based on the decision rules. The predicted radiation dose was calculated based on the decrease in testing.

Fig. 2. The implementation of the criterion of no need for noninvasive cardiac testing in an advanced diagnostic pathway.

The advanced diagnostic pathway using high-sensitivity cardiac troponin I guided decision rules and the predicted care in the ROMICAT II trial.

STUDY END POINTS

The primary end point was the proportion of patients who were identified as not needing stress testing or coronary CTA while maintaining 100% specificity and NPV for ACS. Secondary end points included efficiency, safety, and costs.

Efficiency.

We calculated predicted discharges from the ED (defined as the proportion of patients discharged from the ED without admission to an observation unit or hospital), length of stay (defined as the time from presentation in the ED to the time of the discharge), time to diagnosis (defined as the time from the ED presentation until the first positive diagnostic ACS test, or as the time from ED presentation until the final test result that was used to rule out an ACS), and need for noninvasive cardiac testing (defined as proportion of patients who underwent noninvasive cardiac testing). The predicted results were compared with the observed care in the ROMICAT II trial.

Safety.

We calculated the estimated cumulative radiation doses (defined as radiation exposure from testing, including coronary CTA, nuclear perfusion imaging, and invasive coronary angiography), measured in millisieverts (mSv), with the use of standard methods (32).

Healthcare costs.

Healthcare costs during the index care episode were assessed using reports from hospital cost-accounting systems and physician billing records. Cost data were available for 649 of 1000 ROMICAT II patients. Therefore, costs were calculated using regression coefficients 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 the dependent variable and length of stay of >8 h (binary), number of noninvasive cardiac tests, invasive coronary angiography (binary), inpatient hospitalization (binary), percutaneous coronary intervention (binary), and coronary artery bypass graft surgery (binary) as independent variables. The total health-care costs were then calculated by multiplying the predicted cost coefficients with their corresponding parameters for each patient included in our study, i.e., original input parameters for as observed and modified parameters for as predicted analysis.

STATISTICAL METHODS

Statistical analyses were performed using Stata 13.1 (StataCorp LP). Continuous data are presented as mean ± SD or median and interquartile range (25th to 75th percentile). 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 the Wilcoxon rank-sum test for continuous variables, the Fisher exact test for categorical variables, and the Wilcoxon rank-sum test for ordinal variables. For paired data, a paired t-test, Wilcoxon matched-pairs signed rank test, or the exact McNemar test was used. Binominal 95% CIs were calculated using the “exact” method, i.e., Clopper–Pearson intervals (33). For all analyses, a 2-tailed P value <0.05 was required to reject the null hypothesis.

Results

DERIVATION COHORT

hs-cTnI measurement was available for 353 of 368 patients in the ROMICAT I trial. The baseline characteristics of the patients are summarized in Table 1 of the online Data Supplement. Both ROMICAT I and II populations had similar baseline characteristics. ACS during the index hospitalization was diagnosed in 30 (8.5%; 95% CI, 5.8%–11.9%) patients (myocardial infarction, n = 8; unstable angina pectoris, n = 22). Six of the 8 patients with myocardial infarction underwent revascularization (6 percutaneous coronary intervention). Twelve of the 22 patients with unstable angina pectoris underwent revascularization (10 percutaneous coronary intervention, 2 coronary artery bypass grafting). The diagnostic accuracy of hs-cTnI for ACS is summarized in Table 2 of the online Data Supplement.

VALIDATION COHORT

Serial hs-cTnI measurements were available for 255 of 1000 (25.5%) patients in the ROMICAT II trial (Fig. 1). The baseline characteristics are summarized here in Table 1 and also in Table 1 of the online Data Supplement. ACS during the index hospitalization was diagnosed in 20 (7.8%; 95% CI, 4.9%–11.9%) patients (myocardial infarction, n = 5; unstable angina pectoris, n = 15). Four of the 5 patients with myocardial infarction under-went revascularization (3 percutaneous coronary intervention, 1 coronary artery bypass grafting). Eleven of the 15 patients with unstable angina pectoris underwent re-vascularization (11 percutaneous coronary intervention). The diagnostic accuracy of hs-cTnI for ACS is summarized in Table 2 of the online Data Supplement. Patients diagnosed with ACS were older, more often men, and had more cardiovascular risk factors. Baseline demographics, cardiovascular risk factors, medication, presenting symptom, testing results, and ACS prevalence of this subgroup were similar when compared with ROMICAT II patients without available hs-cTnI measurements (see Table 3 in the online Data Supplement).

Table 1.

Baseline characteristics of ROMICAT II patients included in the study stratified by acute coronary syndrome during the index hospitalization.

	No ACS (n = 235)	ACS (n = 20)	P
Age, years	52.8 ± 7.7	55.5 ± 9.1	0.209
Female gender (%)	105 (44.7)	1 (5.0)	<0.001
Cardiovascular risk factors (%)
Hypertension	117 (49.8)	9 (45.0)	0.817
Diabetes mellitus	37 (15.7)	1 (5.0)	0.326
Dyslipidemia	94 (40.0)	12 (60.0)	0.099
Former or current smoker	108 (46.0)	15 (75.0)	0.018
Family history of premature CAD	85 (36.2)	11 (55.0)	0.147
Number of cardiovascular risk factors (%)			0.037
0 or 1	94 (40.0)	4 (20.0)
2 or 3	121 (51.5)	12 (60.0)
≥4	20 (8.5)	4 (20.0)
Relevant prior medications (%)
Aspirin	52 (22.1)	6 (30.0)	0.412
Beta-blocker	37 (15.7)	5 (25.0)	0.341
Statin	61 (26.0)	6 (30.0)	0.791
Chief complaint (%)			0.802
Radiating or nonradiating chest pain or angina equivalent	196 (83.4)	19 (95.0)
Arm, jaw, shoulder, or epigastric pain	17 (7.2)	1 (5.0)
Shortness of breath	6 (2.6)	0 (0.0)
Other	16 (6.8)	0 (0.0)
Heart rate, beats/min	76.8 ± 13.4	75.4 ± 15.2	0.677
Blood pressure, mmHg
Systolic	144.2 ± 23.1	151.1 ± 22.0	0.193
Diastolic	83.1 ± 13.0	85.2 ± 13.2	0.514
Body mass index, kg/m2	28.9 ± 4.8	28.1 ± 3.3	0.276
Acute coronary syndrome (%)	0 (0.0)	20 (100.0)
Unstable angina pectoris	0 (0.0)	15 (75.0)
Myocardial infarction	0 (0.0)	5 (25.0)

DERIVATION OF THE CRITERION FOR NO NEED OF NONINVASIVE CARDIAC TESTING IN THE ROMICAT I TRIAL

Receiver-operating characteristic curves of the combination of baseline hs-cTnI value and number of cardiovascular risk factors for the diagnosis of ACS during the index hospitalization are summarized in Fig. 1 of the online Data Supplement. Each curve represents the diagnostic performance (sensitivity and 1 – specificity) at different levels of hs-cTnI combined with <1 to <6 cardiovascular risk factors. The combination of hs-cTnI threshold of <4 ng/L and <2 cardiovascular risk factors, which was found for 120 (34%) patients, permitted the exclusion of ACS during the index hospitalization with a sensitivity of 100% (95% CI, 88.4%–100%) and NPV of 100% (95% CI, 97.0%–100%) and best specificity of 43% (95% CI, 37.9%–48.7%).

VALIDATION OF THE CRITERION FOR NO NEED OF NONINVASIVE CARDIAC TESTING IN THE ROMICAT II TRIAL

Overall, 74 of 255 patients (29%) fulfilled the criterion of baseline hs-cTnI <4 ng/L and <2 cardiovascular risk factors for no need of noninvasive cardiac testing (Fig. 3). An additional 12 (5%) patients had either no change or a decrease of the second hs-cTnI at 2 h with baseline hs-cTnI below the 99th percentile and <2 cardiovascular risk factors. We used this as an additional criterion for no need of noninvasive cardiac testing. Hence, overall, 86 of 255 patients (34%) fulfilled the no-test criterion, and none of these patients had ACS during the index hospitalization with a sensitivity of 100% (95% CI, 83.2%–100.0%) and NPV of 100% (95% CI, 95.8%–100.0%).

Fig. 3.

The triage of patients with suspected acute coronary syndrome based on the criterion of no need for noninvasive cardiac testing using the results of high-sensitivity cardiac troponin I and cardiovascular risk factors in the ROMICAT II trial.

PREDICTED EFFECTS OF IMPLEMENTATION OF THE CRITERION FOR NO NEED OF NONINVASIVE CARDIAC TESTING ON THE EFFICIENCY, SAFETY, AND COSTS OF PATIENT MANAGEMENT IN THE ROMICAT II TRIAL

Need for noninvasive cardiac testing.

Assuming that all 86 patients fulfilling the criterion of no need for noninvasive cardiac testing would not undergo stress testing or coronary CTA, the need for cardiac testing would decrease by 33.7% (95% CI, 27.9%–39.9%) as compared with observed care in the ROMICAT II trial (P < 0.001; Table 2).

Table 2.

Cardiac testing, radiation exposure, disposition, length of hospital stay, time to diagnosis, and healthcare costs as predicted using decision rules based on hs-cTnI and cardiovascular risk factors vs as observed in the ROMICAT II trial.

	ROMICAT II as observed (n = 255)	ROMICAT II as predicted (n = 255)	P
Cardiac testing (%)a			<0.001
No testing	0 (0.0)	86 (33.7)
1 test	203 (79.6)	127 (49.8)
≥2 tests	52 (20.4)	42 (16.5)
Invasive coronary angiography (%)	23 (9.0)	22 (8.6)	1.000
Intervention (%)
PCIb	14 (5.5)	14 (5.5)	1.000
CABG	1 (0.4)	1 (0.4)	1.000
Cumulative radiation exposure, mSv/patientc	10.2 ± 11.9	7.7 ± 12.4	<0.001
Disposition			<0.001
ED discharge	62 (24.3%, 95% CI, 19.1–30.1%)	128 (50.2%, 95% CI, 43.9–56.5%)
Observational unit admission	150 (58.8%, 95% CI, 52.5–64.9%)	92 (36.1, 95% CI, 30.2–42.3%)
Hospital admission	41 (16.1%, 95% CI, 11.8–21.2%)	34 (13.3%, 95% CI, 9.4–18.1%)
Left against medical advice	2 (0.8%, 95% CI, 0.1–2.8%)	1 (0.4%, 95% CI, 0.0–2.1%)
Length of hospital stay, hours			<0.001
Median (25th-75th percentile)	21.4 (8.0–28.6)	8.2 (2.0–27.1)
Time to diagnosis, hours			<0.001
Median (25th-75th percentile)	8.2 (5.8–22.7)	5.7 (1.0–20.4)
Healthcare cost per patient in US$
Median (25th-75th percentile)	2698 (2698–2698)d	2698 (655–2698)d	<0.001
Mean ± SD	4066 ± 4858	3342 ± 5089
Non-ACS patients only
Median (25th-75th percentile)	2698 (1836–2698)	1837 (655–2698)	<0.001
Mean ± SD	2922 ±1425	2136 ±1634

^aCardiac testing included coronary computed tomography angiography, exercise treadmill test, stress echocardiography, nuclear myocardial perfusion imaging, and invasive coronary angiography.

^bCABG, coronary artery bypass grafting; PCI, percutaneous coronary intervention.

^cRadiation exposure included exposure from coronary computed tomography, nuclear myocardial perfusion imaging, and invasive coronary angiography.

^dBecause all changes in costs occurred in patients below the median of costs, observed and predicted median costs values are identical.

Patient disposition.

Assuming all patients fulfilling the criterion for no need of noninvasive cardiac testing would be directly discharged from the ED, the proportion of ED discharges would increase significantly from 24.3% (95% CI, 19.2%–30.1%) to 50.2% (95% CI, 43.9%–56.5%; P < 0.001; Table 2). Simultaneously, the proportion of patients admitted to the observation unit would be reduced from 58.8% (95% CI, 52.5%–64.9%) to 36.1% (95% CI, 30.2%–42.3%; P < 0.001; Table 2).

Length of stay and time to diagnosis.

As the result of the increase in ED discharges, the time to diagnosis or exclusion of ACS would decrease from 8.2 to 5.7 h (P < 0.001; Table 2), and the length of stay would be reduced by >50% from a median of 21.4 h to a median of 8.2 h (P < 0.001; Table 2 and see also Fig. 2 in the online Data Supplement).

Cumulative estimated radiation dose.

The decrease in the need for cardiac testing (coronary CTA, n = 43; myocardial perfusion imaging, n = 17; invasive coronary angiography, n = 1) could significantly decrease the mean cumulative radiation dose from 10.2 mSv to 7.7 mSv (P < 0.001).

Healthcare costs.

As the result of lower need for noninvasive diagnostic testing and the subsequent favorable changes in management, healthcare costs could decrease significantly in the overall cohort and in patients without ACS (Table 2). Because all changes in costs occurred in patients below the median of costs, observed and predicted median costs values are identical. However, both the mean values and the distribution of costs changed significantly.

Discussion

Using data from 2 prospective clinical trials in patients presenting to the ED with acute chest pain, we derived and validated the criterion for no need of noninvasive cardiac testing, defined as baseline hs-cTnI <4 ng/L and <2 cardiovascular risk factors, which could safely obviate the need for testing in 34% of patients for whom ED physicians decided to rule out the presence of obstructive CAD or myocardial ischemia. Our results further suggest that implementation of hs-cTnI could significantly improve efficiency by reducing number of hospital admissions and length of hospital stay, with lower radiation doses and healthcare costs.

This study expands our previous work, which showed that concentrations below the limit of detection of an hs-cTnI assay not only excluded ACS (including unstable angina pectoris) with 100% sensitivity and NPV, but also excluded significant coronary stenosis in those undergoing coronary CTA (24); higher hs-cTnI values were associated with presence and extent of CAD and myocardial perfusion deficits (25, 26). We derived and validated a criterion that combines hs-cTnI with cardiovascular risk factors to establish a threshold for no need of noninvasive cardiac testing. The combination of hs-cTnI with risk factors is important, as the main question in these patients who do not have myocardial infarction at presentation is not only to rule out ACS but also to rule out obstructive CAD and myocardial ischemia, both of which are closely associated with underlying cardiovascular risk. Overall, 29% of patients met the criterion for no need of noninvasive cardiac testing based on the baseline hs-cTnI measurements. Our threshold of <4 ng/L is similar compared with earlier studies, including that of Shah et al., who derived and validated a threshold of hs-cTnI <5 ng/L at the time of presentation, which permitted safe discharge of almost two-thirds of patients from the ED, with an NPV of 99.6% for myocardial infarction or cardiac death during the index hospitalization and 30-day follow-up (34). Similarly, a rapid 1-h protocol with an hs-cTnI cutoff value of <6 ng/L also demonstrated a high NPV of 99.8% for exclusion of myocardial infarction (23). In addition to the baseline hs-cTnI threshold of <4 ng/L, we included the criterion of changes in serial hs-cTnI measurements at 0 and 2 h, which identified an additional 5% of patients with no need of testing. It is important to note that none of the 86 patients (34%) who met the no need for testing criterion had ACS, including myocardial infarction and unstable angina pectoris.

Our results suggest potentially large benefits for patients and healthcare providers and systems, including a reduction of testing in 34% of patients and reductions in hospital length of stay. Currently, ED providers face the dilemma of a lack of an objective test to help triage these patients without performing stress test or coronary CTA to exclude myocardial ischemia or obstructive CAD. Although outpatient follow-up and testing is an alternative approach based on the guidelines in the US (2), outpatient follow-up with noninvasive cardiac testing is poor, with fewer than half of patients undergoing the test (4). Hence, availability of an objective triage marker for noninvasive cardiac testing such as hs-cTnI would ease the pressure on physicians to perform these tests in every patient and would constitute a tremendous decision support. This hope is supported by the results from a recently published trial reporting the initial combined experience with hs-cTnI and coronary CTA in Europe (Better Evaluation of Acute Chest Pain with Coronary Computed Tomography Angiography—BEACON trial). The study suggested that serial hs-cTn could be used for identification of patients who can be safely discharged, and this approach could lead to a decreased need for subsequent cardiac testing (35).

Although our data based on a comparison of observed clinical care with conventional troponin vs predicted care with hs-cTnI suggested a significant benefit of hs-cTnI for the evaluation of patients presenting with suspected ACS in the ED, it remains to be determined whether these benefits from observational studies will translate into the real-world care of acute chest pain patients. Recently published data demonstrated that use of hs-cTn in patients with suspected ACS was associated not only with an increased rate of noninvasive cardiac investigations but also fewer in-hospital adverse events (36). Thus, the availability of measurable hs-cTnI values in virtually every patient without myocardial infarction begs the provocative hypothesis that there may be circumstances under which healthcare costs may be increased. Ultimately, prospective studies in which hs-cTnI will be used for clinical decision-making will be needed to answer this question.

Strengths of the current study include that the results are derived from 2 prospective clinical trials with patient populations who had similar distribution of risk factors, CAD, and rate of ACS. Furthermore, the results on potential savings in effectiveness and cost may be generalizable to US populations, as they are based on patient level data from a randomized multicenter comparative effectiveness trial in the US and include both functional and anatomic testing strategies.

Several limitations are noted. First, the measurements of hs-cTnI were available only in a subpopulation of patients in the ROMICAT II trial. Because of the inclusion criteria in the ROMICAT II trial, we studied only patients with conventional troponin below the 99th percentile, which introduces a selection bias. Second, the results of hs-cTnI measurements were not available to clinicians in the ROMICAT II trial. The hs-cTnI assay used in our study is not currently approved for clinical use in the US. Third, whether the results of our study can be replicated with other hs-cTn assays, such as the hs-cTnT assay that was recently approved by Food and Drug Administration, will need to be studied to derive appropriate triage threshold and management recommendations. Fourth, the potential decrease in the number of patients requiring noninvasive cardiac testing represents the best-case scenario. Similarly, estimated decreases in the length of stay, time to diagnosis, and estimated radiation dose would be most likely smaller in clinical practice. Fifth, we did not study whether implementation of hs-cTnI could increase cost in some patients, especially in those with increased hs-cTnI.

In conclusion, a combination of serial hs-cTnI measurements and cardiovascular risk factors identifies a criterion for no need of noninvasive cardiac testing in acute chest pain patients for whom obstructive CAD or myocardial ischemia needs to be ruled out. hs-cTnI could be an important decision support for healthcare providers to increase efficiency and safety and to decrease costs for the treatment of these patients.