Frequency, compliance, and yield of cardiac testing after high-sensitivity troponin accelerated diagnostic protocol implementation

Abstract

Background:

Among persons presenting to the emergency department with suspected acute myocardial infarction (MI), cardiac troponin (cTn) testing is commonly used to detect acute myocardial injury. Accelerated diagnostic protocols (ADPs) guide clinicians to integrate cTn results with other clinical information to decide whether to order further diagnostic testing.

Objective:

To determine the change in the rate and yield of stress test or coronary CT angiogram following cTn measurement in patients with chest pain presenting to the emergency department pre- and post-transition to a high-sensitivity (hs-cTn) assay in an updated ADP.

Methods:

Using electronic health records, we examined visits for chest pain at five emergency departments affiliated with an integrated academic health system 1-year pre- and post-hs-cTn assay transition. Outcomes included stress test or coronary imaging frequency, ADP compliance among those with additional testing, and diagnostic yield (ratio of positive tests to total tests).

Results:

There were 7564 patient-visits for chest pain, including 3665 in the pre- and 3899 in the post-period. Following the updated ADP using hs-cTn, 862 (23.5 per 100 patient visits) visits led to subsequent testing versus 1085 (27.8 per 100 patient visits) in the pre-hs-cTn period, (P < 0.001). Among those who were tested, the protocol-compliant rate fell from 80.9% to 46.5% (P < 0.001), but the yield of those tests rose from 24.5% to 29.2% (P = 0.07). Among tests that were noncompliant with ADP guidance, yield was similar pre- and post-updated hs-cTn ADP implementation (pre 13.0%, post 15.4% (P = 0.43).

Conclusion:

Implementation of hs-cTn supported by an updated ADP was associated with a lower rate of stress testing and coronary CT angiogram.

1. Introduction

In the United States (US), nearly 6 million persons annually present to emergency departments (EDs) with a chief complaint of chest pain, representing about 4% of all visits [1]. The variation and subtlety in acute coronary syndrome (ACS) presentations create a challenging task for emergency clinicians to manage these persons when they present to the ED since only approximately 5% of them will have a final diagnosis of ACS [2]. As a result, much over-testing occurs for the relatively small number of cases that actually have ACS, while on the other hand, owing to its high risk, missed myocardial infarction (MI) continues to be one of the most common reasons behind malpractice lawsuits filed against emergency medicine clinicians [3]. Accordingly, clinicians in the ED must use their clinical judgment to piece together several elements of a patient’s presentation to determine the appropriate disposition: history, physical exam, electrocardiogram (ECG), and cardiac biomarkers all inform this judgment, but none is sufficiently sensitive enough to rely on in isolation to adequately exclude the presence of ACS in the relatively short amount of time permitted in an ED visit.

These diagnostic dilemmas persist even with the recent development of high-sensitivity cardiac troponin assays (hs-cTn), so further adjunctive cardiac stress testing or coronary computed tomography angiography (CCTA) may be requested in patients with normal or inconclusive cTn results. Prior guidance advised it was reasonable to obtain stress testing or CCTA before discharge or within 72 h in patients with normal serial ECGs and cTn results [4]. However, the American Heart Association and American College of Cardiology recently updated guidelines for this scenario to recommend the use of accelerated diagnostic protocols (ADPs) featuring a hs-cTn assay, including no routine stress testing for patients deemed to be low-risk after their ED evaluation [5]. While stress testing and CCTA can play a critical role in further risk stratifying patients presenting to the ED with chest pain of suspected cardiac origin, overuse of these tests adds costs and potential harm to patients [6]. Moreover, keeping patients in the hospital is an increasing burden in the setting of historic hospital crowding, and this problem is exacerbated in health systems where expedited outpatient testing may not be accessible.

At present, about two out of every three hospitals in the US use an older generation cTn assay for the assessment of patients with chest pain and anginal equivalents [7]. Early real-world experiences with the hs-cTn assays in the US have demonstrated notable advantages of improved risk stratification, more rapid cycling intervals of repeated testing (e.g., as little as one hour between tests), and an evidence-based single hs-cTn strategy in low-risk patients [8–12]. However, by definition, the high-sensitivity assays yield detectable cTn values in more than half of patients – a radical shift in practice from older-generation assays when clinicians rarely detected cTn values in populations without chronic illnesses such as kidney disease [13].

Since the Food and Drug Administration (FDA) first approved hs-cTn assays for use in the US in 2017, about a third of US hospitals have adopted hs-cTn along with updated ADPs, [7] and evidence of their influence is growing. We previously demonstrated that initiation of hs-cTn with an updated ADP in our integrated hospital network resulted in an increase in the likelihood of any downstream testing but a reduction in the use of invasive and/or costly downstream services [12]. In this investigation, we sought to understand this finding further. To do so, we performed a pre-post analysis to understand the association of implementing an ADP that includes hs-cTn, with the use of subsequent cardiac testing, as well as whether these tests were compliant with ADP recommendations.

2. Methods

2.1. Implementation details

We performed a retrospective observational cohort study at 5 EDs in a large integrated health system to describe the use of cardiac testing following the adoption of a new ADP that includes hs-cTn. In the spring of 2017, we created a multi-disciplinary and cross-specialty implementation team to develop an ADP featuring a hs-cTn assay for a large integrated healthcare system located in the US Northeast [14]. Following clinician education, updates to the electronic health record (EHR) order entry, and in-laboratory validation, a system-wide transition to the Roche 5th-generation (hs-cTn) assay was completed on April 1, 2018, replacing a conventional cTn in the ADP (See Supplemental Appendix Fig. 1: Contemporary Troponin Accelerated Diagnostic Protocol for details). The ADP utilizes sex-specific hs-cTn cutoffs and advises hs-cTn sampling at arrival and additional testing one hour later [15,16]. In some cases, subsequent testing may be performed three hours later; conversely, a single test in low-risk patients presenting with >3 h of symptoms from arrival and a result <6 ng/L (below the level of quantification) was permissible or if an alternative diagnosis was confirmed after that first result (See Supplemental Appendix Fig. 2: High-Sensitivity Troponin Accelerated Diagnostic Protocol for details). Notably, in the post hs-cTn period, the updated ADP bundled low and intermediate-risk HEART scores (0–6 points) and recommended discharge without further cardiac testing if the absolute and delta hs-cTn values were reassuring [17]. The Mass General Brigham Institutional Review Board approved this study with a waiver of consent, which also follows the Strengthening the Reporting of Observational Studies in Epidemiology reporting guideline [18].

2.2. Selection of participants

ED visits for adult patients (≥18 years old) presenting with suspected ACS among five Mass General Brigham hospital EDs (Massachusetts General Hospital, Brigham and Women’s Hospital, Newton Wellesley Hospital, Brigham and Women’s Faulkner Hospital, and North Shore Medical Center) between April 1, 2017 – March 31, 2019 were included in the present analysis [12]. We defined “suspected ACS” as patients presenting to the ED with a chief complaint of chest pain (based on nurse documentation) who subsequently underwent further cardiac testing for acute coronary syndrome diagnosis or risk stratification. We selected a 1-year pre-period of April 1, 2017 – March 31, 2018, and a 1-year post-period of April 1, 2018 – March 31, 2019. Persons presenting with chest discomfort who lacked a primary care physician in the Mass General Brigham health system at the time of their visit were excluded given challenges to abstract relevant data. If patients had multiple ED visits for chest pain during the study periods, we chose one visit per unique presentation at random for analysis due to the influence of multiple ED visits for the same complaint on subsequent care. Lastly, we excluded those with ST-segment elevation myocardial infarction on ECG at any point during their ED visit and those with stage 4 or 5 kidney disease (GFR <30 mL/min) since cTn elevations in these individuals have comparably low specificity for acute MI and absolute cTn values designated by ADPs typically are not useful in this population. Both populations are excluded from the Mass General Brigham system-wide ADP guidance.

2.3. Classification approach

In the pre-period we classified low-risk persons identified by the ADP receiving further testing as noncompliant and intermediate or high-risk persons receiving further testing as compliant. In the post-period, per the updated ADP, we used a combination of absolute and delta hs-cTn values combined with risk category to classify compliance of further testing. Those individuals with normal absolute and delta hs-cTn values plus HEART low to intermediate risk as well as those with a delta hs-cTn <5 ng/L receiving further testing were classified as noncompliant. We designated further testing in all other scenarios as compliant; for more detail on our classification of risk, see Supplemental Appendix Table 1: Definitions of “Compliant” and “Noncompliant” Care for additional detail of our classification approach. To identify the yield in each group of tests (ratio of positive tests to total tests) information from Supplementary Fig. 4 for “Test-Strata” (severe-abnormal and moderate-abnormal), records of “CABG”, “Stent”, “Cath”, and medication-changes was used; subgroup analyses by test type were performed.

Since our study population only included those patients who received additional cardiac testing, by definition, we did not investigate compliant care among low-risk patients who did not receive further testing and potentially noncompliant care among high-risk patients who did not receive further testing. Furthermore, a high-risk patient hospitalized for further management without further cardiac testing could have been compliant in a scenario where they went straight to cardiac catheterization or medical management optimization based on recent abnormal testing prior to the index ED visit.

2.4. Data source and abstraction

Data were extracted from the Mass General Brigham Enterprise Data Warehouse. This warehouse holds structured and unstructured EHR data including clinician notes, patient demographic data, and diagnostic codes. Trained research associates performed chart review of the EHR and entered abstracted data to a REDCap database. Information on coronary disease risk factors, elements of the chest pain history, patient age and cTn timing and results were obtained. Research associates all had credentials to access the EHR (Epic), including required Epic orientation, HIPAA training, and Good Clinical Practice (GCP) certificates. Study onboarding included a REDCap orientation session with the Principal Investigator and several charts co-abstracted with the Principal Investigator. The research associates flagged any data elements they were unsure about for the Principal Investigator, who performed the final adjudications.

Chart abstraction elements to define a slightly suspicious, moderately suspicious, or highly suspicious chest pain history are listed in Supplemental Appendix Fig. 3: HEART Score History and ECG Scoring Criteria. A board-certified emergency physician (CWB) interpreted ECGs and determined the degree of repolarization disturbance, presence of pacemaker rhythms, left ventricular hypertrophy, bundle branch blocks, and other abnormalities that result in points among the various risk scores. If multiple ECGs were performed (including in a clinic on the same date of service prior to ED arrival) all were reviewed, and the tracing that resulted in maximum points was used to calculate the ECG component of the final risk score. Other data from the EHR used to inform risk such as race, gender, highest ED creatinine/GFR, clinical signs and symptoms of heart failure, initial ED disposition (home, transfer, ED observation or admission), and final ED disposition (home, transfer or admission) were abstracted.

Results from stress testing (e.g., exercise tolerance test, Sestamibi nuclear perfusion stress test, stress echo or other) or CCTA were classified as normal, mildly abnormal, moderately abnormal or severely abnormal (see Supplemental Appendix Table 2: Stress Test Result Classification Criteria) [19]. If applicable, reports from cardiac catheterization, surgical reports of coronary artery bypass grafting, or evidence of change in medical management (e.g., initiation or up-titration of nitrates, change in beta blocker, calcium channel blocker or aspirin therapy) were also identified and results classified [5,20,21].

2.5. Outcome measures

The primary outcome of this analysis was the rate of cardiac stress testing or CCTA following patient-visits for suspected ACS between the pre and post hs-cTn time periods, as expressed by the number of tests per 100 patient-visits. Two secondary outcomes were the compliance by providers in ordering subsequent cardiac imaging and the diagnostic yield among those who had testing. In addition, we measured the 30-day all-cause mortality among the group of individuals discharged from the ED without any additional cardiac testing.

3. Data analysis

Descriptive statistics of patient-visits in pre and post groups are presented – these included demographics, clinical characteristics, and sampling of at least one cTn test during the visit. The yield (ratio of positive studies to the total number of studies) was analyzed by using Pearson’s chi-squared and Fisher exact tests between two groups of pre and post and the two groups determined to be compliant and non-complaint. To examine temporal trends around compliance rates in both comparison periods, the Cochran Armitage trend test was applied.

All P values are two sided with a result of <0.05 considered significant. We performed all analyses with JMP software version 16 (JMP Statistical Discovery LLC).

4. Results

A CONSORT diagram is depicted in Supplementary Appendix Fig. 4. There were 7564 individual visits for chest pain in this analysis, including 3665 in the pre-and 3899 in the post-period. Following hs-cTn introduction into the updated ADP, 862 (23.5 per 100 patient visits) visits led to testing versus 1085 (27.8 per 100 patient visits) in the pre hs-cTn period, (P < 0.001). In the post hs-cTn period, 876 tests were performed (742 stress tests and 134 CCTAs) versus 1113 tests (1026 stress tests and 87 CCTAs) in the pre-period (P < 0.001).

As illustrated in Table 1, comparing the pre and post-period study groups who all underwent further cardiac testing, individuals with suspected ACS undergoing stress testing or CCTA showed a non-significant increase in risk factors for coronary artery disease, such as advanced age and male sex; they also had more prevalent heart failure. Black persons tended to be less likely to go onto further stress testing compared to patients of other races. However, the only characteristic with a statistically significant difference was age (P = 0.04).

Table 1.

Patient characteristics of pre and post-high sensitivity troponin assay introduction.

		Total patient-visits for chest pain		Patient-visits for chest pain followed by stress testing or CCTA
		Pre-period (N = 3665)	Post-period (N = 3899)	Pre-period (N = 1085)	Post-period (N = 862)
Mean age, years (SD)		56.3 (17.1)	55.7 (17.4)	64.6 (13.1)	66.3 (12.7)*
Sex, %	Female	1942 (53.0%)	2132 (54.7%)	546 (50.3%)	412 (47.8%)
	Male	1723 (47.0%)	1767 (45.3%)	539 (49.7%)	450 (52.2%)
Race, %	White	2710 (73.9%)	2826 (72.5%)	856 (78.9%)	683 (79.2%)
	Black	387 (10.6%)	389 (10.0%)	65 (5.9%)	40 (4.6%)
	Hispanic/Latinx	111 (3.0%)	131(3.4%)	48 (4.4%)	37 (4.3%)
	Other	457 (12.5%)	553 (14.2%)	116 (10.7%)	102 (11.8%)
Primary language, %	Non-English	369 (10.1%)	392 (10.1%)	109 (10.1%)	88 (10.2%)
	English	3274 (89.3%)	3498 (89.7%)	970 (89.4%)	771 (89.4%)
	Unknown/declined	22 (0.6%)	9 (0.2%)	6 (0.6%)	3 (0.4%)
Insurance, %	Commercial	2188 (59.7%)	2214 (56.8%)	558 (51.4%)	398 (46.2%)
	Medicare	1017 (27.8%)	1081 (26.1%)	420 (38.7%)	373 (43.3%)
	Medicaid	368 (10.0%)	586 (15.0%)	82 (7.6%)	77 (8.9%)
	Veteran/State	62 (1.7%)	46 (1.2%)	21 (1.9%)	9 (1.1%)
	Uninsured/missing	30 (0.8%)	35 (0.9%)	4 (0.37%)	5 (0.58%)
Existing heart failure, %		233 (6.4%)	267 (6.9%)	100 (9.2%)	93 (10.8%)
Existing cardiovascular disease, %		835 (22.8%)	838 (21.5%)	356 (32.8%)	279 (32.4%)
Existing diabetes, %		620 (17%)	612 (15.7%)	269 (24.8%)	213 (24.7%)
Received at least one troponin test during visit, %		3213 (87.7%)	3380 (86.7%)	1083 (99.8%)	857 (99.4%)

^*P = 0.04 between pre and post cohorts.

Among those who had further testing, the ADP featuring contemporary cTn in the pre-period classified 207 (19.1%) of patients as low-risk and the ADP featuring hs-cTn classified 461 (53.5%) as low-risk (e.g., no further cardiac testing recommended) (P < 0.001). Following hs-cTn implementation, Fig. 1 shows the share of further cardiac testing that was protocol-complaint fell from 80.9% to 46.5% (P < 0.001), while the share of testing that was noncompliant increased from 19.0% to 53.5%, over twofold. Despite this, as illustrated in Fig. 2, the diagnostic yield (ratio of positive tests to total tests) among all tests was similar at 22.2% and 21.8%, (P = 0.83). In Figs. 3 and 4, the diagnostic yield is further divided into compliant and noncompliant categories. The difference in the yield of compliant tests was not statistically significant in the pre and post periods (24.5% to 29.2% [P = 0.07]). The yield among noncompliant tests was not significantly different, from 13.0% to 15.4% (P = 0.43), respectively. Among patients with noncompliant but positive tests, 18.5% and 18.3% underwent subsequent coronary revascularization, respectively (P = 0.98).

Fig. 1.

Compliance with Accelerated Diagnostic Protocol (ADP).

MIBI = Sestamibi nuclear perfusion stress test, ETT = exercise tolerance test, hs-cTn = high sensitivity troponin, MI = myocardial infarction, CI = confidence interval, CABG = coronary artery bypass graft, CT = computed tomography, MRI = magnetic resonance imaging, PCI = percutaneous coronary intervention, CCTA = coronary computed tomography angiogram.

Fig. 2.

Overall cardiac testing yield*.

*Yield = ratio of positive tests to total tests, MIBI = Sestamibi nuclear perfusion stress test, ETT = exercise tolerance test, hs-cTn = high sensitivity troponin, MI = myocardial infarction, CI = confidence interval, CABG = coronary artery bypass graft, CT = computed tomography, MRI = magnetic resonance imaging, PCI = percutaneous coronary intervention, CCTA = coronary computed tomography angiogram.

Note that some patients had more than one stress test (e.g., a positive ETT followed by a positive MBI or a positive CCTA followed by a negative MIBI) so the sum of tests by modality exceeds the total number of compliant/noncompliant classifications and number of patient visits; however, the sequence of positive test followed by another positive test occurred in only five cases and none of those were in a low-risk category by HEART score.

Fig. 3.

Compliant cardiac testing yield*P = 0.49.

*Yield = ratio of positive tests to total tests, MIBI = Sestamibi nuclear perfusion stress test, ETT = exercise tolerance test, hs-cTn = high sensitivity troponin, MI = myocardial infarction, CI = confidence interval, CABG = coronary artery bypass graft, CT = computed tomography, MRI = magnetic resonance imaging, PCI = percutaneous coronary intervention, CCTA = coronary computed tomography angiogram.

Note that some patients had more than one stress test (e.g., a positive ETT followed by a positive MBI or a positive CCTA followed by a negative MIBI) so the sum of tests by modality exceeds the total number of compliant/noncompliant classifications and number of patient visits; however, the sequence of positive test followed by another positive test occurred in only five cases and none of those were in a low-risk category by HEART score.

Fig. 4.

Noncompliant cardiac testing yield*.

P = 0.35.

*Yield = ratio of positive tests to total tests, MIBI = Sestamibi nuclear perfusion stress test, ETT = exercise tolerance test, hs-cTn = high sensitivity troponin, MI = myocardial infarction, CI = confidence interval, CABG = coronary artery bypass graft, CT = computed tomography, MRI = magnetic resonance imaging, PCI = percutaneous coronary intervention, CCTA = coronary computed tomography angiogram.

Note that some patients had more than one stress test (e.g., a positive ETT followed by a positive MBI or a positive CCTA followed by a negative MIBI) so the sum of tests by modality exceeds the total number of compliant/noncompliant classifications and number of patient visits; however, the sequence of positive test followed by another positive test occurred in only five cases and none of those were in a low-risk category by HEART score.

In the pre hs-cTn period, all noncompliant testing was among low-risk HEART (0–3 points) patients. In contradistinction, that was not true in the post hs-cTn period (Table 2). In the post hs-cTn period, we observed that 88.2% of the noncompliant testing in the post-period was among persons with an intermediate HEART score of 4–6 points. In addition, only 137 low-risk HEART score patients had noncompliant cardiac testing in the post-period, versus 205 in the pre-period.

Table 2.

Interaction of HEART score and testing yield in post-period.

HEART Scores	Compliant Stress Test Yield % positive (number positive)
	Overall	MIBI	ETT	Coronary CTA	Stress Echo	Other
Low Risk 0–3	19.0% (4)	28.6% (2)	0.0% (0)	25.0% (2)	NA	0.0% (0)
Intermediate Risk 4–6	24.7% (47)	26.3% (40)	12.0% (3)	37.5% (6)	100.0% (1)	0.0% (0)
High Risk 7–10	34.7% (66)	33.5% (56)	50.0% (6)	62.5% (5)	0.0% (0)	75.0% (3)
HEART Scores	Noncompliant Stress Test Yield % positive (number positive)
	Overall	MIBI	ETT	Coronary CTA	Stress Echo	Other
Low Risk 0–3	4.2% (5)	5.3% (2)	5.9% (2)	2.0% (1)	NA	0.0% (0)
Intermediate Risk 4–6	18.8% (62)	17.3% (39)	22.2% (14)	35.8% (19)	0.0% (0)	0.0% (0)
High Risk 7–10	30.8% (4)	36.4% (4)	0.0% (0)	NA	NA	NA

In the sensitivity analysis, we did not observe any pattern in the monthly compliance trend to suggest an adjustment period to the new ADP followed by a sustained higher compliance rate towards the end of the post-period (P = 0.68). Similarly, there was no significant difference in adverse events between the pre- and post-test groups. 30-day all-cause mortality was <0.1% (n = 3) and 0.0% (n = 0) in the pre-group and post-group, respectively (P = 0.11). 30-day hospitalization for a primary hospital discharge diagnosis of myocardial infarction was <0.1% (n = 3) and < 0.1% (n = 3) in the pre-group and post-group, respectively (P = 0.68).

5. Discussion

In this study across 5 EDs in a large integrated health system, in the year after the implementation of hs-cTn, individuals presenting to the ED with suspected ACS were less likely to receive cardiac testing compared to the prior year when older generation cTn tests were used. This finding is consistent with other investigations but runs counter to the expectations of staff prior to switching over to a hs-cTn assay [12,22,23]. Nonetheless, clinicians were less likely to adhere to a diagnostic recommendation not to test based on the updated ADP, driven primarily by testing among intermediate risk HEART score patients with reassuring hs-cTn results. On the other hand, testing among persons with a low-risk HEART score fell dramatically, demonstrating that the overall reduction in testing largely likely came from this group. The yield of cardiac testing trended upwards when clinicians were compliant with the chest pain ADP after implementation of hs-Tn. This finding suggests that updated ADPs driven by hs-cTn cut points and delta values may better identify patients with occlusive coronary disease more likely to benefit from further cardiac testing versus ADPs with conventional cTn. Lastly, yield of ADP-noncompliant testing was low in both pre- and post-hs-cTn ADP implementation. This study suggests updating a chest pain ADP to incorporate hs-cTn may safely result in less downstream cardiac testing, particularly among the lowest risk, yet clinicians may not trust the guidance not to test patients at intermediate risk via commonly used risk scores despite reassuring hs-cTn patterns. The findings from this study identify education of clinicians regarding outcomes in this group as being a potential intervention to further reduce unnecessary testing.

As clinicians caring for those with suspected acute MI recognize advantages from hs-cTn assays and seek to transition from older-generation versions of these tests, leaders must anticipate the common challenges experienced by clinical staff making this switch [24,25]. Failure to appropriately plan the transition will likely lead to resistance and frustration from clinical staff, including a risk of overuse of costly resources such as consultation, further cardiac testing and hospitalization [26]. To smooth this transition, local leaders from key stakeholder groups such as emergency medicine, cardiology, laboratory, internal/hospital medicine and informatics should convene a workgroup to plan a timeline for implementation, make key decisions (e.g., 99th percentile cut points, gender-based or not), generate a shared protocol with repeat testing timeframes and disposition recommendations and create in-service tools for staff education in advance of the go-live date. Lack of trust or familiarity with the new ADP, ongoing concern for unstable angina or unique patient-specific factors such as expectations for further testing by the patient or outpatient clinicians combined with a lack of confidence in the reassuring hs-cTn pattern may drive additional testing among individuals with a low-risk hs-cTn pattern.

With tools providing increased sensitivity to detect risk, the ability to categorize individuals as low-risk (with more potential patient and system-level harms from further testing) allows for reduction in unwarranted variations in clinical practice. Focus can instead be placed on higher-risk patients more likely to yield abnormal testing that results in a change in management that impacts patient-centered outcomes. Our analysis suggests that patients most likely to fulfill these criteria are those with hs-cTn absolute and delta results that preclude expedited discharge. Shared decision making in such grey-zone presentations is another well-studied approach that can both limit low-yield testing and help both patients and emergency physicians feel comfortable with a specific testing strategy [27]. The use of hs-cTn-leveraged ADPs may assist clinicians to both consider certain dangerous conditions in their differential diagnosis and adequately evaluate them via an evidence-based testing strategy. Our finding that positive findings among noncompliant testing did result in coronary revascularization in a subset of patients highlights the value of tailoring ADP recommendations to individual patients when there is a compelling rationale to do so.

The future state of how hs-cTn will be used in the ED continues to evolve. For example, it is still unclear whether a hs-cTn-only ADP or one integrated with a risk score is the optimal strategy; furthermore, a 1-h versus 2-h delta approach appear to yield similar results [17,25]. Recent investigations of a hs-cTn-only approach focus on the negative predictive value for acute MI and not on the sensitivity for 30-day MACE [28]. Point of care hs-cTn platforms will soon be introduced that will allow the initial sample to be drawn and result closer to the triage time, facilitating gains in ED length of stay that have been seen in Europe but not yet clearly demonstrated in the US [23,29]. The role of clinical decision support integration into the EHR is another emerging area ripe for integration of chest pain ADPs including delta hs-cTn interpretation and the calculation of risk scores [30]. Artificial intelligence and machine learning tools also show promise as a future area of practice change to further optimize the ED evaluation of patients presenting with chest pain [31].

5.1. Study limitations

The study results should be interpreted in the context of its limitations. This pre-post study is not designed to make causal inference and should not be interpreted as such. We did not include patients without further cardiac testing at the index visit, so we were unable to calculate their care as compliant or noncompliant with ADP guidance. Accordingly, we were unable to calculate the total rate of compliance among all patients presenting with chest pain to the ED in the pre and post periods. Additionally, our approach cannot exclude other impactful temporal changes in care between the pre and post periods; although we avoided extending the analysis period into the COVID-19 pandemic period to avoid an influence on our analysis. Further, our previous analysis did not show an impact from including a one-month washout period in the analysis [12]. Our data reflect the experience of a single health system, although we did study a combination of two academic and three community EDs with variable access to resources such as observation units and further cardiac testing. Additionally, we restricted the study cohort only to patients with a chief complaint of chest pain and with a primary care physician (PCPs) within our health care system. As a result, our findings may not be generalizable to patients with non-chest pain anginal equivalents, such as dyspnea or to patients without PCPs. More specifically, patients with an ongoing provider relationship may differ from those who do not, limiting generalizability. Further, retrospective HEART score determination may differ from prospective scoring, especially for elements such as chest pain history that may have been variably documented [32]. The difference between ADPs in the pre and post periods with respect to intermediate risk patients is another factor influencing the comparison of ADP concordance between the 2 periods. In the pre period, we considered all stress testing among intermediate risk patients to be compliant. However, in the post period, we bundled intermediate risk patients who fulfilled rule out hs-cTn absolute and delta values with both low- and intermediate-risk patients (e.g., HEART 0–6) and recommended discharge without further testing. As a result, the care of the same intermediate risk patient (HEART 4–6) kept for stress testing would have been classified as compliant in the pre period and noncompliant in the post period. Accordingly, this phenomenon likely suppressed the compliance rate in the post period.

6. Conclusions

The first year of hs-cTn assay implementation supported by an ADP was associated with fewer net cardiac testing studies. Among those patients in whom ADP-compliant testing was performed, the imaging yield showed a non-significant pattern of increase. These results further extend a reassuring message that transition to hs-cTn testing is not associated with inappropriate increase in utilization of healthcare resources.

Abbreviations:

Hs-cTn

high sensitivity troponin

MI

myocardial infarction

AMI

acute myocardial infarction

CI

confidence interval

CPT

current procedural terminology

CABG

coronary artery bypass graft

OPPS

outpatient prospective payment system

ED

emergency department

ECG

electrocardiogram

CT

computed tomography

CCTA

coronary computed tomography angiogram

MRI

magnetic resonance imaging

PCI

percutaneous coronary intervention