The significance of historical troponin elevation in acute heart failure: Not as reassuring as previously assumed

Historical cardiac troponin (cTn) elevation is commonly interpreted as lessening the significance of current cTn elevations at presentation for acute heart failure (AHF). Evidence for this practice is lacking. Our objective was to determine the incremental prognostic significance of historical cTn elevation compared to cTn elevation and ischemic heart disease (IHD) history at presentation for AHF.


INTRODUC TI ON
3][4] "Transitioning a larger proportion of patients to the outpatient setting" from the ED, by improving risk stratification at ED disposition, has been identified as a key "unmet need" in AHF research by the Society for Academic Emergency Medicine (SAEM) and Heart Failure Society of America (HFSA). 5rdiac troponin (cTn) elevation in AHF represents a longstanding and high-stakes problem for ED risk stratification, being present in roughly half of patients. 6Around 10% of AHF presentations involve concomitant AMI, 2,7,8 a majority of which present without electrocardiogram ST-elevation. 2The other 40% of AHF patients with cTn elevations have nonischemic myocardial injury, which has also been associated with adverse events across multiple studies. 6EM and HFSA recently singled out the question of whether "all AHF patients with a troponin elevation absent suspected acute coronary syndrome require admission to the hospital?"as one of the four highest priority topics for AHF research in the ED setting. 9cTn elevation, regardless of being due to ischemia or injury, is included as a risk factor suggesting hospital admission in clinical decision rules 10 for ED disposition decisions in AHF.Thus, a common practice is to risk stratify AHF patients by comparing current cTn to historical values in the electronic medical record (EMR). 11The rationale behind this practice aligns with the goal of differentiating when an elevated cTn in AHF should prompt hospital admission versus ED discharge: (1) it is likely that not all cTn elevations are equally significant for risk prediction, (2) it may be possible to identify subgroups with elevated cTn who would versus would not benefit from hospital admission, and (3) identifying these patients may facilitate fewer unnecessary hospitalizations.
The practice assumes that cTn elevations are less serious (i.e., safer) in patients with historical (i.e., "chronic") elevation, yet literature examining the validity of this assumption is lacking, despite its ubiquitous application.
We hypothesized that elevated cTn in the ED would be associated with adverse events in AHF patients, regardless of prior historical values.Our primary objective was to assess the incremental prognostic value of historical cTn elevation after accounting for ED cTn and other predictors of adverse events.A secondary objective was to determine whether prior evaluation for ischemic heart disease (IHD) moderated prognostic significance of historical cTn, such as by delineating subgroups wherein the value of hospital admission for ischemic testing was greater.

Study design and setting
We performed a preplanned secondary analysis of two prospective observational patient cohorts 2,3 from five urban academic hospitals during 2017 through 2021 in two U.S. states.Hospitals included three Level I and two Level II trauma centers, with ED yearly visits between 60,000 and 100,000.Study design for both cohorts, REED-AHF 3,12 and CLEAR-AHF, 2 have been published previously.These prior reports and the current article were prepared in accordance with the STROBE guidelines.Research was institutional review board approved at Wayne State University and Indiana University.

Participants and data sources
Detailed inclusion and exclusion criteria, data collection practices, and data sources in REED-AHF 3,12 and CLEAR-AHF 2 have been published previously.Participants presenting to the ED, including on weekends and outside the hours of 8 a.m.-5 p.m., were enrolled after informed consent if the treating clinician suspected possible AHF (as reported by the physician to the screening research assistant).
Diagnosis was adjudicated by two blinded experts (with a third to break ties) to arrive at the final cohorts.Patients with ST-elevation myocardial infarction (STEMI) and cardiogenic shock at ED presentation were excluded from both studies.To meet the requirement for independence of observations in our statistical analysis, duplicate patients were excluded.
Conclusions: Historical cTn elevation in AHF patients is a harbinger of worse outcomes for patients who have not had a prior IHD workup and should prompt evaluation for underlying ischemia rather than reassurance for discharge.With known IHD history, historical cTn elevation was neither reassuring nor detrimental, failing to add incremental prognostic value to current cTn elevation alone.

K E Y W O R D S
acute decompensated heart failure, electronic medical record, emergency medicine, incremental prognostic value, ischemic heart disease, risk prediction, troponin

Follow-up and outcome adjudication
Patients were followed for 90 days or longer by telephone and EMR follow-up of serious adverse events (SAEs). 2,3,12SAEs were checked against statewide health information exchange data to capture events occurring at facilities outside the participating health systems.AMI, as defined by the 4th Universal Definition of MI, 13 and AHF-related readmission were adjudicated by two reviewers blinded to one another, with a third expert to break ties.Adjudication of AMI and diagnosis were performed by two separate groups of investigators, to prevent bias that could arise from the same person adjudicating diagnosis and outcomes.ECG findings in the ED, if commented on by the ED clinician (e.g., "no ST-depression," "T-wave inversions present but unchanged from prior ECG," "Smith-Modified Sgarbossa Criteria not present") were coded as interpreted at that time.If the physician documentation did not explicitly describe one or more ECG findings or their chronicity (i.e., new vs. present on prior ECG), two investigators reviewed the ECG and any prior ECGs independently, with disputes resolved by discussion or a third adjudicator.Source documents for other data points were reviewed by two or more blinded adjudicators.

Outcomes
The primary outcome of this analysis was a composite at 90-day follow-up ("COMP90") of death/CPR, mechanical cardiac support (MCS), intubation, new or emergent hemodialysis, and/or AMI/ percutaneous coronary intervention (PCI)/coronary artery bypass grafting (CABG).This specific composite was chosen based on its use in the STRATIFY decision instrument, 14,15 one of three structured risk scores endorsed by the American College of Emergency Physicians for aiding ED disposition decisions in AHF, 10 and other recent AHF literature on ED patients. 2,3Secondary outcomes included the same composite at 30-day follow-up ("COMP30"), 30-day AMI/ PCI/CABG, and adjudicated AMI at the index encounter.

Primary predictor: troponin elevation
cTn values were collected from the index ED visit and the most recent historical cTn value available in the EMR (up to 2 years prior).Elevation was defined as a value exceeding the manufacturer reference range.Since 2007, all cTn assays worldwide have been standardized to an upper reference limit (URL) defined by the 99th percentile value of that assay tested in a healthy reference population.While some interassay differences remain due to assay precision, scaling of both conventional sensitivity and high-sensitivity troponin assays to the standard of each assay's 99 th percentile URL is recognized as the preferred approach for achieving uniformity in clinical diagnosis and trial endpoints by the International Federation of Clinical Chemistry, the European Society of Cardiology, the American Heart Association, the 4th Universal Definition of MI, and others. 16The 99 th percentile URL standard has been employed in virtually all analyses of cTn's significance for AHF over the past 20 years. 6,16

Secondary predictors and covariates for adjusted analysis
Estimated glomerular filtration rate (eGFR), left ventricular ejection fraction (LVEF), age, and history of a prior workup for IHD were considered as potential confounders based on their association with cTn levels and/or AHF SAEs. 2,6,13,16Each variable was therefore included in the multivariable-adjusted analysis.IHD history was also considered as a potential moderator.An interaction term between IHD and troponin was included in the multivariable models to test the hypothesis that historical troponin elevation has different prognostic implications depending on whether a patient had a positive, negative, or unknown IHD history at the time of the index encounter.
A "positive" IHD history was defined as documentation prior to the index ED visit of known coronary artery disease (CAD), AMI, PCI, CABG, and/or cardiologist diagnosis of ischemic cardiomyopathy.
Patients without history of CAD/AMI/PCI/CABG/ischemic cardiomyopathy were further subdivided into "negative" (a documented negative CAD workup via cardiac catheterization and/or cardiologist diagnosis of nonischemic cardiomyopathy) or "unknown" (no prior CAD workup and the patient has not been diagnosed previously as having or not having IHD).
Other variables felt to be informative for the ischemic secondary outcomes were included in the descriptive, but not multivariable, analyses.These included first (earliest) cTn versus highest/peak value in the ED, "delta" value (percent change) if multiple ED troponins were ordered, symptoms (chest pain, shortness of breath) at presentation, various ECG findings, and utilization rates of cardiac catheterization.

Description of the cohort
All statistical analyses were conducted in R (R Studio 2021.09.1).Variables were described as median/interquartile range (IQR) or percent (n).
Descriptive comparisons utilized the Wilcoxon rank-sum or chi-square tests.For this descriptive analysis, cohort characteristics were stratified by (1) the primary outcome (Table 1), (2) the secondary outcomes (Table 2), and (3) elevated versus normal initial cTn in the ED (Table S1).
An alpha of 0.05 was assumed for hypothesis testing.

Multivariable analysis to assess the incremental prognostic significance of historical troponin and IHD history
Logistic regression was used to evaluate the outcomes versus historical cTn (elevated vs. not elevated) adjusted for ED cTn, age, LVEF, eGFR, and IHD history.An interaction term between IHD history and historical cTn was included to test whether history of IHD (and history of IHD workup) altered prognostic value.Model results were reported with covariate adjusted odds ratios (aOR) with 95% confidence interval (95% CI).

Ischemic versus nonischemic adverse events
Next, we assessed whether the association between elevated cTn and the primary outcome was driven by ischemic adverse events, nonischemic adverse events, or both (Figure 2).Patients experiencing AHA for risk stratification in AHF. 17 It has previously been validated to predict both in-hospital mortality 18 and death or cardiovascular events at long-term follow-up (median 3 years after admission). 19

Sensitivity analyses
cTn, both the ED and the historical values, were also evaluated as continuous variables in a sensitivity analysis to assess whether magnitude of cTn elevation affected the results.The ratio of ED to historical troponin (as a single variable substituted for both terms in multivariable modeling) was tested for a similar reason.Over the

TA B L E 2 (Continued)
study period, multiple conventional sensitivity (cs-cTn) assays and high-sensitivity (hs-cTn) assays came in and out of use at the study institutions.To partially adjust for changing assays within and between institutions over time, all cTn values were converted to a percentage of the assay 99 th percentile URL.We felt this method best reflected clinical reality, particularly for the common clinical practice of comparing current cTn value to prior EMR values, since institutions including our own change cTn assays as frequently as every 2-3 years.
While the 99th percentile of a healthy reference population should theoretically be equivalent between hs-cTn and cs-cTn sensitivity assays, 16 changing standards for a "healthy reference population" have led to observations that some hs-cTn assays may detect a greater proportion of patients above the 99th percentile URL than with cs-cTn. 16Therefore, we performed sensitivity analyses to see if the primary results changed with hs-cTn versus cs-cTn.Additionally, we compared ED cTn positivity (elevated vs. not) and continuous value (as multiple of URL) in the sample by cs-cTn versus hs-cTn, after regression adjustment for patient sex and prior history of cTn elevation on a cs-cTn assay to account for selection bias (e.g., differential gender and cTn historical positivity by cs-cTn vs. hs-cTn).In a 2023 analysis of 550 U.S. hospitals in the American College of Cardiology's NCDR Registry, less than one-third had implemented hs-cTn as of the most recently available data. 20However, this percentage is expected to increase in the coming years as implementation of hs-cTn in the United States increases.Therefore, we felt the results including both cs-cTn and hs-cTn would be timely and informative, particularly in evaluating whether the results of our current research question changed based on assay sensitivity (i.e., this sensitivity analysis).

Sample size determination and missing data
Sample size was determined to target 10 events per variable in logistic regression for the primary outcome and Cohen's ω of 0.3 (99.9% power) to 0.154 (80.3% power). 21Missing historical cTn was evaluated as "not elevated," and a sensitivity analysis was performed with and without patients missing historical cTn to evaluate validity of this assumption.Patients with missing ED cTn were excluded (n = 6); other model covariates had 0% missingness.For further details please refer to the supplementary materials.
Adjudicated AMI was present for 29 patients at index visit (9%), and 33 (10%) experienced AMI/PCI/CABG by 30 days.Elevated ED cTn was associated with historical cTn elevation (p < 0.05), ECG criteria suggestive of ischemia, lower eGFR and LVEF, greater proportional change in delta cTn, greater current/prior cTn ratio, and greater rates of the primary outcome (Table S1).

Patient characteristics by primary and secondary outcomes
Table 1 presents characteristics stratified by the primary outcome, and Table 2, by the secondary outcomes.Greater initial ED cTn value (as multiple of URL), elevated ED cTn, greater change on serial/delta troponins, and higher rates of new ECG abnormalities in the ED were associated with the primary outcome (Table 1) and every secondary outcome (Table 2).The primary outcome was also associated with older age, lower eGFR, higher BUN, lower sodium, and higher GWTG-HF risk score (Table 1).The composite at 30 days (COMP30), but not the primary outcome, was associated with historical cTn elevation (Tables 1 and 2).
AMI at the index visit accounted for 29/33 (88%) patients with AMI/PCI/CABG by 30 days, while 4/33 events occurred after discharge from the index encounter.The initial ED cTn was elevated for 23/29 index AMIs, while five (21%) experienced their first cTn elevation on a serial lab draw during the ED or inpatient course (Table 2).
Patients with 30-day AMI/PCI/CABG or index AMI were older and had lower eGFR (Table 2).Among those experiencing either outcome, only 45% reported chest pain in the ED, compared to 37% without an ischemia-related event (p = 0.3, Table 2).
A minority (17%) of patients with an adjudicated AMI at ED presentation received coronary catheterization during the index hospitalization compared to posthospital at 30-day (34%) and 90-day (38%) follow-up (Table 2).Over half of all catheterizations were diagnostically negative for AMI and did not result in PCI/CABG (Table 2).
Catheterization within 90 days was similarly common in those with versus without a known IHD workup before the index encounter as well as those with a positive versus negative prior workup (all p > 0.05).

Multivariable analysis to assess the incremental prognostic significance of historical troponin and IHD history
Prognostic value of historical cTn elevation for predicting the 90day primary outcome, after adjusting for ED cTn elevation, age, eGFR, LVEF, and IHD history, depended on whether patients had a history of workup for IHD at the time of the index ED visit (p interaction = 0.04, Figure 1).In those with a prior IHD workup, regardless of the prior workup being positive or negative, historical cTn did not add significant predictive value (p > 0.05) to the risk factors at presentation (Figure 1).In patients without a prior workup for IHD, an elevated historical cTn predicted fivefold greater odds of the 90-day outcome (aOR 5.27, 95% CI 1. 23-22.4).Similar relationships, after multivariable adjustment, were present for historical cTn elevation versus each secondary outcome (Figure 1).ED cTn as a continuous multiple of the URL was also significantly associated with greater odds of each outcome (Figure 1).The ratio of the ED and historical troponins had virtually identical predictive value as the ED cTn alone (for primary outcome; ratio aOR 1.45 [95% CI 1.13-1.87],ED cTn alone aOR 1.39 [95% CI 1.12-1.73]).
None of the observed relationships between cTn (i.e., whether historical vs. ED, binary vs. continuous) and outcomes differed significantly when comparing results by hs-cTn versus cs-cTn (all p > 0.05) or when excluding missing historical cTn, in a sensitivity analysis.

Ischemic versus nonischemic adverse events
Figure 2 presents the primary outcome composite stratified by ischemia-related adverse events (AMI, PCI, CABG), versus nonischemic events, at 90 days.Significantly more patients experienced  Historical and ED Troponin vs. Outcomes ischemic adverse events when ED cTn was elevated versus normal (absolute risk difference [RD] +12.6%, 95% CI 5.2%-19.9%,p = 0.001).Among 40 total patients with at least one ischemic adverse event, 11 (27.5%) also experienced a nonischemic adverse event.A similar number (n = 43) experienced a nonischemic adverse event without any ischemic events.The proportion of patients with strictly nonischemic outcomes was numerically greater for those with elevated versus normal ED cTn (Figure 2) but not statistically significant (RD +3.1%, 95% CI +10.8% to −4.6%, p = 0.497).

Historical troponin association with ED clinician decision making
ED clinicians were significantly more likely to discharge a patient with an elevated ED cTn if there was a history of prior troponin elevation (78% vs. 32%, p = 0.025).After multivariable adjustment for ED cTn, historical cTn, IHD history (ischemic, nonischemic, or unknown) and GWTG-HF risk score, a greater ED cTn was associated with greater odds of admission versus discharge (aOR 1.51 per multiple URL, 95% CI 1.05-2.19).However, compared to patients with high ED cTn and normal historical cTn, there was greater multivariable adjusted odds of discharge if both ED and historical cTn were high (aOR 0.970 per multiple URL, 95% CI 0.94-0.99,p interaction < 0.05; Figure 3).This effect was present regardless of IHD status, which was not associated with ED discharge versus admission decision after multivariable adjustment.
Among those discharged (n = 32, 11%), AMI/PCI/CABG occurred in three discharged patients (8.3%), including two patients (5.6%) adjudicated to have AMI at the index encounter.Figure 4 presents the composite outcome at 30 days stratified by ED disposition and ischemic versus nonischemic adverse events.Compared to patients admitted to the hospital or observation, patients discharged from the ED had a nearly threefold lower rate (7.5% vs. 2.7%) of nonischemic adverse events at 30 days.Conversely, the rates of ischemic adverse events were similar between ED discharge versus observation/inpatient disposition (8.3% vs. 9.8%).
F I G U R E 2 Ninety-day adverse events stratified by ED troponin and ischemic versus nonischemic.Patients experiencing the 90-day composite are stratified by (1) elevated versus normal troponin and (2) whether one or more ischemic adverse event occurred for the patient (AMI/PCI/ CABG).AMI, acute myocardial infarction; CABG, coronary artery bypass grafting; PCI, percutaneous coronary intervention.

DISCUSS ION
In this preplanned secondary analysis, we evaluated the incremental prognostic value of historical troponin values when risk stratifying AHF in the ED (Figure 5).Our study has two primary findings.
First, a prior history of cTn elevation predicted greater adverse event risk in patients without a prior IHD workup (Figure 1) and did not reduce the risk associated with elevated ED cTn when IHD history was known.This runs directly counter to common practice patterns, 11 including what we observed in the current study: patients with an elevated ED cTn had greater multivariable-adjusted odds of discharge if they had a history of prior cTn elevation, while IHD history was not associated with ED disposition.Second, the adverse event risk associated with elevated cTn in the ED was primarily driven by patients with ischemic adverse events (AMI/PCI/ CABG; Figure 2), suggesting that accurate identification of ischemic versus nonischemic cTn elevations may aid in risk stratification where comparison to historical values does not.Patients experiencing AMI/PCI/CABG within 30 days, 88% of whom experienced their ischemic event during the index presentation, were just as likely to be discharged from the ED as admitted to inpatient or observation (8.3% vs. 9.8%; Figure 2).Together these results suggest hospital admission for urgent coronary evaluation in AHF patients is valuable among patients with recurrent elevated troponin and no prior ischemic workup.Likewise, with a known IHD history the clinical diagnosis of ischemic versus nonischemic cTn elevation should not be overly predicated on comparison to historical cTn, which appears to add little incremental value once IHD history is known.Current ED disposition patterns do not appear to reflect this approach, calling into question current ED risk stratification paradigms regarding cTn in AHF.
Prior data support our findings that historical cTn elevation is generally not reassuring in heart failure and may indicate a need for an IHD workup 6 .In a study of 196 chronic heart failure patients, recurrent low-level cTnI elevation was found to be a worse prognostic indicator than intermittent spikes in troponin. 23The patients' minimum troponin level across three or more all-cause hospitalizations, but not maximum values or trends between visits, predicted nearly 13-fold higher mortality. 23Our results here in an ED setting also corroborate prior literature suggesting that persistent elevation of cTn after hospital admission predicts worse AHF outcomes, 24 while persistent cTn elevation after inpatient discharge correlates with adverse cardiac remodeling and underlying heart failure. 25ile there are likely several possible explanations for why historical cTn elevation did not ameliorate ED troponin risk, the most likely is perhaps that patients with chronic troponin elevations are simply sicker.Lack of a prior workup for IHD appeared to moderate the relationship between historical troponin and adverse events (Figure 1).Patients without a prior IHD workup who have a documented prior troponin elevation likely represent individuals at high risk for underlying IHD who nevertheless have "slipped through the cracks."To that end, the fivefold greater odds of adverse events observed in this group suggests that an unknown IHD history together with ED cTn elevation should prompt hospital admission and evaluation for underlying IHD.Among those with a known IHD history, a lack of incremental value (i.e., either positive or negative) for the historical cTn was observed even when considering the magnitude of difference from the historical troponin.Most tellingly, a ratio of ED and historical troponin had virtually the same predictive value as the ED troponin alone.
In contrast, historical cTn elevation was associated with greater odds of ED discharge after adjustment for covariates (IHD history, This underlines a key opportunity for improvement: current ED practices operate under an assumption that preexisting or "chronic" troponin elevations are "safer" than when troponin is newly elevated and less heavily consider the relationship between IHD history and cTn elevation.In a recent survey of ED physicians' AHF disposition decision-making practices, attending physicians rated "higher troponin value in relation to past values" as more important to determining disposition than "troponin value, regardless of prior value" and as the third most important factor influencing disposition overall. 11Only "low blood pressure" and "new oxygen requirement" were rated as more important by physicians to their disposition decision making.
Despite how common this assumption may be, this is, to our knowledge, the first report to evaluate it in ED patients with AHF.
The ED disposition patterns we observed here and the disposition attitudes reported previously 11 directly conflict with what our results suggest: that admission is likely valuable with elevated historical and ED cTn specifically when IHD history is unknown.
The association of ED cTn elevation and adverse outcomes was predominantly driven by a 12.6% greater incidence of ischemic adverse events (Figure 1, Table 2).A more modest 3.1% risk difference occurred for adverse events in the absence of any ischemic outcomes (AMI/PCI/CABG), which failed to reach statistical significance.Given this contrast, accurate identification of ischemic versus nonischemic presentations appears to be a critical step toward the broader goal of identifying when and how cTn elevation in AHF informs the value of hospitalization versus ED discharge.Namely, while both ischemic and nonischemic cTn elevations are associated with adverse events, 6 the former is a much larger contributor to adverse event risk.It is possible that after excluding ischemia, patients with elevated cTn in the ED and a prior workup for IHD may be appropriate for ED discharge assuming no other indications for admission.This would seem to be corroborated by recent data from Fermann et al. 26 of AHF patients discharged from the ED, in whom AMI had been clinically excluded, which found no association between cTn and adverse events.
Unfortunately, several observations from the current study underline the real-world challenges in achieving these goals.First, ED patients who experienced 30-day AMI/PCI/CABG were just as likely to be discharged as they were to be admitted to the hospital (8.3% vs. 9.8%, p = 0.8), including for two patients adjudicated to have experienced an AMI during the index visit.To the extent that ED disposition decisions reflected lower 30-day adverse events in discharged versus admitted patients (11% vs. 17%), this was driven almost entirely by fewer nonischemic adverse events among discharged patients and not by a difference in ischemic events (Fig- ure 2).Second, despite excluding cardiogenic shock and STEMI, we observed a 9% rate of AMI at the index visit (Tables 1 and 2).This is similar to rates in multinational AHF registries dating back over a decade, 7,8 which reinforces that acute ischemia in AHF remains common.Third, around half of all cardiac catheterizations performed during the study were negative (Table 2), further suggesting the difficulty in noninvasively selecting patients for criterion standard workup and treatment for IHD.Fourth, AMI presentation was frequently atypical with regard to chest pain in the ED, which was both insensitive (44.8%) and nonspecific (62.8%).Overall, differentiating patients with ischemic events versus nonischemic cTn elevations in AHF remains challenging.

LI M ITATI O N S
This study was limited as it was a secondary analysis of previously collected data; however, this dataset was prospectively collected as part of other AHF research and the hypotheses preplanned.Verifying and adjudicating 90-day outcomes was performed by independent reviewers and subject to human error.However, each outcome was compared after blinded review and adjudicated by a third party to resolve conflicts.The prior troponin values were only those that were available in the system EMR, and the time frame between the prior value and the ED varied as much as a few days to nearly 2 years.
It is plausible that differential timing (inpatient vs. outpatient, end of hospitalization vs. beginning) of the historical troponin may affect clinical ramifications.However, these discrepancies between available information for different patients are a clinical reality that clinicians face every day.The same caveats, and rationale, apply to the fact that troponin assays changed throughout the enrollment period.
The 99th percentile URL, while imperfect for reasons previously discussed, is the current international standard by which both cs-cTn and hs-cTn are benchmarked.Multiple sensitivity analyses did not show a difference in results by troponin assay, including between cs-cTn and hs-cTn.The lack of a difference between ischemic adverse events among admitted versus discharged patients was based on a relatively small number of events, and thus we cannot rule out that random error or errors in adjudication led to this result.An event rate of 24% for COMP90 also may have limited statistical modeling for all possible confounders.Lastly, while discharge from the ED was positively associated with elevated historical cTn but not IHD history, we cannot say that this was a causative relationship for physician decision making based on the nonexperimental design of the study (i.e., we can only state association).

CON CLUS IONS
Acute heart failure patients with historical cardiac troponin elevation, in the absence of a prior ischemic heart disease workup, may benefit from hospital admission given an association with greater adverse events.Urgent coronary evaluation may be beneficial given that adverse event risk appears driven by ischemic outcomes.It may be reasonable to arrange an outpatient ischemic workup for some of these patients with close follow-up, but with a likely self-selected group for poor access to care, feasibility of avoiding admission may be limited.When ischemic heart disease history is already known, comparison to historical cardiac troponin does not appear to reliably distinguish high versus low-risk patients or ischemic versus nonischemic cardiac troponin elevation during the current ED visit.These findings directly contradict current disposition practices in the ED and do not support the common belief that cardiac troponin elevation in acute heart failure is safer when accompanied by historical cardiac troponin elevation.Opportunity for improvement is highlighted by the discordance between clinical outcomes we observed and current practices, with ischemic adverse events particularly common but frequently missed in the ED.

*
Adjusted to: Mean GWTG−HF Risk Score IHD history = Non−ischemic the Relationship Between ED Troponin and Discharge Probability

outcomes Composite outcome at 30 days AMI, PCI, and/or CABG at 30 days AMI at index presentation No (n = 284) a Yes (n = 57) a
11t differently, a common belief underlying ED physician standard practice is that cTn during an encounter for AHF has limited incremental prognostic value if a historical troponin is elevated.In a recent survey of ED attending physicians on their disposition decisions in AHF, physicians rated comparison of the ED cTn to a historical cTn as the third most important factor determining whether they would discharge a patient or admit them to the hospital.11Wesought to evaluate the validity of this assumption in the current sample, in case physicians in the current cohort behaved in a way systematically different from what would be expected under ED usual care.If historical cTn had the effects on ED physician decision making described above, we hypothesized that two observations would be true in the current sample: (1) Patients with elevated cTn in the ED would be more likely to be discharged if historical cTn was also elevated and (2) the multivariable-adjusted odds of discharge from the ED would increase as historical cTn increased relative to the ED cTn.To test the former, patients discharged from the ED with an elevated cTn were stratified by elevated versus normal historical cTn and compared by descriptive statistics.To test the latter, a logistic regression model for ED discharge versus admission (dependent variable) was fit for ED cTn, historical cTn, IHD history, and the Get With the Guidelines Heart Failure (GWTG-HF) risk score.The latter is a risk prediction score based on admission characteristics (blood pressure, age, blood urea nitrogen (BUN), sodium, heart rate, chronic obstructive pulmonary disease history), which is endorsed by the ACC and
Acute MI, Percutaneous Coronary Intervention, and/or Coronary Artery Bypass Graft Non-ischemic adverse events: Intubation, New or Emergent Dialysis, Mechanical Cardiac Support, Death *Occurence of any ischemic adverse event(s), with or without non-ischemic event(s), counted here as `Ischemic` (e.g.PCI + Death = Ischemic) 90-day Outcomes Stratified by Ischemic vs. Non-Ischemic Adverse Events* 11ute MI, Percutaneous Coronary Intervention, and/or Coronary Artery Bypass Graft Non−ischemic adverse events (Blue): Intubation, New or Emergent Dialysis, Mechanical Cardiac Support, Death *Occurence of any ischemic adverse event(s), with or without non−ischemic event(s), counted here as 'Ischemic' (e.g.PCI + Death = Ischemic) 30−day Outcomes Stratified by Ischemic vs. Non−Ischemic Adverse Events* Central illustration.After considering baseline characteristics at time of presentation for ADHF (including current cTn, eGFR, LVEF, Age, and IHD history), the incremental prognostic value of historical cTn in the EMR is dependent on the patient's IHD history.In those without a prior IHD workup, elevated historical cTn increases risk of multiple adverse events.When IHD history has already been established (whether ischemic or non-ischemic) historical cTn does not add incremental prognostic value beyond current cTn and the other baseline variables on their own.AHF, acute heart failure; AMI, acute myocardial infarction; CABG, coronary artery bypass grafting; CPR, cardiopulmonary resuscitation; cTn, troponin; eGFR, estimated glomerular filtration rate; EMR, electronic medical record; IHD, ischemic at the index visit, and GWTG-HF risk score).We suspect that this finding reflects physician beliefs that historical cTn elevation is a reassuring prognostic factor, similar to what doctors have said in surveys,11yet contrary to what our data suggest about the relationship between adverse events and historical cTn elevation.Meanwhile, when compared to the association between historical cTn and disposition decisions, other factors associated with ischemic adverse events were underutilized (e.g., new ischemic ECG findings, serial ["delta"] troponin, IHD history; Table2).In a majority of patients discharged from the ED, for example, the clinician did not obtain a F I G U R E 5 heart disease; LVEF, left ventricular ejection fraction; PCI, percutaneous coronary intervention; ROC AUC, receiver operating characteristic area under the curve; URL, upper reference limit.cTn