By Michael H. Crawford, MD, Editor

SYNOPSIS: A prospective observational study of 30-day outcomes in patients with suspected acute coronary syndromes triaged in EDs before and after the implementation of the HEART score showed that there were fewer hospitalizations and a very low risk of death or myocardial infarction among low-risk patients who were discharged without further cardiac testing.

SOURCE: Mahler SA, Lenoir KM, Wells BJ, et al. Safely identifying emergency department patients with acute chest pain for early discharge. Circulation 2018;138:2456-2468.

Since fewer than 10% of patients seen in EDs with chest pain are diagnosed with acute coronary syndrome (ACS), healthcare workers in this setting often use accelerated diagnostic protocols such as the HEART score (History, ECG, Age, Risk factors, and Troponin) to improve the accuracy of chest pain triage. Although shown to increase the percentage of such patients discharged early and to decrease cardiac testing, length of ED stay and cost, prior studies were insufficiently powered to provide prospective safety and efficacy data.

Investigators from the Wake Forest School of Medicine conducted a prospective study at three North Carolina EDs. The study included 8,474 adult patients with possible ACS from 2013-2016 but without evidence of ST-segment elevation myocardial infarction (STEMI) on ECG. Data were accrued 12 months before and 12 months after implementation of the HEART score. A wash-in period to train providers in the use of the score separated the two data collection periods. The study population included patients with known coronary artery disease (CAD). Patients with CAD or acute ischemic changes on ECG (new T-wave inversions or ST depression in contiguous leads) were immediately classified as not low risk, and no HEART score was calculated. Also, troponin was measured at time zero and three hours later. Patients with HEART scores ≤ 3 were designated low risk and discharged without cardiac testing. The primary effectiveness outcome was hospitalization in 30 days. The primary safety outcome was death or acute MI in 30 days. There were multiple secondary endpoints.

There were 3,713 pre-implementation and 4,761 post-implementation patients. The HEART score identified 31% of the post-implementation patients as low risk. Their 30-day death or MI rate was 0.4%. Hospitalization rates at 30 days were lower by 6% in the HEART cohort compared to the pre-implementation cohort (56% vs. 62%; adjusted odds ratio [aOR], 0.79; 95% confidence interval [CI], 0.71-0.87). Also, identification of MI during the index visit was higher in the HEART group (6.6% vs. 5.7%; aOR, 1.36; 95% CI, 1.12-1.65). The authors concluded that using the HEART score to triage suspected ACS patients in the ED was associated with fewer hospitalizations, better identification of acute MI, and a very low rate of death or MI among low-risk patients who were discharged without further testing.

COMMENTARY

The major implication is that the HEART score, as deployed in this study, can be used to identify low-risk patients who can be discharged safely without further cardiac diagnostic assessments such as stress tests or angiography. In this study population, 31% of subjects were low risk and demonstrated a death or MI rate of 0.04% in 30 days. There is a general consensus that an ED-accelerated diagnostic protocol for ACS should demonstrate a missed MI rate of less than 1% to be medicolegally viable. In this study, the HEART score easily made this cutoff. Other reports have reproduced these results such that more than 50,000 patients have now been reported on.¹ However, a recent study from the Netherlands revealed no change in the percentage of discharged early patients and a major adverse event rate in 30 days of 2%.² There are some critical differences between the Netherlands study and the Mahler et al study. In the latter study, two troponins were measured three hours apart, whereas the authors of the Netherlands study used only the initial troponin. Also, in the Mahler et al study, patients with known CAD or ischemic ECG charges were automatically not low risk. In addition, the HEAR part of the HEART score was presented on the electronic health record (EHR) as a series of binary questions to reduce subjectivity.

The major limitation of the Mahler et al study is that it was not a randomized trial. Secular trends and provider maturation may have affected the results and reduced the differences between the groups. Also, the study was EHR-based and subject to the accuracy of those records. Further, events after 30 days were not analyzed. In addition, the differences in hospitalization, testing, and length of stay were modest; however, these differences, taken across the entire nation, would reduce healthcare costs. Finally, an analysis of using two troponins over three hours alone revealed almost-as-good results for 30-day death and MI vs. HEART (sensitivity, 92% vs. 98%; specificity, 88% vs. 40%, respectively). This highlights the importance of troponin in the HEART score. However, over-reliance on troponin is not wise. Early in an ACS event, even a three-hour troponin could be negative. Thus, the superiority of the HEART score makes sense.

Using the HEART score as deployed in the Mahler et al study with two troponin measures at least three hours apart and the automatic exclusion of patients with known CAD or ischemic ECG changes from the low-risk group should change ED practice, as it is extremely safe to discharge low-risk patients without further testing. Many EDs admit intermediate-risk patients to a chest pain observation unit and high-risk patients to the hospital. Whether these tactics are equally as effective and safe remains to be studied.

REFERENCES

1. Sharp AL, Wu YL, Shen E, et al. The HEART score for suspected acute coronary syndrome in U.S. emergency departments. J Am Coll Cardiol 2018;72:1875-1877.
2. Poldervaart JM, Reitsma JB, Backus BE, et al. Effect of using the HEART score in patients with chest pain in the emergency department: a stepped-wedge, cluster randomized trial. Ann Intern Med 2017;166:689-697.